help-with-hw-design

[knotty tiger]

so the Type-C spec says the receptacle shell must be connected to the PCB ground plane, so i guess those designs out there with EMI filtering on the shield conductors are non-conforming. (the spec doesn't say whether the signal ground pins need to be connected to the ground plane, so i guess EMI filtering could happen there.) it does also say all GND pins must be connected together at the receptacle

[woven bluff]

correct me if I'm wrong, but I don't think USB have signal ground

[knotty tiger]

it's combined signal and power ground. non-SuperSpeed USB is nominally differentially signaled (D+/D-), but it does have some important single-ended states. also, there's communication over the CC lines that's referenced to ground

[last prairie]

this is what I ended up doing I dont know if this is right...

[knotty tiger]

i think you should also tie all the VBUS pins together. also connect the shield pins to ground. D+/D- should also be connected to their counterparts. if you don’t connect the same signals together, you won’t get working cable flip support

[last prairie]

aaahhhh

@knotty tiger ok that makes sense. I will do that

[last prairie]

So I connected all pins of the same type together. My concern is this creating "ground loops" but I don't know how those work anyway.

[supple pollen]

Ah, that's a change. The earlier ones seemed to use the shield as a static drain and generally connected it to ground via a large (like 1 meg) resistor and sometimes a small parallel capacitor.

[knotty tiger]

yeah, USB 2.0 only says

The shield must be terminated to the connector plug for completed assemblies. The shield and chassis are bonded together. The user selected grounding scheme for USB devices, and cables must be consistent with accepted industry practices and regulatory agency standards for safety and EMI/ESD/RFI.

[last prairie]

connected the shield pins, forgot about that. So in theory this should work no matter how the cable is flipped?

[knotty tiger]

yeah, looks right for USB 2.0 on Type C

there are joke/prank USB C devices that behave differently depending on cable orientation, but outside of that, supporting cable flipping is kind to users (and required by spec)

[last prairie]

should have said this before but this is the circuit chunk that I am trying to replicate:

[knotty tiger]

that could be a different receptacle that connects different instances of the same signal together internally, or maybe it's taken care of in the footprint

[last prairie]

@knotty tiger its from the QT Py SAMD21 schematic. This is what eagle says about the part:

[last prairie]

when I look at the BOM it gives:
USB_C_CUSB31-CFM2AX-01-X

[last prairie]

@knotty tiger I don't know if this is how it works but I have each vbus connected in series (?) like this:

[unique patio]

that's not in "series", they are just all tied together

with a single run

i guess you could say the pads are in series

you might want to look at some of the adafruit samd21 designs

[last prairie]

what I was thinking was signal coming out of one of the vbus pins (if that's what is meant to happen) would go through the other vbus pins meaning that a diode would be needed.

[last prairie]

I am coppying the qtpy atsamd21

[unique patio]

there is no signal on VBUS, it's just supply voltage. They are mean to be all tied together

[last prairie]

yeah so the voltage coming from one of the vbus pins will be distributed through all the other vbus pins right? If the current runs through something that is designed be outputing something wouldn't that result in a short or something breaking right? probably not I just dont understand but still. :/

[unique patio]

There are multiple VBUS pins so that you can turn the USB-C connector over, since it's reversible: https://www.allaboutcircuits.com/uploads/articles/Fig1m11292018.png

the VBUS pins are all supposed to be tied together

[last prairie]

ooohhhhh

[unique patio]

if not all of them are connected, you'll still get power through at least one of them

[last prairie]

ok I get it, they are not all connected and the ones that are are both outputing

[unique patio]

they might or might not be, but if they are, they are all supposed to be the same

they are not supposed to be different from each other

[last prairie]

ok that clears so many things up thanks alot.

[cloud hinge]

I have a project that will handle a constant current of 25A per channel, 4 channels. I'm in the process of putting the PCB together and have chosen 2mm/6oz copper. Two layer. I've tried putting the power channels on one layer and the signals on bottom. This type of board is a bit expensive so please provide me with any critiques you have to my design. The 'Lug' is the big power input lug. With so much copper around it, will it be impossible to solder? Any other insights are much appreciated.

[soft nacelle]

For that high of a current it could be a good idea to make the traces copper pours instead. I usually make each "trace" a pour on as many layers as possible and stitch it with regularly placed vias on a 1mm grid. Soldering could be difficult yes, though with plenty of flux and a good temperature controlled iron it would be fine. Using thermal spokes could be a great idea to improve solderability

something like this:

[tough matrix]

what kind of connector is J1?
looks like for each channel it has one through-the-hole pin, spaced 0.1", which seems unsufficient for 25A. Is the connector rated fro such current?

and how wide are your power traces?

[cloud hinge]

The contacts on the connector are rated at 25A but I’m not so sure the PCB side connection is sufficient. J1 will most likely be swapped out for a Deutsch DTP and use doubled pins.

[cloud hinge]

I like this idea. Do you see any more issues with noise doing it this way? In my design I will have some current sense wires surrounded by PWMd 25A traces.

[tough matrix]

doubled pins on deutch DTP sounds good

[last prairie]

I am copying the schematic from the QTpy atsamd 21, this is what I have made. Can anyone spot any errors before I proceed? (its my first PCB)

[tough matrix]

Looks like VDDCORE pin of the IC. is not connected to 3.3v bus?

also, it is generally advised to put 0.1uF caps next to each power input pin of the chip. QT Py skips some of them just because they do not have room for it- it is a really small board - but if you have room, better add it.

[last prairie]

That's what's happening in the reference schematic, I just copied it. But vddin and vdda are connected to the 3.3v bus.

[tough matrix]

indeed, you are right

[last prairie]

What does vddana even mean? I just realised I had no Idea. So if I wanted to install those capacitors, it would be on the VDDIN and VDDANA pins right? In what way does in benefit the device, It is clearly not essential\ but if it would make it better I am willing add it.

[supple pollen]

One popular cost saving approach is to use ordinary 2oz copper and leave the traces without soldermask, so you can thicken them with solder or even solder wire along them to boost their current carrying capability.

[supple pollen]

VDDANA is the analog supply voltage. If you're using the analog peripherals, it's common to add a little extra decoupling/filtering between the logic and analog supplies to reduce interference.

[tough matrix]

0.1uF caps are used to filter out noise.
How essential it is depends on how noisy your power input is.
But since these caps cost next to nothing, I'd add them.

I also suggest you check ATSAMD21E18 datasheet: https://ww1.microchip.com/downloads/aemDocuments/documents/MCU32/ProductDocuments/DataSheets/SAM-D21-DA1-Family-Data-Sheet-DS40001882H.pdf

in particular section 45.2:

and also this:

[supple pollen]

That little inductor in the top diagram is the "decoupling" I was referring to. There are a few ways of doing this, that's a common one.

[rigid plume]

Hi folks what would be the standard logic voltage for the Wiegand RFID card readers like this one? https://www.amazon.com/UHPPOTE-125KHz-Wiegand-Connect-Control/dp/B01B18CI3G/ref=sr_1_6?crid=UNNQJ75R07KL&keywords=rfid+wiegand&qid=1703973128&sprefix=rfid+wiega%2Caps%2C182&sr=8-6 Thanks

[last prairie]

Here I am following the recommended layout you gave me and have started by laying out all the pins in question out, but I cant find VDDIO. What is its purpose, why is it not on my footprint and what pin should I use for it?

[tough matrix]

the datasheet is common for different versions of samd21.
The one you are using - SAMD21E18 - has fewer pins, so it only has VDDIN, VDDCORE and VDDANA, but doesn't have separate VDDIO.

in case you are curious what these symbols mean, the answer is here:

[last prairie]

the actual QTPY uses the same ATSAMD21E18, so should be fine right? I will just skip the pin...

[tough matrix]

yes

[last prairie]

ok thanks

[last prairie]

@tough matrix from what I have found out that squiggly symbol is an inductor, one that I don't understand is how I can choose the suitable one for this circuit. I am not familiar with this part...

[tough matrix]

I had used this ferrite bead:

https://www.digikey.com/en/products/detail/murata-electronics/BLM21PG331SN1D/1915299

but honestly, this you can really skip.

[cloud hinge]

i've completely forgone traces for the first round.. just put big fat pads in there for 12g wire. This way I can test the components first, and then if I ever follow up I can work on the PCB more.

[soft nacelle]

Noise wise as long as you aren't running any analog or sensitive digital traces over the pours it should be ok imo. If these tracks are referenced to ground having a solid ground layer would be helpful. (Just going to throw out the option of 2 oz 4 layer, having the pour across three layers would give you the same 6 oz of current carrying capacity and a ground layer).

The shorter the path the better, too. Looking again it seems like you may be able to move Q1-Q4 and J2 closer to J1, and move the microcontroller and J3 below or to one side. That would make the high current tracks a lot shorter

[supple pollen]

Standard for Weigand is 5V, but a counterfeit from amazon? Could be anything.

[last prairie]

SO even if I do use this despite it not being necessary it will still work wont it?

[tough matrix]

of course

[last prairie]

great, so this is why I made.

[supple pollen]

You will ultimately have to tie the grounds together

Many designs do so at one point.

[last prairie]

Missed that I meant to do that...

[last prairie]

In this I have this pin connected to a test point like in the QTPY schematic, so if I added this too it I dont know if that test point would still work. Also where is the wire connected to? it says VDD rather than 1.62V-3.63V...

[rigid plume]

Hi folks I was wondering what transformer parameters would I need to look for this TI SEPIC circuit, 12-24v input rated and 12v output

[supple pollen]

The chip data sheet will specify things like the primary, secondary, and leakage inductances, current, operating frequency, and so on, and generally offer some suggested parts. You can also search parts suppliers for the chip number, as they'll sometimes carry transformers designed to work with that chip. Transformers are more complicated than they seem.

You may also be able to use TI's "WEBENCH" software, which has a lot of the parameters and parts databases built in and can help you with optimization and component choice. @rigid plume

[woven bluff]

you can also print your own PCB transformers

[cursive sentinel]

Probably not with the required electrical characteristics.

[last prairie]

this is a before and after of how I have changed the SWCLK pin to include what the data sheet is saying is "critical for reliable operations". I think I have got this right... I still don't understand how adafruit got away with not doing this since it is so important...

[last prairie]

Wait how come we can program the chip through the USB port then? Is there some sort of initial set-up that has to be sent through the SWD interface? If so that would make things harder.

[jade wedge]

So, Adafruit gets away with not having the pullup resistor on SWCLK because it's only critical when operating in an electrically noisy environment. Adafruit is simply assuming that you will not be operating in such an environment. As such the internal resistor that the chip already uses to pull the pin high is sufficient.

Additionally, we get to program Adafruit boards via USB rather than the normal programming interface on the chip because Adafruit has kindly loaded a bootloader onto most of their boards that initializes various USB functions, as well as an endpoint for the computer to communicate with before passing control over to whatever program the user has loaded.

[last prairie]

Ada fruit has added test points to these pins, do they use them for anything before the device is sent off or is it just a complimentary feature?

[jade wedge]

A little of both. I believe they do use them for testing, but they absolutely admit that advanced users are free to use them to fully reprogram the chip. This is something to consider if your program is just slightly too large to fit. By programming the chip directly you can erase the bootloader and reclaim roughly 2kB of program space.

Obviously at that point you will not be able to use the USB unless your program also does the initialization and endpoint setup.

[last prairie]

So before the bootloader is put onto the chip the only way to send data is through the swd pins/protocol. So that means that the bootloader is installed through something like SWD and not USB right?

[jade wedge]

Yep

[last prairie]

Perfect, another question is how come the swclk pin is connected to VDD when is meant for data?

[jade wedge]

It's part of the programming process. During boot up the chip (regardless of bootloader) checks it, and pulling it low is for programming. So they pull it high by default to signal normal operation.

[last prairie]

Huh. I haven’t got the schematic on me right now but I will take a look when I can to better understand what’s going on. Thanks for the info

[last prairie]

So if I inserted a switch in-between VDD and the pin I could stop it from being pulled high and allowing the chip to be programmed (eg to put the bootloader onto it). Right?

[jade wedge]

No. The programming process needs to drive the pin itself. Suitable programmers will pull the pin down as needed.

The pin is internally pulled high, so you can leave the pin floating.

Trying to manually disconnect the VDD connection would also be bad because nobody's perfect, and you don't want to risk forgetting to flip the switch, and shorting something out.

[last prairie]

So leaving it connected to VDD will still allow me to add the bootloader and excluding it will only effect the devices performance in noisy environments. I can see that when it comes to the to the qtpy adafruit has left in the bare minimum for what is needed for the chip to run.

[jade wedge]

Connecting to VDD directly effectively disables programming. After the chip is programmed, this would be fine. I haven't been reading the whole conversation, just mostly this morning. If you're designing your own board, and will be putting on unprogrammed chips, you cannot connect SWCLK to VDD directly. If you expect to work in a noisy environment, you should use the pullup resistor. This will protect the chip from interference, and allow reprogramming.

Are you planning on making a totally custom board, or are you plugging one of Adafruit's boards on top of something you're designing?

[last prairie]

Plugging it on top of something I am designing. Until I build up the knowledge to design one from scratch. I choose the QYpy because it was simple and I could read it.

[jade wedge]

Ok. So for that, on your board you can connect directly to VDD if you really want since you'll have the option of pulling the board off when you need to reprogram it. If instead you want to go the route of having a programming port on your board so you don't have to pull them apart, you will need the pullup resistor between VDD and SWCLK. That way the programmer can still drive the pin, but without risking a short.

Also, make sure that if you do go the route of having a programming port on your board, it branches off from trace between the resistor and chip.

[last prairie]

Ok that sounds way to complicated so why don’t so just use one of these to put the bootloader on before I but the chip on the pcb… (and do it the way adafruit did it)

[jade wedge]

Oh, if you can use that, then I've misunderstood. I thought you were connecting a whole existing board to your design, not a plug on chip. So, let's backup a second.

[last prairie]

Yeah, what I mean is do all the bootloader things with this first, solder the chip to the pcb and then program what I can through the USB port.

[jade wedge]

You plan to be buying individual chips to put on your board?

They won't initially be attached to anything?

Actually I guess the most direct question is what are you buying to start this off with?

[last prairie]

I don’t know if chip is the right word, I am buying the same ones used on the atsamd 21 qtpys from lcsc and reflow soldering them to my pcb.

[tough matrix]

if you need help with flashing the bootloader, feel free to private message me

[jade wedge]

Ok, so sounds like just the chip. So, first problem, they don't come with a bootloader. That's adafruit's doing. Definitely walk to shurik, that's out of my experience.

You will need to put the programming port onto your board. So you will need to leave the SWCLK pin either floating, or with a pullup to VDD.

Personally I would suggest starting off with the qt PY itself, and just socketing that on if you can. Just to start with. Second revision can use a chip directly on the board.

[last prairie]

Thanks, I was thinking about that. I know they won’t come with a bootloader. I have worked with just chips before like the attiny 85 and getting a bootloader onto that. I think the fist thing is to get the shcematic done and then move onto getting a chip that will work.

[jade wedge]

Or a similar board with that chip if you need more pins.

Oh, you've done some of this before, ok, so that's good.

[tough matrix]

note that to flash the bootloader you need a programmer. Either a dedicated one like https://www.adafruit.com/product/3571, or (haven't tried it myslef) you can also use another SAMD21 board and Adafruit's DAP library: https://learn.adafruit.com/programming-an-m0-using-an-arduino/overview

[last prairie]

So your saying to solder a socket to the pcb so I can iterate?

[jade wedge]

Yes. Repeatedly soldering components is not the best, so you'll want to be able to put the chip on once, and leave it there. Adding an appropriate programming port is the way to go for that.

And I mean a socket to plug your programmer into. The SAMD21 does not come in any variation that you can socket.

This is something I'm going to have to deal with myself soonish. I'm planning on moving from the ATTiny85 to the 1616.

[tough matrix]

I leave 5 holes on my boards, connected to SWCLK, SWDIO, GND, nRESET and 3V3. I use staggered hole placement, so for programming I can just plug in a 5-pin male header and it will hold by friction

[jade wedge]

Oh that's clever.

[tough matrix]

if you are constrained for space, you can instead use test points, and connect using pogo pins

[last prairie]

Can’t I just connect this (reply) to one of these: SEGGER J-Link EDU Mini - JTAG/SWD Debugger and once I have loaded the bootloader on the the chip add it to the pcb…

[tough matrix]

sure you can. The only drawback is that it won't be possible to replace the bootloader, but normally you do not need that.

[last prairie]

Can’t I just have a socket like this on the actual pcb so that I dont have to solder it and can take it out if i need to?

[tough matrix]

yes, but it takes up a lot of space.
If you have that much space available, why not just put headers and plug in the the actual QT py board? this way, you do not need to worry about bootloader, USB trace lengths and widths, or about anything else

[last prairie]

I don’t know but it doesn’t feel right….

[last prairie]

They have this on the actual qtpy, it’s the test points

[woven bluff]

something tells me that thing is not cheap

[last prairie]

I have this already done... with the test points being set up so that I have access to all the points for programming. This is what they did in the adafruit QTpy schematic. I changed nothing, I just laid them out this way.

[rigid plume]

Hi do you think this will work

https://www.mouser.co.uk/ProductDetail/Wurth-Elektronik/750313442?qs=x30FU183Eu73efjb%2F5Oi3Q%3D%3D

[last prairie]

@tough matrix maybe thats how they programmed it rather than doing it before they soldered the chip to the board.

[tough matrix]

I am certain that Adafruit first solders the chips to the board and then programs them using testpoints and pogo pin jig.

[pale umbra]

wish I knew about pogo pin jigs before so I didnt have to buy a soldering iron 🤣

[faint rain]

A soldering iron will come in handy sooner or later.

[torn saddle]

Good Afternoon. I have a program using Talkie library to create speech. The hardware and software all works. I have two issues.
Speech not real clear
The library does not have all the words I need.

Questions:
Does anyone have a recommendation for a voice synthesizer that is clear?
I am using an Arduino Zero, most Synthesizer use a serial port. What is best way to add a third serial port to an Arduino zero?

[supple pollen]

You can probably use SoftwareSerial to implement an additional serial port

[torn saddle]

Thanks that solves part of problem

[native plover]

I've got some fairy lights I want to control with a microcontroller. I want to attach just the LEDs (blue part in the picture) to an H-bridge and PWM them, but I'll probably use 5V. Since I've got 60x20mA=1.2A running in parallel at an estimated 3.5V, it'll have to drop 1.5 V, so unless I'm missing something, that would mean a resistor of >=1.25 Ohm with a 1.8 W rating. Is that correct? I feel like I'm missing something here. In their design, they run off 4.5 V and there seems to be only a single 10 Ohm resistor in line with the LEDs (SMD, labelled "100"). How did they pick that? How does it have a sufficient power rating, being this tiny? Did they use a larger-than-neccesary resistor to limit current beyond what the LEDs can handle so they have to dissipate less power?

[supple pollen]

I suspect a few things are going on with that setup. You're right that the larger resistance will equate to lower current. However, the LEDs may be arranged artfully so the current draw is lower (perhaps some sort of series/parallel arrangement). Additionally, there is resistance in the H bridge, and the power supply and wiring as well. There may even be more resistors tucked in with the LEDs.

On top of that, they may not be running at a full duty cycle, which would further reduce power dissipation.

[native plover]

Unless they use LEDs with integrated resistors, I'm fairly sure there aren't any. It's the kind where SMD LEDs (0805 package, I think) are soldered between two copper wires, so not a lot can be going on.

I want to be able to run them at 100% duty cycle (shared between the two directions) and I don't want to waste anything in terms of light output. So a small, high-power resistor would indeed be the correct choice?

Can I get by with using a bunch of larger, lower-power resistors in parallel? Or drop some voltage on something like a diode? I don't have resistors of that wattage 🙂

[supple pollen]

Yes, but I'd rate it conservatively (3 watts, maybe 5), especially if it's going to be enclosed in anything

[native plover]

Alternatively, I could order a resistor in a TO220 package, but that's 10 bucks. Fairly expensive imho.

[supple pollen]

Yes, you can build resistors out of series/parallel combinations of smaller resistors to distribute the power dissipation. I actually did this in the middle of the night to replace a 2Ω 3W resistor I didn't have, in order to get the freezer running so our food wasn't ruined. I later replaced that with a proper resistor (I used a 5W one so it wouldn't get so hot)

Note that the TO-220 housed resistors generally need heatsinks as well (as do other chassis mount resistors)

You can drop voltage with diodes, but they still have to dissipate the power

Resistors are generally a better bet, as they stabilize the current somewhat, evening out variations in the voltage drop of the LEDs

[native plover]

Then some resistors would probably be the easiest solution, they have more surface area and I've already got them here.

Thanks 🙂

[last prairie]

wait so the only extra components I need to program this PCB is the programmer thing. This is good. I put the 5v pin there instead of the 3.3v pin because it would go through the regulator and power the chip anyway.

[woven bluff]

I would just go for SMPS if you want to drop a lot of voltage to drive the leds

SMPS is not scary anymore nowadays

constant current BUCK convertor designed for leds only need 4-6 external components

[pale umbra]

Trying to find a buck-boost that will support from 4 to 8 NiMH batteries @ 2AH with a discharge graph like this: to power servos, a 3.3V MCU and a 5.5V MCU at the same time. I think this would involve using 3x buck-boost that have a min voltage of 4V or so a max output amps of 4A,a quiesent as low as possible, a duty cycle as high as possible with the necessery breakout to use it because they all seems to be smd . The powermax from adafruit dont seem to provide enough amps for some MCUs at low input voltage. The plan is to separate the battery pack into 3 paths each with a buck-boost, 2 of them possibly limited to 500mA to uses the remaining 2A to power servos directly. Ideally Id like the top VIN to be around 12V so I can reuse them for other battery types or to have 8xAA for the stuff to last longer. The alternative is using buck-boost with USB output but that isn't very compact since Id have to use very short usb cables directly or other cables to convert it to usb-c usb-a microusb etc

[supple pollen]

Pulling >4A out of a 2Ah NiMH pack is going to be hard on it. Presumably only the servo regulator will need to be high current, the MCU supplies could probably be low-current units.

For the servo regulator, you could use something like https://www.pololu.com/product/4082, which can provide 5V at up to 3A from a wide variety of input voltages.

Perhaps https://www.pololu.com/product/4083 for the 5V supply, then an ordinary linear regulator from there to the 3.3V supply, or if you really want a switcher at that position, you probably don't need buck/boost, so you could use the https://www.pololu.com/product/2842 buck regulator.

[woven bluff]

if you want to design something compact, sooner or later you'll find existing SMPS boards not going to cut it and end up printing your own

[supple pollen]

Designing an SMPS is somewhat involved, and many of the commercial versions are quite compact. Like all engineering, it's a series of tradeoffs.

[pale umbra]

Is there a particular reason why companies seems to over buck-boost/buck/boost with set output voltage instead of letting peoples choose like regular ones and at more expense than them? Like do one that let you choose the output voltage have huge tradeoffs vs a fixed one and are less complex so less costly ? All adafruit/sparkfun/polulu seems to offer are fixed voltage output

Like I even fouhnd one that dynamically switch to a second voltage when in low power mode but couldnt find a breakout for it

[supple pollen]

Fixed output is easy and stable, just two resistors in the feedback loop. Many projects need one of a small set of popular voltages, so drop-in units are appealing. Making one adjustable involves adding an adjustment potentiometer, which is a more expensive, delicate, and troublesome part than two resistors. Pololu does offer adjustable ones.

[gloomy garden]

A quick update and thank you to @supple pollen for your help on my AC LED driving circuit question a few weeks ago (12/21/23). I am happy to report that with a few lessons learned the hard way (mostly dead capacitors, but that is what prototyping is for!) I have a circuit that has run for over a week with many on-off power cycles to test the transient voltage suppression that the paper I based this circuit on suggested was a primary concern. That said, despite reading several documents on TVSes I still have a question due to my not fully understanding how a TVS works. As a reference, my circuit schematic is below with D2 being the bidirectional TVS and D1 a small-signal diode to act as an LED for the alternating cycle. My question is: A TVS suppresses transient voltage surges, but transient relative to what? The paper indicates that when powering on, a worst case 3x10-5 second 20A forward current surge can result. I can somewhat understand what a TVS is doing in this worst case, but how does it handle an AC power source with a voltage that is constantly varying from Vp-p? Isn't the sine wave constantly "transient"? Apologies in advance if this is basic electrical theory (something I am learning as I go), but I would like to try to fully understand the TVS aspect before I implement a permanent solution. The last thing I want to do is go through the trouble of creating my replacement LED indicator boards and installing them in the electrical panel on our boat (a major pain) and then find out I did something dumb like mis-spec a component out of ignorance.

[supple pollen]

It is constantly changing, so there's current flowing through C1 (phase shifted with respect to the voltage). However, that current is normally reasonably small, so it develops a small voltage across the LEDs, not enough voltage for D2 to conduct. But if the power is applied partway through the cycle, there could be a large sudden change of voltage, which will cause C1 to conduct a similar large pulse of current, which could be more than the (other) diodes can withstand. If the voltage across them becomes high enough, D2 conducts, absorbing the current/voltage spike and limiting the maximum voltage to which the LED and its antiparallel diode are exposed.

[gloomy garden]

Thanks again, that helps my understanding a lot but has led (pun intended) to two additional questions... when you say "If the voltage across them becomes high enough, D2 conducts..." would the voltage at which D2 is "triggered" and starts conducting be the breakdown voltage (Vbr) in the graphic below? Or would it be the maximum stand-off voltage (Vrm)? And looking at the figure, for my 240VAC application should I spec a TVS diode with a Vrm of 240V * 1.4 = 336Vpeak, or 672Vpeak-peak since this is AC? And one last question just occurred to me looking at that figure... I am now more confused about how the clamping voltage (Vcl) factors into the specification of a TVS diode. Based on the paper, the estimated worst-case transient would be a 30us 336V/20A surge. Does this mean I should select the diode based on both the clamping voltage / peak pulse current and an appropriate stand-off / breakdown voltage? And if so, what is the most important factor, the Vrm or the Vcl? Sorry for all the questions... the more I learn about this the more I feel like I am not factoring in all the considerations necessary to correctly spec a TVS diode for my application.

[quartz flume]

https://www.finh.cc/ash
This is a really cool product that will definitely never be released. What SBC would be suitable for running these AI models on the board?

It has to be thin enough to fit in that case. I was originally thinking a Pi Zero 2 W but not sure if it's powerful enough.

It'd be my first machine learning project. Thought it would be a cool, simple one since it machine vision and text generation without crazy implementation. Just displaying them in a UI 🙂

I sincerely doubt 512 MB on the Zero will be enough. Is my only option a 4 or 5 with the ports removed? 😅

[cursive sentinel]

I mean... any phone can do this easily.

[quartz flume]

I know. It's the novelty of the thing. Like a pwnagotchi.

[pale umbra]

you did mean running right ? not training. Running it ? right ?

[quartz flume]

Yeah, running it

I think I said running. Did I say training?

[pale umbra]

no but we dont know how much you know about this topic

and that you might confuse running with the much harder training step

[quartz flume]

I have a basic understanding of machine learning and some experience with SBC's. I think I'll go the route of a Pi 5 without its port stacks. I'm a bit worried about frying a nice board though lol

[pale umbra]

Honestly I think it's mostly a disk space issue for the models if the cpu / ram are ok

Like Im looking at one model with all the training data and it's open-source but each animal takes 2-3gb of data

if you are near a zoo might be worth giving it a shot to see if the hardware you want can actually run an existing model

[quartz flume]

Oh jesus, really? I had no idea the models were that large

[supple pollen]

As the graph shows, it's not all or nothing. It begins to conduct at Vrm, and conducts heavily at Vbr. To protect the diodes, I'd select a low voltage device.

[supple pollen]

You could probably do that with TensorFlow Lite. It doesn't make ridiculous hardware demands. I'd suggest trying it with a Zero 2 and see how it goes. If it doesn't fit, you aren't out a lot of money.

[supple pollen]

Any phone? How long should I crank the magneto?

[supple pollen]

A quantized TensorFlow Lite model is generally 500kB - 3500kB or so

[quartz flume]

That sounds terrific, thank you. Where would you suggest I start on this project?

[supple pollen]

Run through some tutorials, it's a great way to get a feel for the steps and various moving parts. I don't have a link to the (probably obsolete) one I used a couple of years ago, but this one looks decent https://www.tensorflow.org/lite/models/modify/model_maker/image_classification

The "run in Google Colab" option is a great way to get started quickly

[humble swift]

Just a general warning to never eat anything just cos na AI told you it was safe to eat 😛

[jade wedge]

Whaaa? You mean GLADoS might not have my best interests in mind when she suggested making a cake with "Fish-shaped ethylbenzene"?

[woven bluff]

that's because AI's own training data is poisoned to begin with

[humble swift]

I saw a few people using an ai to check some wood mushrooms and im like...no....

[supple pollen]

I loaded the GLADoS voice into my GPS. It's a hoot. "In ... negative ... 600 meters ... take a U-turn"

[jade wedge]

I kinda wish I could do that, but I don't think you can do custom voices with the default android speech synthesizer.

[pale umbra]

I once asked an early gpt about making a recipe with my leftover: shredded chicken, lime, fish sauce, plutonium nuggets, bag of asian-style frozen vegetables and it happily gave me once telling me to sautee the plutonium once the main dish is over without any disclaimer/sermons

[wispy blade]

anyone know a good arduino/metro shield for a display? i like the 20x4 LCD's but they dont come on shields and probably take up too much io(as im rapidly running out no surprise running 2 i2c busses and 2-3 spi's)

[wispy blade]

these(https://www.adafruit.com/product/1651) take up a pin i need

[supple pollen]

Could you remap that pin?

[wispy blade]

im trying that right now

is the metro rp2040's builtin flash connected to spi0 or spi1

[supple pollen]

It's on the QSPI port

[wispy blade]

Oh that makes sense, thanks ^w^

[grand wind]

This might be a really odd question and/or horrible idea, but could I (realistically) solder an FPC cable meant for a connector directly to a PCB using hot air or a hot plate?

[supple pollen]

Some commercial devices are built that way, so I'm guessing it's possible, but I'm unsure what tools and techniques are used. One I'm familiar with is the Nintendo Virtual Boy https://www.projectvb.com/vb/tech/displayfix/beforesolder.jpg @grand wind

[humble swift]

wasnt that cable notoriously dodgy

[supple pollen]

Heh, yup.

[onyx kernel]

hey. im just wondering what VDDIO stands for? I have a ICM-20948 gytoscope with a VDD of 1.8-3.6v and a VDDIO of 1.71-1.95v. As i understand it VDDIO is for the in/outputs of the chip, does that include I2C? So would i need a 1.8v I2C bus? Also this is a project of mine which i started months ago and i just picked it up again and i seem to have designed it with nothing connected to VDDIO. But when testing it back then i connected the i2c to 3.3v so is it fryed?

[supple pollen]

It's the power supply for the I/O pins. Some chips offer separate power supplies for the core and I/O like that.

It can be useful if you want to have your I/O at a different voltage from the core

[onyx kernel]

so its most likely fried when i connect it to a 3.3v i2c bus (with VDDIO floating)

[supple pollen]

I don't think that would necessarily fry it. Presumably the protection diodes would conduct and drag up the VDDIO rail to 3.3V minus a protection diode drop.

[inner rampart]

What's the feasibility of getting the original manufacturer of the 32x8 max7219 dot matrix displays to design one with an esp32 module onboard with a usb port. https://www.amazon.com/HiLetgo-MAX7219-Arduino-Microcontroller-Display/dp/B07FFV537V

[left grove]

sounds possible to do yourself 😝

[supple pollen]

It's easy enough to just hang an ESP32 board on one

[inner rampart]

true but I really want the entire device to be completely assembled by the PCB house.

[woven bluff]

can I replace this part of the ZVS with a LDO?

also, doesn't the primary need some kind of snubber?

[left grove]

you can totally get that done as well.
There are services that offer PCB+parts+assembly so you just send them the files and you get fully assembled PCBs.
I haven't made such things but I've seen people here and on other parts of the internet use them.

[supple pollen]

No, the primary doesn't need a snubber because it's a resonant design: C1 and the primary form a tank circuit with a resonant frequency, and the transistors just keep it oscillating (like pushing a child on a swing).

[woven bluff]

humm, this is also the first time I try the non-center-tapped ZVS version.

Wouldn't it short this way?

[jade wedge]

Not knowing exactly what this is, I suspect "shorting out" is largely the point. This looks like a design to take a DC power source and turn it into high frequency and (presumably) high voltage AC. Transformers of that design don't work with DC. If that is the case, it's less of a short than you might think as the load on secondary side of the transformer should create an inductive load on the primary side.
But, I could be talking out my butt here.

[woven bluff]

those two paths do not pass through primary

[jade wedge]

No, but they will influence it, and be influenced by it.

[woven bluff]

they just charge L1 and L2

[supple pollen]

The idea is that L1 and L2 have high enough inductance that they block the high frequency switching signal and just conduct the DC supply.

If you were to leave one of the transistors on, yes, high current would flow through an inductor and that transistor. But the transistors switch on and off alternately at a high frequency.

[woven bluff]

but they do not supply DC either

this two paths switch on and off the same time, and only one of them is useful

[supple pollen]

What's going to happen is the nodes at the transformer ends of the inductors will end up with a DC potential of approximately VCC/2 with an AC waveform superimposed on it. So current will flow through the inductors as the power supply tries to pull that VCC/2 up to VCC, this is the power supply for the primary, and is what you want. The AC waveform will flow back and forth through the transformer primary and capacitor, propelled by the switching of the transistors.

As long as the switching frequency is high enough in relation to the inductance of the inductors, they'll block most of the AC.

In practice, the majority of the current does not flow along either of the paths you have marked.

[woven bluff]

so the primary will be charged by L1 and L2 mostly?

[inner rampart]

@left grove yes it sounds simple but once you actually try to do it many complications arise. Mostly part sourcing.

[supple pollen]

Not really "charged", but voltage will flow in through L1 and L2 (mostly DC) and out through Q1 and Q2 (DC and AC). What the circuit is doing with this power is getting C1 to charge through the transformer's primary, then it discharges through the primary. All this activity basically gets current bouncing back and forth through the transformer's primary, which is of course the entire point of the thing.

[rigid plume]

Hi folks

I was wondering if this would work

I'm trying to create an high precision current source to do Deliver currents of up to a few amps with precision.
Allow dynamic adjustments at up to 10kHz.
Maintain a stable output at DC when necessary.

based upon

https://grupos.unican.es/dyvci/ruizrg/postscript/Componentes/REF102.pdf

Thanks

[supple pollen]

I don't see a current output anywhere. Is this just the reference stage? It looks like a DC reference, so presumably you'd be doing your dynamic adjustments elsewhere? What sort of precision are you looking for?

[rigid plume]

Well Im new to this and I need to precisely control the output voltage of the current source, which in turn determines the current delivered to our coils.

Around 5 m.v

[supple pollen]

Ah, you're driving an inductive load? At 10kHz, the voltage and the current are going to be out of phase...

5 million volts? At a few amps? So 20 megawatts or so?

[rigid plume]

As in accuracy +5m.v

0.025 - 0.050 %

[supple pollen]

Heh, yeah, I wondered. Big M is "mega", little m is "milli". Big difference.

[rigid plume]

Haha yeap

[supple pollen]

That voltage reference lashup should be more than precise enough, if it's built with appropriate techniques, components, and shielding.

[rigid plume]

Yeap but I think there could be an issue with current

I need to get it up to an amp and this IC would maybe provide 10ma max

[supple pollen]

Right. Turning voltage to current in a precise fashion is fairly complicated when it's significant current. And if you're driving an inductive load, it gets even tougher.

[rigid plume]

Yeap I heard 😢 #

Is there maybe an alternative IC that you reccommend aswell?

[supple pollen]

It reminds me of the amplifiers driving a magnetically deflected vector display. Since the current through the deflection coils determines the position of the beam on the screen, it has to vary quickly and precisely. While there are power op-amps with more current capability, most of them aren't made any more. The usual approach these days is to have an op-amp driving a power transistor, with a current sensing resistor in the load, which completes the feedback loop of the op-amp. Then the op-amp takes care of dealing with a lot of the deviations from linearity.

[rigid plume]

I see 😅 that is a bit complicated do you have any pdfs I can research on or perhaps ref schems

I'm trying to make this board as quick as possible haha

[supple pollen]

If you search for the Vectrex service manual (it's available as a PDF), you can see the schematics of its deflection system (based on LM379 op-amps).

[rigid plume]

Okey I will look into it

Thank you

[supple pollen]

This page includes a some useful info on the subject https://tubetime.us/index.php/2015/06/27/crt-magnetic-deflection-driver-design/

Including a link to a github with the design files

Based on LM4765 power op-amps capable of driving almost 3 amps into a load

Linear Technology offers this useful application note on boosting op-amp outputs https://www.analog.com/media/en/technical-documentation/application-notes/an18f.pdf

Including an amazing ±120V output stage that uses 12BH7A vacuum tubes!

[sweet cairn]

I've implemented the PAM8302A audio amp chip in my design using Adafruit's version as a reference. Can someone please tell me if I've made any mistakes on the hardware side?

My design

Adafruit's design

My schematic

Adafruit's

Previously I've been told that the schematic itself is fine, but I was wondering if there is something else that I'm missing. I've got a testrun of my board and it isn't working so trying to eliminate all possible faults

The datasheet for the chip: https://www.diodes.com/assets/Datasheets/PAM8302A.pdf

One change I've made over adafruit's design is to use tantalum caps instead of mlcc ones.

[supple pollen]

Polarized capacitors for coupling capacitors might cause issues, I'm not sure. Do you have an oscilloscope?

When you say it "isn't working", what do you mean? What color flames do you get?

[sweet cairn]

Yes I can access an oscilloscope although I don't really know how to use one

[sweet cairn]

I'm not getting audio output.

Someone tested the firmware I'm using on the Adafruit PAM8302A board and the same set of daughterboard pins that I'm using and it's working for them (though I think they used a different buzzer)

[sweet cairn]

'flames'? What is that? What do you mean?

[supple pollen]

One of the worst failure modes is flames shooting out of components

[sweet cairn]

Oh like that lol

No i didn't get any flames

[supple pollen]

It could be your "buzzer" doesn't have enough inductance

[sweet cairn]

Is there any way I can test all this out?

I'd just like to know that the issue is

[supple pollen]

Hook it to an ordinary speaker

[sweet cairn]

I have access to an oscilloscope and a multimeter

[sweet cairn]

Hmmm that I don't have rn unfortunately 😅

[sweet cairn]

Also if this is indeed the issue, then what's the solution?

[sweet cairn]

Btw when I say this, I mean that I get zero output at all. Not even static or crackles or anything, anything that was audible to me at least

[supple pollen]

I'd see if the supply pin has voltage and the ground pin doesn't. Then I'd look for PWM on the output leads with an oscilloscope

[sweet cairn]

So this is the supply pins and GND pins of all components while the device is in operation, correct?

Is that safe to do with capacitors?

[sweet cairn]

Output leads of the IC right?

[woven bluff]

wait until you see a high current diode explode, I couldn't hear anything for 5 minutes

[quartz flume]

I have 8000mAh in parallel same-capacity batteries connected to a Powerboost 1000 Basic. I need a way to power them that won't take 20 hours, and I need to make sure I can power them while they're connected. Any advice?

[rigid plume]

Hi folks I was wondering if this DAC circuit will work?

just needs to precisely control the output voltage of the current source

[supple pollen]

Do you mean charge them? There are chargers that deliver (or can be configured to deliver) more than 500mA. Do you mean charge them while they're powering something? That is somewhat more difficult.

[supple pollen]

That will basically work, giving about 20mV resolution with a 5V supply. Note that R6 and R7 form a voltage divider, so the output will be from 0-2.5V. Note that op-amp isn't specified for rail-to-rail input, so it may behave strangely at low output voltages.

[crystal nova]

Not sure if there is a better place to ask, sorta hw design.
Does anybody have any recommendations for mounting a small solar panel (~4in x 5in) to a wall. Something cheap or DIY would be preferred vs a $20 or $30 solution per panel.

[supple pollen]

Nails? Double sided tape? Is this indoors or outdoors? Is this a bare panel or an enclosed one?

[crystal nova]

outdoor, bare panel

[supple pollen]

Presumably you'd want to add some sort of protection. Maybe sandwich the panel between a couple of sheets of glass or plastic that extend past the edges of the cell and use those to mount it?

[crystal nova]

Interesting thought

[humble swift]

You can get exterior mounting tape that might do it, depends on if you think the panel will be fine uncovered

[wispy blade]

should i be using a schottky, zener, or rectifier diode for preventing reverse voltage/polarity when connecting to an unknown power source?

my intuition wants to use a rectifier/silicon diode but thats really the only thing ive learned about

[supple pollen]

An ordinary rectifier will work, but will incur a voltage drop (about 0.6V for ordinary silicon diodes). The advantage of a Schottky diode is a lower voltage drop. A zener diode has the ability to conduct in the reverse direction at a specified voltage, it's not particularly useful as a series polarity protection diode (but they're popular as a parallel polarity protection diode as they protect against both overvoltage and reverse voltage, but you'll want current flow protection such as a fuse if you're employing a parallel polarity protection scheme).

[wispy blade]

so i assume ill want a schottky for carrying 3.3v to something like a microprocessor to reduce voltage drop

[wispy blade]

also dunno if this belongs here, but if anyone here is experienced in kicad, is there any way i can have two busses which share pin/wire names? (the pcb wants to tie j1's pin 5 directly to gpio12)

[supple pollen]

Yes, Schottky or an ideal diode for that use case

[wispy blade]

great, thanks again!

[supple dagger]

is this qty two 2.2uf caps in series?

[rigid plume]

Thank you very much our application isn't too critical

Another question if you can look at :), is this the correct way to get - voltage

[rigid plume]

yeap

[supple pollen]

Or parallel

[supple pollen]

I'm having trouble reading the IC name at that resolution. But it looks like you have a missing diode and a 0Ω resistor shorting the -15V signal to ground.

[jade wedge]

I kinda wish TI's power designer would just say if those capacitor pairs should be series or parallel.

[pale umbra]

Can someone please remind me of the big chinese parts seller, dont remember if it's taiwan/hong kong/china but they seem to have everything and I think they make an app to make a circuit from which you can order parts as well. I always forget it but they sell low-grade parts that digikey/etc wont even make available and I found a mcu that sells for like 1 us$ and Im curious

[supple pollen]

I strongly suspect parallel (to increase capacitance and lower inductance and ESR). https://e2e.ti.com/support/power-management-group/power-management/f/power-management-forum/654554/webench-tools-tps61021a-webench-tool-query-cout-cap-is-showing-quantity-3-what-does-it-mean

Series would halve the capacitance and double the ESR and inductance.

If you click on the part, you should be able to get to an "alternate part selection" dialog where it gives the calculated limits on capacitance, ESR, ripple current, etc. You can switch parts or use the parameters to look for other possibilities.

[rigid plume]

On the ti webench it shows it being connected to gnd and there wasn't a diode on webench ic is lm25088mh

[rigid plume]

If you do find it please tell me, it sounds interesting

[supple pollen]

Ah, okay, that chip uses a different circuit configuration, so you're fine on the diode. I'm going to assume R16 is 16 milliohms (16mΩ) instead of 16 megaohms (16MΩ). However, R20 still seems problematic, as it's shown as zero ohms and connected from the "-15V" node to ground, which would be a short circuit.

[supple pollen]

The usual Asian parts supplier I see is LCSC, but I don't know anything about an app.

[jade wedge]

Ah, that makes sense. The larger capacitince alone would make more sense from what I know about buck/boost power circuits. You want to ensure a smooth supply.

[pale umbra]

yeah it was LCSC but they dont have any phyplus PHY6222 microcontroller on an eval board. I guess my only bet would be to buy a fake 2$ "smart watch" with it maybe in it. oh well i give up. It was a low power 3.3V ble chip with nice specs so I wanted to try it but the datasheet is only 46 pages so I guess I would have been on my own a lot. Seems well supported (in chinese) but I cant find anywhere to buy it.

[golden wasp]

Hi, I'm building a project based on the Adafruit Feather ESP32 V2. It seems that I'll be needing 12V to drive some LEDs and I plan to use some small stepper motors to rotate some dials to display things. I'd use the USB port only to upload code and debug, during normal operations no USB is needed. Is it a good approach to use a small 12V DC wall adapter and then have a circuit that converts the voltage to 5V and hook that 5V up on the USB and ground pins of the Feather? What would you use for this? A buck converter?

[wispy blade]

what you're looking for seems to be a voltage regulator. not that im very experienced but i think a buck converter is a subcategory of voltage regulators, or would serve the same purpose as one

[pale umbra]

can you give some more details on the led / why they are 12V ?

I feel like if I suggest getting a led driver chip you'll probably say you already have one and that this mght be a led array. Right ?

[golden wasp]

I have pushbuttons with LED-s and it says that you have too hook them up on 12V.

I don't have a dedicated LED controller, I'd just use the PWM pins of the ESP32

I haven't calculated yet how many pins I need but I see that it's easy to run out of pins fast.

[woven bluff]

can something like this work as UVLO?

when your board only has one IC, and the IC only has one GND pin, is star grounding still a thing?

[quartz flume]

That is what I meant, I apologize. I think what I'm going to do is design a PCB for it. Found a nice little chip that allows for 4 batteries in series to be charged through PD while also powering something. Automatically detects battery voltage as well 🙂

[woven bluff]

5V charger's are limited to 2A, so it will take at least 8 hours or so to fully charge it depending on charge curve. Any faster you will need PD. Most modern charger chip have load sharing (some even have back-boost function you won't even need a boost regulator) so you can power. If you want pre-made boards I'm not aware any of those can meet your demand.

also any chip with high power and advanced functions are likely only FN or BGA packaged. It will be steep if you have never designed PCB before.

[quartz flume]

I'm gonna just an MP2760 chip for it 🙂

Should work well. Buck output down to 5V and run into my components

[woven bluff]

I don't think that one has PD negotiation so you will also need to program a MCU to control it.

[woven bluff]

I tested the cheap mmWav radar HLK-LD2410. It's much better than PIR, but eats tons of power...

[alpine zephyr]

Hey folks. I'm having a problem that I don't know how to troubleshoot, so I figured I'd ask here. I'm doing a project with a pyportal which I need to power off a battery. I followed the guide linked below to power the pyportal off of a lipo. I've tried connecting both a battery, a microusb cable, and both to power the pyportal and the same thing happens: the unit starts, and my program appears to load, because it renders my interface, but hangs there. When I power through USB, it works perfectly, but through stemma, it doesn't connect to WiFi and get going. Any thoughts on what might be going on? I can't plug into the computers USB to read the output, because then I'm not powering off the battery charger!

https://learn.adafruit.com/portable-pyportal/overview

[supple pollen]

I've seen that sometimes when the program tries to send or receive something over the (USB) serial port, and hangs if the USB isn't connected.

[alpine zephyr]

Ooooh so removing print statements could do it?

[supple pollen]

If you don't need to change color, you can save pins. You can also use external driver chips to power the LEDs.

[supple pollen]

Yes, along with any serial initialization statements

[alpine zephyr]

Ok. Thank you so much I'll try this

[alpine zephyr]

So I tried this, and I think it's on the right track, but how do I prevent any serial communication with CircuitPython? Like is there a command i can run at the top of my program to cut the serial connection? I ask because libraries I'm using, like for wifi, include print statements for connection failures, etc, and I want to suppress those as well.

[alpine zephyr]

Ach this is not the right channel for that question, sorry! Moving to the help with CP channel

[dry pelican]

almost definitely. I've seen a buck converter blow up because 2 22uF caps in series were substituted for a 10uF

never forget your decoupling

[pale umbra]

anything else to know ? I thought this was the point of buying an IC to do it that they would take care of everything especially ti/analog devices...

[supple pollen]

ICs can do a lot, but switching power supplies are tricky things, and while the ICs and design software can corral some of the quirks, there's still need for some domain knowledge to implement one successfully. It's not quite like the simplicity of 3-terminal linear regulators, where they have almost everything baked in for you (but they too can do nutty things with the wrong input or output capacitor setup in some circumstances).

You can buy packaged switching power supply modules that abstract even more of the design for you. Personally, I'm quite fond of them, as they often contain optimizations (like custom designed magnetic components) that are impractical for me to employ.

[cursive sentinel]

Yeah, switching regulators rely pretty heavily on the external passive components. Physics doesn't allow us to build good inductors or capacitors on a chip.

[knotty tiger]

at least not at power levels that are useful for most applications…

[pale umbra]

It's just that I expected buck-down to be plug and play. I hope the datasheet would say I needed caps for it or anything else I need for it. Oh well

[supple pollen]

The data sheet should specify that in detail. Most manufacturers also offer application notes that explain things more thoroughly

[cursive sentinel]

Internal charge pumps are commonplace.

[knotty tiger]

chip-scale inductors are harder than capacitors, but they do exist

[cursive sentinel]

Yes

And they are horrible in comparison to their discrete counterparts.

[supple pollen]

I suppose it depends on which parameters you need, but in general, yes, they're a distinctly inferior substitute

[knotty tiger]

on-chip charge pumps for EEPROM/Flash programming are commonplace, but tend to only use switched capacitors instead of inductors

[cursive sentinel]

Indeed

(side note: charge pump implies the use of capacitors rather than inductors)

[pale umbra]

is there something like a digital galvanometer chip that would replace an analog one with the big needle and display to use with an LCD instead of having an imprecise needle/ink analog meter? Did a search on digikey but there dont seem to be any (or it's called something else)

[knotty tiger]

like with a normal graphic LCD, or one with custom patterns?

[pale umbra]

with a normal graphic LCD to display the value. This would be for an esr meter and Im hoping to have more precision between 0 and 2 ohms that looking visually

ie: trying to have a better measurement in the zone circled than trying to determine visually where it is between 0 and 2

[knotty tiger]

when driving a normal graphic LCD, there’s usually a microcontroller involved. i’m not aware of any special purpose analog meter emulation chip. you can get stuff like LED bar graph driver chips, but they have less precision than what you want

[pale umbra]

yeah I was trying to find a digital galvanometer chip that would basically convert the analog measurement to a digital one with an ADC in it so I could read it from a microcontroller and display it. But it seems I would have to get a different schematic completely. Anyway question 2) Youtuber is saying to make sure the resistors are high precision(0.1%) and from the same batch. I'm not sure how to make sure of that when ordering from say digikey and when there are many resistors to get:

[dry pelican]

I don't think same batch will matter that much (I think the youtuber is saying that because maybe the same batch resistors will have similar tempcos and drift the same amount). Precision definitely does matter though. Also, if you order from digikey, they should give you resistors from the same batch since they will come from the same reel

[knotty tiger]

you can also buy more than you need, and hand-match using a high-precision multimeter

you still want to buy the high-precision stuff, because they tend to be more stable in terms of tempcos, etc

[pale umbra]

would brands matter here ? From what I can tell most digkey stuff seems to be middle-end at least they dont really sell really cheap random brands.

and I dont need that many resistors for this so might be worth going for quality over quantity since it's for a measure instrument

[uncut briar]

Hello, I am designing a board with the rp2040. I am following this doc: https://datasheets.raspberrypi.com/rp2040/hardware-design-with-rp2040.pdf#page9
In the schematic it shows a 100nF cap at the VCC pin for the flash device, but the device I picked has no indication, from the datasheet, that the capacitor is needed. Is this capacitor required anyway?

[pale umbra]

But they explained 1 page before that such things are decoupling capacitors and must be as close to the power source as before and the diagram looks exactly the same as the page before it. They probably didnt mention it again for brevity

As per the rules from 2.1.2 flash has one power pin so one 100 nF cap ("Because, of this, it is important to place decoupling close to the power pins. Ordinarily, we recommend the use of a 100nF capacitor per power pin")

[uncut briar]

Yeah, but I think that is referring to the power pins in the rp2040 chip, no?

[pale umbra]

It's a rule normally for anything that has a power pin to have decoupling because no power supply is perfect unless a datasheet says otherwise

and to put it as close to the power source as possible

See mostly 2nd paragraph

[supple pollen]

I suspect that youtuber is talking through their hat. While such precision resistors exist, they're not cheap (except from surplus vendors where you can sometimes get them at competitive prices, but you don't get a choice of value) and rarely actually needed.

[pale umbra]

or I got confused with 1% 😄

I'm trying to do the esr meter from QNZ magazine from AARL april 2014 to test caps

[supple pollen]

Measuring ESR is a fairly approximate venture anyway, but 1% resistors are cheap and common and generally more than good enough.

[pale umbra]

art of electronics 3rd edition dont even say why for the 100 nf decoupling caps...

while the quote from above is from my practical electronics for inventors

[distant raven]

I tend to see that 100nF is a common value because it can get you in a rough order of magnitude of closeness to effective decoupling values. This usually ends up being more evident in designs where you see them paired with 10,4.7,2.2, and 1 uF capacitors

There was a video I watched some time ago that demonstrated this, but I can’t find it for the life of me

[pale umbra]

nah they really said either resistors from the same batch or uses 1% (around 0.15$ for each from a good brand like panasonic for throught-hole) or 0.1% (around 0.75$ each for throught-hole)

[distant raven]

But basically it comes down to effectiveness on managing transients and cost

100nF are ridiculously cheap and generally take up very little space

[pale umbra]

also that circuit uses both AC and DC, I finally get to play with AC lol

[distant raven]

Also using a bunch of 100nF in close proximity can sometimes help with DC derating in a smaller footprint than larger packages of similar value. But there is a big caveat to that derating claim as well

If you can do big packages, don’t hesitate

Some people I follow have found the best space to derating ratio is found between 0805 and 1206 MLCC

[supple pollen]

The derating depends on the dielectric, which in turn depends on the capacitance and volume. The upshot of that is, you can get less derating for the same footprint with a taller package.

[wide patio]

https://www.signalintegrityjournal.com/articles/1589-the-myth-of-three-capacitor-values

[supple pollen]

Oh, I was looking for that article a while back! I really like its treatment of the subject matter.

[woven bluff]

I have never seen this in datasheets or application notes, only in schematics made by hobbyist.

[onyx kernel]

Hey i could use some help designing a pcb with an RF Antenna. Im taking this Picocklick C3 as a reference and im wondering why the FEED line gets connected to GND at the antenna. also when i use the same antenna in EasyEDA it only has two pads, the one for the FEED line and the so called Soldering terminal (which isn't connected to anything) But there is no GND Pad besides the FEED line.
The antenna im using: KH-RFECA3216060A1T_V09 (C411566 on LCSC)
This is the Picoclick (my reference) https://makermoekoe.gitbook.io/picoclick-c3/hardware
The first pic is the PCB of the Picoklick the secone one is the antenna on my PCB

[supple pollen]

I don't see where the feed line is connected to ground, so I'm a little lost

[onyx kernel]

its "not" connected on the gerber file/footprint but the antenna only has one solder pad on each side so you "short" the feedline to gnd when soldering it on

[supple pollen]

According to the data sheet, one terminal is the antenna connection, and the other is just for mechanical stability (not connected)

[onyx kernel]

yes thats how ive done it. give me a sec ill try to find a pic of how the same antenna is soldered to the picoclick

[supple pollen]

Oh, I think I see what you're referring to: a single pad bridges from the feed point to the adjacent ground

[onyx kernel]

yes

how should that work and why isnt my footprint like this

[supple pollen]

I'd expect something more like this

It's like the first part of the data sheet doesn't agree with the second part

[onyx kernel]

on a nother note how bad is this rf setup? i know its horrible, but i don't have much space and i don't need a huge range. There is 2 layers of GND between top and bottom

[unique patio]

do you mean there is a trace antenna there? The keepout area for a trace antenna should not have ground planes underneath

or same for chip antenna: https://www.lairdconnect.com/resources/white-papers/how-to-implement-chip-antennas

[onyx kernel]

Wrong wording... its everywhere but under the antenna

[unique patio]

is it a chip antenna? The mfr will have guidelines

[onyx kernel]

but there is virtually none clearance between the antenna and the rest and also the feed line is not optimal. I know the guidelines but how bad is it to not follow them.

[unique patio]

the guidelines are for Reasons. If you don't follow them you can compromise the performance of the antenna, and also maybe allow RF to get into your own circuitry

I'd suggest that you start with the guidelines as an absolute, and redo the layout around it.

without being an expert or experienced you can't know what will work and what won't

[onyx kernel]

Version 5.0 of the pcb🥹 nah joking desinging pcbs is fun 🙂 Thanks

[unique patio]

good luck!

[onyx kernel]

ty

[rigid plume]

Hi folks I have a question on my schem design

This is supposed to get the 10v output from an precise voltage output source (100ma) and boost it to 1A

Does this look like it will work?

based on

[unique patio]

if you are using a design program all the points with the same name will get tied together. I don't think that's what you mean here (all the "+10V")

[rigid plume]

yes Im just using it as an example

[unique patio]

how precise is precise? Why not just use a 10V 1A regulator?

note the caveat in the text you posted

pointing to 4.26

[rigid plume]

it's supposed to be for an magnetic field tuning system

[unique patio]

is it supposed to be 10.0000 V, or is supposd to be very stable?

stability is important for both, but just stable is different than stable and precise

[rigid plume]

well I just need it to be precise within a few 100 mV at most

and it does have to be stable but not as stable

there is room for the stableness

[unique patio]

how is the field tuned? Are you varying the voltage or something else?

I am no expert on this, but it sounds like you just want a well-regulated power supply

[rigid plume]

dynamic adjustments at up to 10kHz.

[rigid plume]

well it needs to be very precise and I can't find solutions that are to deliver upto an 1A

[unique patio]

so are you varying the 10V to tune, or something else

[rigid plume]

yes to tune

[unique patio]

you mean the specs on various voltage regulators are not stable enough?

[rigid plume]

i mean they are not as precise

[unique patio]

you mena they are not necessarily exactly 10V?

there are trimmable regulators

[rigid plume]

I see

[supple pollen]

It seems like what you want is a voltage to current converter. The op-amp configuration you show isn't going to do that, but it's similar to what I was telling you about earlier, with a boost transistor and a current sensing resistor in an op-amp's feedback loop.

[unique patio]

over to you @supple pollen I did not see the previous discussion

[rigid plume]

Thanks

[hollow cosmos]

Is there a way to control the Trinket via Bluetooth in near real time, bi-directionally?

How about with USB?

[rigid plume]

Well I was just looking to boost the output current of the precision votlage circuit

[supple pollen]

Yes, that is a voltage to voltage converter with an output capable of sourcing a fair amount of current, but my understanding is that you wanted to regulate the current, not the voltage.

[supple pollen]

It's easy enough with USB, I don't know about Bluetooth

TI offers this nice writeup on a converter that uses a 0-10V input to program a current from 0-1A https://www.ti.com/lit/an/sboa325/sboa325.pdf

[hollow cosmos]

From an iPhone to the Trinket?

[rigid plume]

Thank you very much I will look into it

[supple pollen]

If you want to control it with an iPhone, you would presumably have some sort of software on the iPhone to do the controlling.

[unique patio]

not a Trinket, but the nRF boards we have can talk Bluetooth to the iPhone. You could use the Bluefruit Connect app

any particular reason it should be a Trinket?

[supple pollen]

Here's a beefier version capable of 5 amperes https://www.ti.com/tool/CIRCUIT060047

[hollow cosmos]

I have 4 solenoids to control from the iPhone (which the Trinket can do with its 4 PWM outputs), just a “set solenoid #3 to push” message. Plus, I need to send info back to the phone about overheating, jams, and other status. Bluetooth seems like overkill.

USB gives me Trinket power too (I still need 4a to power the solenoids, from batteries or wall adaptor), but when I connect the Trinket to the phone, the Trinket’s power lights don’t illuminate, so I figure the USB drivers are not recognized.

The Trinket doesn’t power on from the phone.

[rigid plume]

thanks

[unique patio]

USB devices say how much power they want. The Trinket may be asking for more than the iPhone wants to supply

what kind of Trinket?

classic or M0?

3.3V or 5V classic?

[hollow cosmos]

M0 3.3v

[unique patio]

what kind of dongle are you using on the iPhone?

[hollow cosmos]

It was just a standard lightning to USB type A, then a short female A to the micro B adaptor. Do you think it’s the cabling?

Is the USB host supposed to send out a trickle charge to activate the Trinket’s USB signature return?

[unique patio]

you need an OTG thingie to make the Trinket visible as a device. Above, an Apple Lightning to USB camera adapter, connected to a Trinket M0, which is lit up. Phone is showing contents of CIRCUITPY drive

i don't think the iPhone is going to pass USB CDC serial. Its USB connectivity is very limited, for security and marketing reasons

we normally do all this kind of stuff with BLE

https://learn.adafruit.com/bluefruit-le-connect

[hollow cosmos]

Is the Trinket’s BLE fast and reliable?

[unique patio]

the Trinket has no BLE. You need something like a Feather nRF52840 or Itsy Bitsy nRF52840

The CircuitPython support on ESP32-S3 for BLE is very limited right now. Arduino is possible, but I haven't tried it

CircuitPython BLE projects https://learn.adafruit.com/adafruit-circuit-playground-bluefruit/getting-started-with-ble-and-circuitpython

https://www.adafruit.com/product/3093 might be interesting

don't your solenoids need motor drivers or external transistors?

[hollow cosmos]

If I understand the question, these solenoids have internal springs so there’s no “reverse” direction to energize. I was also looking into latching solenoids that only expend energy when changing states.

A motor driver would be, for example, an H drive, that allows for reversing the direction of travel?

[unique patio]

i just mean a high-current driver of some kind here. The CRICKIT has that on board

sorry, I didn't really mean motor driver

[hollow cosmos]

At 1 amp, I’ll need something heftier, like a TIP120.

Can I drive 4 amps with a cricket and no external transistors?

[unique patio]

• 2 x Bi-directional brushed DC motor control, 1 Amp current limited each, with 8-bit PWM speed control (or one stepper)
• 4 x High current "Darlington" 500mA drive outputs with kick-back diode protection. For solenoids, relays, large LEDs, or one uni-polar stepper

no

so the CRICKIT doesn't solve your problem

https://www.adafruit.com/product/5757 but only 2.5A max

https://www.adafruit.com/product/5648

[hollow cosmos]

That was what I was thinking. Even the 8-way Darlington chip maxes out around 1.6 amps across all circuits, due to heat dissipation issues.

[unique patio]

5648 is hefty enough, you need four. Or just wire up your own. Don't forget kickback diode because they are inductive loads

what do these solenoids control, door latches or something?

[hollow cosmos]

They move tiny “bumpers” to divert tiny cardboard pieces (0.5 grams) moving down a chute into the appropriate bin. Adafruit is out of the mini ones, so I’m trying out the next size (which is marked as “small”).

[unique patio]

would servos work? Less power consumption

[hollow cosmos]

I’m also looking into this new device called a “flap actuator” that is a coil on a paddle that attacks or repels a magnet.

[unique patio]

you can convert a servo rotary movement into a linear movement with the right linkage

not sure it's fast enough for you

[hollow cosmos]

Servos would work, but they require power to keep a position, right? A latching solenoid only uses power to switch positions.

Speed is accomplished by a lever.

[unique patio]

they use a lot less power, and the gearing mostly keeps them in place. If they get displaced, then the motor runs. tiny cardboard pieces are not going to move the positions

solenoids and stepper motors are very powr hungry

[hollow cosmos]

Is there a better tech that I’m missing? (Besides the flat actuators?)

[unique patio]

there are linear servos ("linear actuators")

[hollow cosmos]

The linear servos I found were really slow. I think they were built for precision rather than speed.

[unique patio]

i don't completely understand the use case. pieces of cardboard are going down a chute? you want to sort them into multiple paths?

[hollow cosmos]

Yes, multiple paths. Like a coin sorter.

[unique patio]

so a regular servo could move a paddle to the left or right to choose a path. Any friends who are mechanical engineers?

[hollow cosmos]

Hey, that’s a great idea. With a single servo I could choose between 4 paths instead of using 3 solenoids, if I could get a fast one.

[unique patio]

there are various servo coin sorter projects; haven't found the perfect example in a websearch yet, but you could keep looking

[hollow cosmos]

And with a fifth position, I could block the path totally so the camera has a chance to image the piece (which is what determines which path it takes). That replaces ALL the other parts with a single servo! Brill!

[unique patio]

here's a project I didn't even know we had: https://learn.adafruit.com/machine-learning-with-marshmallows-and-tiny-sorter

not same mechanics, but uses a servo

arduino or CircuitPython for you?

[hollow cosmos]

That’s amazingly similar! Anduino seems overly complicated, and appears to have so many bugs. Is it worth it?

[unique patio]

CircuitPython is easier if you are not already a C/C++ programmer or familiar with Arduino. https://learn.adafruit.com/search?q=servo%2520circuitpython projects with servos

i will leave you to do some reading and cogitating

#help-with-circuitpython if you have CircuitPython questions

[hollow cosmos]

Even with the most advanced IDEs — and I’ve used them all — I still like editing a python file and using print() statements.

[unique patio]

#help-with-projects is more about projects like this. This channel #help-with-hw-design is more about circuit design

and PCB's, etc.

[hollow cosmos]

Thanks so much! Very helpful. You’re employed by adafruit, I suppose?

[unique patio]

yes

[hollow cosmos]

One bit of feedback I have for the company is how hard it is to find info on the website. I must have read 100 pages before I found out what “Express” means, for example, or what Cricket stands for. Or that I could find out what a board does by searching “Introducing …”. It’s so overwhelming that I feel you are losing sales.

A glossary would be more helpful than a search result — a glossary tells people what is possible to search for, that is, what topics are considered interesting.

[unique patio]

Yes, you have to bounce back and forth between the product descriptions in the shop and the product guides on https://learn.adafruit.com. We've tried various top-level guides in the past, but they're very hard to keep up to date. I agree it's overwhelming. But the idea of a glossary is interesting. I will bring it up internally.

[hollow cosmos]

I think you go around the room and ask each person for a list of 10 things they’d like customers to know about the website. Then combine them and add organization.

As for keeping it up to date, leverage the tags (look for patterns), saved pages, and the users. Add a “nominate this page” and have users type a reason to promote it to the glossary. DYI folk are willing to help out — your audience is unique, I think.

[unique patio]

✔️

[sweet cairn]

Hi I'm trying to replicate Adafruit's PAM8302A in my own design but it isn't working. I've already asked help about it here but narrowing down the cause of the issue is tough 😅

This is Adafruit's PCB

And this is mine

One big change compared to the Adafruit board is that I'm using polarized tantalum caps. There was a suspicion that that might be the issue. However according to the datasheet I should be fine, I have the polarities correct I think

A change/mistake I did notice though, is that pins 1 and 3 on my input trimpot are interchanged

So on Adafruit's design, the trimpot pin 1 is audio - and pin 3 is audio +

On my design pin 1 is audio/audio+ and pin 3 is GND (the audio - pin has to be tied to GND if it's not being used which is what I've done wherever audio - is used)

Idk exactly how trimpots work, especially with audio, so is this interchanging a reason why it may not be working?

[supple pollen]

Probably not: that would just make the adjustment "go the other way"

[dry pelican]

do you have a multimeter (or even better an oscilloscope)?

[woven bluff]

are there single-row led matrices?

[left grove]

Isn't that just a Neopixel/WS2812B strip?

(or led strip in general)

The electrical matrix how it's wired doesn't really have to be the same as how it's physically laid out. Like you could but 4 LEDs in a line but wire them like a 2x2 matrix

[supple pollen]

Basically a bunch of LEDs with a common anode or a common cathode? Yes.

[unique patio]

See https://www.adafruit.com/search?q=led+bar

[modest grail]

in eagle is there a way to add a reference picture?

[woven bluff]

humm, I'm making a light printer, but those bars are too wide. I'm looking for something with smaller aspect ratio

[supple pollen]

I don't know what a light printer is, but there are lots of options.

[woven bluff]

it's a display device using your brain as buffer to store vertical line scans.

[supple pollen]

Ah, like a VirtualBoy. I've seen 1x1024 arrays on chips, as well as the ones in LED printers that are similar.

[woven bluff]

back in the day I got my first date by making one of those on a 8051 chip...

now I'm in a similar situation and I think such device may come in handy again.

[supple pollen]

Here's one vendor's description https://www.prpopto.com/wp-content/uploads/2018/08/Monolithic-LED-Arrays-Technology-Overview-20160616.pdf

For a cheap and easy to obtain option, you could use an OLED display and just light one line. An SPI interface one should be able to update fast enough.

[woven bluff]

actually, I don't need that kind of resolution, the last one I made was only 16 pixels high, using a bunch of 0603 LEDs

[sweet cairn]

Yes I have a multimeter and I have access to an oscilloscope (it's not mine but I can use it for a couple of days). I did some basic checking with a multimeter following guidance provided by a friend. It was a little rushed so I'm not fully sure how accurate it was, but we discovered that the buzzer and the inputs on the IC - my friend told me the voltage on them is lower than expected. But again I'm not sure how much we can trust my readings because I did it in a hurry. I can do it again properly perhaps

[unique patio]

We have a number of projects that do persistence of vision: https://learn.adafruit.com/search?q=pov

[fickle patio]

Yes, but it’s not fun. You’ll need a two color bitmap, and getting the scale right can be challenging. https://ceworkbench.wordpress.com/2014/09/03/importing-bitmapped-pcb-images-into-eagle-cad/

[modest grail]

...i...I have a 3d model that I need the buttons lined up with and...i...😿

[modest grail]

need help designing a USB audio adaptor around the VS1053B chip. No I cannot just buy one, need to be built on a board with other stuff

[supple pollen]

A good starting point would be the breakout board

[modest grail]

yeah I tore off everything from a sparkfun board except the audio jack and power circuit.still don't know how to connect it to usb

[modest grail]

this https://www.adafruit.com/product/2264 connected through spi to this https://www.adafruit.com/product/1381tgen connect to a speaker/ headphone jack. Strip it down to whats needed, eliminate and duplications, wire directly to the pi

[celest terrace]

Any recommendations on making a low profile, compact USB-C to Micro USB adapter, male-male. I have a small keyboard I need to run into a Raspberry Pi Zero 2 W and there is not enough room to use a standard cable. Thinking two break out boards wired together might be easy enough.

[tight kelp]

Any idea why Kicad doesn't want to route / connect this differential pair? I have to move the U1 really far away for it to be happy to connect the pads.

[modest grail]

can someone look and tell me if this will work?. The sections marked usb, Ethernet, gpio, and poe are just connections to the pi 5 I'm using.

its an attempt to wire these (https://www.adafruit.com/product/1381 & https://www.adafruit.com/product/2264) together to create a USB audio adaptor and wire it to one of the pi 5 usb ports

I do not have the board files for this, just the schems

[sullen loom]

Hello, is there anyway for the PI pico to drive 1.8v going OUT. Need it for a project im doing to force buttons to be active high which is 1.8v

[latent jungle]

For a distance that short, it probably isn't worth using the differential router. There was an issue a couple of months ago with the DP Width constraint. I found I had to make it 0.15 or 0.25, instead of the default 0.20 for some reason.

[tight kelp]

Oh. Thanks for the info. I moved it away and it had to be moved really far for it to route. I will try fiddling with the DP width.

[supple pollen]

Two ways come to mind. One is using 1.8V for IOVDD, which would make the GPIOs 1.8V ones (there's already a 1.8V regulator available for the core voltage). The other is an ordinary logic level shifter.

[sullen loom]

hm all of my buttons for deactivation need 1.8v LOL

i have 6 1.8v voltage regualtors

better yet, i do need both input and output to possibly work

[supple pollen]

I haven't looked into the details of the Pico, but I'm guessing by bodging a couple of traces, you could disconnect VDDIO from the 3.3V supply and connect it to the 1.8V core supply, at which point you'd have a Pico with 1.8V I/O

[sullen loom]

ill look into it, cuz my power source rn is 3.2 and i think it made outputs into 3.3v still

cuz i did some research and its still magically brings it to 3.3v still even with 1.8v to 5v

[pale umbra]

If you look here, the regulator adafruit seems to use the most, the graph seems to says you could get fixed 1.8 to 5V with it: #general-chat message

The graph seems to be at 3.2V being the minimum for getting 3.3V out and the regulator itself is 1.5V to 6V

[supple pollen]

Oh, I misremembered, the core voltage is 1.1V, not 1.8V, so you'd need a separate source of 1.8V, but you said you had some regulators. Looks like IOVDD is on pins 1, 10, 22, 33, 42, and 49 so you'd have to bodge a few wires (or just replace the 3.3V supply)

[pale umbra]

with 6 options for fixed output

If i may about it as well, is there a gotcha here or I can safely use this adafruit schematic part in any project I regulate:

[sullen loom]

hm, im not sure what the 3.3v supply is

let me go get my pi pico

[pale umbra]

Give me a sec to check if there is a 1.8v option I think there is one

[pale umbra]

ok earlier youj said you want to provide 1.8V out of a pico. Im saying you might be able to use the 3.3V or 5V output pin with a AP2112K-TRG1.8 which provide up to 400mA output at fixed 1.8V from 1.5 to 6.0V

THis would have to be validated by one of the helper because Im not 100% sure of the side effect but that is what adafruit seems to do (schematic above as proof)

[supple pollen]

I don't think the AP2112K can boost voltage, but yeah, it'll make 5V into 1.8V all day long

[sullen loom]

im new to hardware so this is all news to me

[supple pollen]

The Pi Pico regulator is somewhat more complicated, as it's a switching regulator

[pale umbra]

ah that would explain the 3.3V graph that says it provide the input voltage until you are over 3.3V then it provides 3.3V 😦

but we are not boosting here

we are using the 3.3V / 5V pin out to the chip to be regulated down to 1.8V

[sullen loom]

yea

[supple pollen]

Note: the schematic (and a whale of a lot of detailed information on the chip and Pico) are available here: https://datasheets.raspberrypi.com/pico/pico-datasheet.pdf

[pale umbra]

so I think that chip could help you regulate 3.3V/5V out of the pico to 1.8V assuming there are no safety side effects on the pico, and you'd possibly need a current-limiting diode to make sure it doesn't go above your 1.8V device since the chip can output 400mA

for your use case. Think it's worth looking into

For my uses it seems to be the answers to all my problems. so Im going to get 3x of them for 3.3V and 1 for 5V for uses with 5xAAs (NiCD = 5.3-5.5V(dead voltage) to 6V so no boost occuring)

[supple pollen]

The RT6150 that the Pico uses is a buck-boost chip so it can do both, but it's more complicated to use

[pale umbra]

I'll keep that in mind too. Seems easier to reuse the solution of professionnals like adafruit and raspberry pi than try to figure out impementing this on my own

[supple pollen]

Yeah, I try not to re-invent the wheel (too much)

[pale umbra]

Just need to figure out what R1 is for in the adafruit schematic but I suppose it's a bridge resistor between two pins and depends on output voltage and explained somewhere in the datasheet (100k in adafruit schematic)

I like the usb-out buck-down/buck-up kit but it is more expensive and it wont do eventually for motors (neither will that chip but I'll try to find a schematic from adafruit/sparkfun/etc for this)

[supple pollen]

It's just a pull-up resistor to enable the regulator by default (but you can override it by pulling it down if you want to shut down the regulator). If you don't need the ability to disable it, you can just tie the enable pin to the IN pin.

[pale umbra]

also the chip is legged so it's perfect since I'm allergic to tiny smd

[supple pollen]

For powering motors, I just buy modules from Pololu with whatever voltage and current I want.

[pale umbra]

I know technically it's smt but I can just put a glob of solder on the tip of the leg and a wire/header pin where as that is not really possible with a tiny QFN32

[sullen loom]

now i just realized lol

oh wait nvm

it would still boost the power

1, 10, 22, 33, 42, and 49

just need the 1.8v

[supple pollen]

I did design my own converter, it worked, but I ended up not using it, as the smaller module from Pololu was easier to work with!

[pale umbra]

Im interesting in 40A-110A max currrent r/c motors eventually so hobby-grade electronics 5A motor modules dont really do. I realize I'll need to find specialized NiCd batteries for this that do 20C but these exists for r/c. tl:dr: Im trying to keep my options open so I can use the same batteries for electronics projects as well as r/c projects and the same motor controllers if possible

[supple pollen]

I like to keep my options open too, so I've been replacing the connectors on RC battery packs with the same connectors my other projects use.

[pale umbra]

That is also why I buy 50V caps even if so far the max I uses is 12V

[supple pollen]

Good idea. Also the 50V caps won't have voltage derating when used at 12V. I use 50V and 100V caps for a whole lot of things, the exceptions are electrolytics (which you don't want to undervolt too much or they'll deform) and high voltage circuitry (not a problem for most people)

[pale umbra]

Is there a shock hazard with 100V caps and having to discharge them with a resistor even if used between 3.3V to 48V ?

[sullen loom]

i wasnt mentioning the out pin, talkoiig about the GPIO pins

[supple pollen]

No, 100V is just the maximum, they won't charge themselves to 100V without 100V being available

[sullen loom]

if u were referring to that

[pale umbra]

so Id need like a perforated contacts PCB to have a discharging circuitI'd made for that ?

[pale umbra]

oh you are logic shifting, I thought this was more powering. Sorry 😦

[supple pollen]

I don't bother with a discharge resistor for low voltage capacitors.

[pale umbra]

then that is even easier. I could tell you what PFC level shifter madbodger told me to use a year ago but Im sure they remember lol

[sullen loom]

yea, need to be able to set 1.8v for high

and 0 for low

[pale umbra]

anyway Ill let madbodger handle it since he has the context. if needed I can try to find what PFCYYYYY chip/level shifter I was told to use before that also works with bidirectional I2C

[supple pollen]

I don't think Leo wants an I2C shifter, more an ordinary logic level shifter.

[pale umbra]

right, it was this one, but the chip supports much more than what they says and can also be used with bidirectional I2C along with general logic level between any voltages since the low side get configured to the supplied power and so does the high side. I think it support up to 60V or so on the high side despite the presentation they do (I care more about what the chip datasheet says than marketing claims)

[sullen loom]

hm

[pale umbra]

No need to refute my claim if I am completely wrong, dont want to waste your time

simpy presenting options if they might fit the use

[sullen loom]

so im still able out low 0v and inpout 1.8v and 0

Cuz thats all i really need to be able to do for each pin

Read the controllers input, which is high at 1.8v low at 0v

and for my chip to also output 0v and 1.8v

[pale umbra]

I would think so, the LV and HV pin on opposite side is for controlling reference voltage.. I have no idea how you would say that 0V is 0 and 1.8V is 1 (If you are refering to setting the digital frontier values). But it seems the device is generic with 4 logic channes, not necessarily forcing you to use I2C and as far as I know i2C uses varying votage which can be 0V so the reference voltage is the maximum

I'd need to dig to find out the exact chip used but it's pretty common

[sullen loom]

hm

[worldly schooner]

If USB and 3.3v logic aren’t required, the pico can be powered directly by a 1.8v regulator for 1.8v logic. It makes it a bit more complex to program, but requires minimal hardware modification and extra components.

[high scarab]

Hello, I am making this..

But I am facing little problem

There are two problem

1. Intensity of the LEDs at the is low even at full brightness (may bcos I am powering then in series), I just want to know do I have to power them in parallel
2. The power adaptor I was using got fired Idk why (it was 5v - 1amp) I don't know proper way to calculate the how much volt or watt adaptor is required.

Help pls!

[unique patio]

https://learn.adafruit.com/adafruit-neopixel-uberguide especially https://learn.adafruit.com/adafruit-neopixel-uberguide/powering-neopixels.

That's a lot of LED's. You'll want to make heavy-duty connections at both ends of the chain, and maybe at intermediate points. Otherwise there will be voltage drop along the chain. You can probably measure it. If you don't have a multimeter get one; easy to find at larger hardware stores and online.

5v 1A is almost certainly inadequate if more than a few LED's are on at one time. The power supply was probably sagging, and may have broken because you overloaded it.

[high scarab]

Do I also have to use a level shifter?
I am user using node MCU esp8266 which is of 3v!

[unique patio]

it might work, it might not. Depends on the particular RGB LEDs. What are you programming it in?

[high scarab]

I am making my iot project,
I flashed wled ( https://github.com/Aircoookie/WLED ) in it so that I can control it with any device

[high scarab]

I have connected 600 LEDs in series!. And after calculating according to the available on the website!
600 leds x 20 milliamp ÷ 1000 = 12amp

[jade wedge]

Quick correction for you, 20ma will be the current drawn by one channel of one of your RGB LEDs. If you turn that whole thing on at full, you won't draw 12A, but instead 36A because you have 600 RGB packages, and each one has 3 separate LED elements.

[high scarab]

Yes, but I don't think I am going to use it at maximum intensity.

[jade wedge]

Hopefully not. That would be... Excessive.

[high scarab]

But I am still confused 😓

I need a resistance for the data pin, and a capacitor to provide a stabilized voltage.

Idk do I need a level converter or not

I read the article, and I am still confused, do I used 5v-2amp adaptor which I currently have, or do I have to use 12volt -5 amp led adaptor

I am just have poor skill in electrical and electronics

😞.

I don't want to blow my whole project

[jade wedge]

If you're powering them from a 5v source, then officially yes, you need a level shifter. A signal will be considered high if it rises to 0.7*supply. Meaning that you need a signal of 3.5v to be considered logic high. Using 3.3v logic will not meet this. In practice, many of them will still work, but you're operating in "no man's land" where logic is undefined.

You absolutely do not want to use the 12v adapter as that is for other LED strips that are arranged differently than what you have.

The root problem as I see it is that you're simply running too many LEDs for the LEDs that you've sourced. They simply weren't intended to be used like this. As stated earlier, you can definitely do this, but you're going to have to add additional power connections along the strip, and your power supply is going to need to be much better.

[high scarab]

So, how much volt and amp adapter I should use?

(sorry if my questions are looking dumb)

[unique patio]

Learn Guide on level shifting: https://learn.adafruit.com/neopixel-levelshifter

[supple pollen]

They're 5V LEDs, so you want a 5 volt supply. How many amps you need depends on how many LEDs can be lit at a time.

[jade wedge]

Also there's no harm in sourcing a power supply of a higher amperage. The amp rating of a power supply is simply the max it can provide.

[unique patio]

when playing snake, what is the longest snake that's going to be displayed? And do you really want to use max brightness? NeoPixels at full brightness can be blinding.

[high scarab]

This animation

[unique patio]

so more than half

what kind of installation is this for? An art installation? Just for home use? etc.

[high scarab]

I can also display any text with any colour background. So there maybe a 100% chances that I am going to use all LEDs.

[unique patio]

OK, then you need 600 * 60mA, a 36-amp 5V supply (at least), which is large and expensive, and needs heavy-duty wiring

[high scarab]

I am making my iot project, and if teacher got impressed it is going to be installed in the iot classroom 🙃.

[unique patio]

consider using fewer LED's at lower brightness if this is a first project

[high scarab]

Yes, it is my first project

[high scarab]

Ok

[high scarab]

It there any other way I can change the wiring for series to parallel?

Would it help?

[unique patio]

it is in parallel. Each NeoPixel is connected to the 5V and ground rails. But the rails are long, and need to drive a lot of NeoPixels. So you need to feed power at several points with heavy duty wires.

the data lines areserial, but the power lines are parallel

are you using a pre-existing snake program?

You can also break the strip into multiple shorter strips and drive each one, but the software has to take that into account.

[high scarab]

Yes, there are so many pattern written in the program.

[unique patio]

do you have a link to the existing project?

if there was a complete writeup there, it would explain how to do the power wiring

[high scarab]

So, this will eliminate the use of level shifter?

[unique patio]

no, you need a level shifter for each strip

some boards have built-in level shifters: https://www.adafruit.com/product/5650

https://www.adafruit.com/product/3975

[high scarab]

https://github.com/Aircoookie/WLED

[unique patio]

https://www.adafruit.com/product/4537

[high scarab]

I am using this

[unique patio]

ESP8266 has very few GPIO pins, so not very suitable for breaking up the strips

[unique patio]

did you already buy the strips?

[high scarab]

Yes

[high scarab]

I have done this much

[unique patio]

so that is a video of your own project?

[high scarab]

Yes

[unique patio]

ok, so you just need a much better power supply, and you need to run heavy wires from the power supply to both ends of the 600-LED chain, and possibly several points in between.

[unique patio]

is that with the 1A supply?

why was it working at all?

were you running it at low brightness?

did you mount all the leds yourself or did you buy them on a board?

[high scarab]

Nope, It was directly powered by a nodemcu

[unique patio]

so 5V from the nodemcu to the array?

[high scarab]

LEDs at the end (or bottom) are little dim

[high scarab]

3v

It is esp8266

[unique patio]

did you run a power line from the nodemcu to the array, or did you power the array separately?

[high scarab]

Run a power line from nodemcu

[unique patio]

3.3V !?

I am surprised the regulator on the nodemcu didn't give up

[high scarab]

Bcus my 5v-1amp adaptor got smoked while powering

[high scarab]

Me too ..

[unique patio]

ok, don't do that. power the array separately, but you need to tie the grounds together. what country are you in and where do you buy things from?

[high scarab]

India!..

[unique patio]

what city?

[high scarab]

Ghaziabad, near Delhi

[lime herald]

WLED has automatic current limiting, dividing the max current available by the number of LEDs. I don't remember what its default assumption for current is, but that's probably why it's worked at all.

[high scarab]

Yes it have that

[spice zenith]

@high scarab very late to the thread here, but the solution I'm using to solve exactly this problem is to use a 20v laptop power supply, and then multiple buck converters to drop the 20v down to 5v. 36A @5v is 180 watts, so a good starting point would be a 240w (e.g.) Alienware laptop power supply. They're about $100 per unit. 50w to 75w buck converters can be had fairly cheaply from Amazon and other suppliers.
The one HUGE advantage of running a 20v main bus for the power is you need much less beefy cable to move it around, and as a free bonus, the buck converters are very forgiving of input voltage. So if you lose a volt or two along the 20v bus, no big deal.

[high scarab]

But led strip is of 5v🙂

[unique patio]

are you tying the buck converter outputs together? If they aren't exactly matched they might bother each other. Or you separated the strip into separate chunks?

[spice zenith]

Yes. The Buck converters drop the 20v down to 5v

[high scarab]

And + I have to use additional circuit for decreasing voltage

[unique patio]

@spice zenith long discussion in subthread

[spice zenith]

No I do NOT tie the buck outputs together. I explictly separate the 5v power segments, and just run gnd and data the whole length.

[high scarab]

Why 5v adaptor is of 2amp only

😞.

I just checked all the charges!.

[spice zenith]

moving to thread.

[minor flax]

I believe this is a fileter that is inline with the headphone jack on the right, just wondering what I should be looking for as a replacement.

[supple pollen]

That circuit position looks a little like a common mode suppression choke, but that would be unlikely for a headphone jack. It also looks like a transient protection module, but that's unlikely too. It's a weird looking circuit arrangement in general. Do you know what any of the other components are? The GJ8Y ones look like diodes or capacitors. The black ones look like resistors. The rightmost ones look like possibly capacitors.

[hushed smelt]

My guess is an isolated resistor pack. Would have to do a resistance test to find the exact value as it’s too small for etching the value in the face.

[jade wedge]

To narrow things down, I'd at least probe it out with a meter to see what positions those traces are going to in the connector. I'm assuming that there are one or two more on other layers since your typical headphone jack has at least 3 or 4 connections total. More if there's a dedicated presence sensor.

[minor flax]

Top and bottom are left and right audio I've never been able to identify the GJ8Ys but in circuit they read as 240uF? The filer has 0ohms on the bottom 450 ohms on the top :S. the black components read as 100K ohm resistors in circuit, can't read the far right ones

[woven bluff]

is there a reason to avoid metallic cores?

ferrite cores need way more turns to have the same inductance

[supple pollen]

When you say "metallic", do you mean powdered iron, or laminations, or what? Ferrite has a lot more permeability than powdered iron. Laminations can suffer from circulating currents which makes them lossy.

[dry pelican]

Iron cores are also not as good at high frequency

[onyx kernel]

would this be the correct way to connect a dpdt switch?

[supple pollen]

The connections are correct, but USB might not enjoy that

[cursive sentinel]

USB1.1 won't care at all. You might be able to eek by with 2.0 as well.

[latent jungle]

what are you trying to accomplish

[onyx kernel]

well i have 2 mcus on the same pcb with one usb hub. and i want to choose which one to program with the switch. I think a usb hub would be overkill

its a esp32 C3 and a esp8266 btw. The esp8266 is a sort of base station, that the pcb with the esp32 C3 plugs in to for charging and programming. The esp32 C3 also connects to the base station via ESP-Now to transmit its data. Till now i always programmed my esp boards with a nother esp and jumper wires, but for this project i wanted to get fancy 😄

[rigid plume]

Hi folks I have a question on my custom igniter design schematic for an mini rocket

Will this work, the only question I have is if I've connected the SPST correctly https://www.mouser.co.uk/ProductDetail/Carling-Technologies/2M1-SP1-T1-B1-M2QE?qs=fLlAb0goH0dj4%2FU5qxsPsw%3D%3D

[supple pollen]

The schematic symbols don't make it clear. You could theoretically have a shared power supply for all of them. One thing you want to watch out for is current capability, a lot of ignitors draw a lot of current and you want your switches to be capable of it.

[rigid plume]

Thanks

This is the SPST

I haven't really worked with switches

I think the 3rd pin is for gnd? not 2nd

[supple pollen]

That appears to be an SPDT two-position switch. In one position, it connects pin 2 to pin 1, in the other position, it connects pin 2 to pin 3. So that schematic symbol may be wrong (or the pins may be numbered differently).

[rigid plume]

I see, thank you

[dry pelican]

Is this a good way to route the PGND and AGND signals separately and then connect them at one point (under the main buck-boost IC)? I'm working on version 2 of my Zeptodrive constant current LED/Laser diode driver. Version 1 worked but had some stability issues, so I was thinking that separating AGND and PGND like this would help lower noise and improve stability in addition to some other changes.

and schematic

[celest terrace]

I have a project I am working on that uses a Raspberry Pi Zero 2 W. This only has a single Micro USB input but I have two peripherals, both USB-C, that require power as well as data, and am very limited on space within my enclosure. As far as I’m aware this requires a hub. I’ve made my own custom usb cables without issue but not sure how to go about doing 2 (USB-C male) into 1 (Micro USB male). Any advice, or hardware recommendations? I have the breakout boards/plugs already just not sure what needs to go in between to keep the data all correct.

[nocturne stag]

Hmm, I guess this is just possible, if you also create a switch, that can switch between the cable, else I cant imagine how that should work. Or you have to connect one of the Cables to the pins of your Raspberry pi Zero 2 W. But I dont know much about it, so dont take me too seriously.

[unique patio]

Is this too big? https://www.adafruit.com/product/3298

[celest terrace]

Unfortunately too large.

[fervent lance]

Does anyone know the hole size dimensions to mount the analog joystick and the same for the thumb stick?

I'm in texas and for some reason all of our shipping services are spazzed out and deliveries are now almost a week late (this is why I ask, otherwise I would use my mics to figure it out)

Here are the links to the 2 products I mentioned
https://www.adafruit.com/product/3102
https://www.adafruit.com/product/512

I'm trying to finish designing my front panel so I can send it off to get machined

[nocturne stag]

The sizes of the first one: 2.7" tall x 2.1" long x 2.1" tall
sizes of the second one: 1.5" wide x 1.5" long x 1.25" tall

I even have a tip for your future from my life experience (even im not really old) sometimes it helps to scroll down to the technical details of something you want to get information about. EDIT: Ouch, trying to gibe a little bit, and totally messing up… gonna learn from that

[fervent lance]

yeah....that's real smart...I've even seen that information. However, like I asked, I need "the HOLE size dimensions to MOUNT" them

not the overall product dimensions

[nocturne stag]

Oh, I guess I pretty much messed up here.

[fervent lance]

no worries

[nocturne stag]

Im gonna try to measure it, looking at the Pictures and then comparing to the board sizes.

[fervent lance]

I thought about that...and I'm sure im close....but I need actual sizing

Like, the problem with the thumbstick will be, if the hole isn't big enough, It will hinder full range motion

the opening on the xbox thumbstick is 0.765"

[nocturne stag]

Okay, im a little bit confused now, but you are talking about the size of the screws i guess (sorry im not really good at english)

[fervent lance]

No....I am having a custom panel cut to mount the controls to. The joy stick and thumbstick are mounted from below. So there HAS to be a hole cut to allow the sticks to pass through

I need to know the sizing and tolerances for those before I send off the design to a machine shop to have the panels fabricated

[nocturne stag]

Okay, then I think I cant help, im really sorry.

[fervent lance]

I just thought I would ask here, if anyone has those controls and tell me the "stick hole" sizes I should use.

No problem, thanks for trying.

[heavy jasper]

For at least the 3102, I looked up the manufacturer part number shown in the photos and found this: https://module-center.com/administrator/files/UploadFile/D202X-R4.pdf

Which seems to include a dimensioned drawing.

And even if not a completely dimensioned drawing for the size of the stalk (in case you’re mounting under a super thick panel) at least enough that you might be able to trace over it in CAD with some known dimensions and then get some info from that.

For the thumb joystick, I can’t guarantee but it sure looks similar to: https://www.sparkfun.com/products/9032 and they appear to have 3d models linked.

[fervent lance]

excellent! I will check those out! Thank you!

[fickle quarry]

Hey folks, Im working on an update to my controller and despite a few revisions Im not able to get the BMP388 barometer working. (I've had it working with a standalone Adafruit STEMMA QT, so dont think its code related)
I'm on the RP2040 and have a vibration motor driver working fine over I2C but despite that the BMP388 isnt found over I2C. Assuming the soldering is done fine (JLCPCB did the assembly) is there anything you see off with my schematic?
Much appreciated.

[onyx kernel]

Just in theory would this work in that it turns off both poles?

those are N-Channel mosfets

[fickle quarry]

BMP I2C errors

[olive sierra]

Hi everyone ! I have a noob question on easyEDA, whatever I connect to via, it get an error in the DRC "Object 1 is disconnected from other objects of the same network". I don't understand forum posts talking about it and what causing it

[olive sierra]

Playing with the menu and setting the object into the same net seems to solve the issue, I tough that wouldn't matter or be automatic

[onyx kernel]

check which net those tracks with a yellow x are. looks to me like theyre in the wrong net

[dim light]

Hopefully I'm asking the right people for help. I built an a computer controlled silicone adhesive dispenser and a turn table. I use a linear actuator, motor driver, momentary push button, raspberry pi 4, touch screen for the silicone pump/control. The turn table is simple, it's built using a geared electric motor, knob pwm controller, and a switch that switches rotation direction. Both systems are powered by separate external 12v power supplies.

Now here's the problem. Simply plugging in the momentary button without pressing it activates the silicone pump. This only occurs if the silicone pump and computer are near the turntable. If you change the direction of the turntable that will trigger the silicone pump. I have double and triple checked everything is grounded properly. I think I'm dealing with electrical noise but I don't know how to fix it. Thanks in advance for the help.

[jade wedge]

If you don't have access to an oscilloscope, I might suggest trying to temporarily change the pin you're using to read the button into an analog in, and seeing what it reads as. If you're failing to pull it up/down properly, you may find that you're getting a voltage that's just barely close enough for the system to consider it a valid digital signal. If that's the case, you probably need to adjust your pull up/down resistors, or possibly add shielding to the button wires, or possibly just adding a ferrite bead to the wires.

[dim light]

Yes I have access to an oscilloscope. I was under the impression the raspberry pi 4 doesn't have an analog pin. This problem frequency occurs when the button is plugged in but still occasionally happens without the button.

[supple pollen]

No, that's not really going to work. Not only will Q5 not turn on because you need a P-channel MOSFET to switch the high side, but its parasitic diode will conduct in any case. You don't need to switch both sides.

[supple pollen]

Seems like there may be a missing pull-up/pull-down resistor so any random noise triggers things. Additionally, you could have a ground loop problem if your 12V supplies and your Pi supplies disagree on the zero volt reference.

[jade wedge]

Yeah, sounds like the button wire is acting as an antenna.

[fervent lance]

would you say that it's a good idea to use about a 220ohm resistor on every 10k pot I'm using (pulling analog values from)?

[supple pollen]

It depends on the circuit position. Sometimes you'll find a fixed resistor in series to prevent going all the way to zero ohms in any direction. Sometimes you'll have a series resistor to the wiper to limit current (but if it's just going to an analog input, that's not terribly useful unless you're guarding against the input being switched to being an output). Or you could wire it as a pull-up/pull-down but a low value like that would swamp the effect of the potentiometer, so that's probably not useful at all.

[fervent lance]

I didn't think it would have been useful for my application, you just confirmed what I was thinking with that last sentence

[olive sierra]

Thanks ! it was the issue !

[onyx kernel]

Thought so. But how could i do it? I played around a bit in circuit JS but i couldn't get anything to work. Oh and just in theory i know it wouldn't make much sense for a led ;D

[supple pollen]

The usual approach is a P-channel MOSFET in the hot lead, driven by a level shifting circuit for its gate

[low anchor]

what sort of processor would y'all use to replace a greenhouse controller? I've been designing with the ESP32 but I'm worried about its capability (it needs to sync via WiFi regularly to a base station, while managing all sensor & user inputs) as well as long-term maintenance of the software (I'm a university student and I graduate in a few months). Based on that I've been thinking about redesigning with something like the Pi Zero 2 W, but I'm wondering if the microSD card would be an issue for long-term durability since I've had a lot of poor experiences with removable flash storage

[pale umbra]

I dont see any argument against the capacbility of an esp32 here 😄

In fact it's low power for uses as a data collection IoT device

vs an SBC that will uses much more power

[low anchor]

The other thing is we might want to do some fairly complex models for driving all the greenhouse controls at the cheapest cost, and that'd be better on the zero's processor

[pale umbra]

But it would be it sending data to a base station, not being polled

[low anchor]

instead of requiring the 18+ boards to sync with the base station and having the base station compute everything

Plus my other thought is having a command line accessible on the pi zeros would make updating easier as well as connecting other 3rd party sensors via wifi or bluetooth

[pale umbra]

one doesnt prevent the other

[low anchor]

Circuitpython also didn't seem like it was a very great experience on the esp32, it has a lot of weird quirks since it doesn't just show up as a USB drive. maybe I do a pi pico instead

[pale umbra]

My esp32-s2 are all connected to my pi 4 8gb 24/7 , the pi acts a base station, the iot datalogger are battery operated

and I use arduino mostly since I dont know python much and because of the large amount of library

[low anchor]

hmm

[pale umbra]

their roles is to scream if something happens, otherwise log it and send it by wifi every minute or so so maybe not your use case 🤣

[low anchor]

wouldn't writing custom drivers for whatever the 3rd party sensors are be a PITA on an esp32?

[pale umbra]

custom drivers ? no they are all supported as sketch on the arduino ide/arduino C

and they are all very simple, I have 4x qtpy esp32-s2 and they only have a single sensor, and a button, really simple stuff, they send the data and then delete it on ack

[low anchor]

i mean like if there were a 3rd party sensor that's already its own complete product

although I suppose that's a weird use case we probably wouldn't run into, considering if it supports wifi we can just connect it to the base station that way

ugh now I can't decide between circuitpython and Arduino 😂

oh I originally leaned towards CP because it would be easier to make the esp32 download a file from a server and run it, I'm not even sure that's possible on arduino

so I could have each controller dynamically adjust what code & software definitions it uses for the connected hardware, because each room of the greenhouse is different

[low anchor]

what happens if I connect USB-C to a feather board while it's powered from somewhere else too

[dry pelican]

It depends on where the external power is coming from

[pale umbra]

some feather will try to recharge battery connect to them on the bat pins with usb also connected, might be bad if it try to recharge alkaline batteries

[low anchor]

hmm. I'm trying to figure out how to power a feather board in my design. I have a 12V power supply since that's the voltage used to switch the relays

[pale umbra]

12V to usb converter/dongle for simplicity sake

[low anchor]

what if I need to plug in USB while it's running to access a serial monitor or something?

I'm thinking maybe I just route 5V to the USB pin (since my old design for the esp32 already had a 5V regulator) and then make a USB cable with a diode on the power line to prevent backfeeding haha

or, this might be a dumb question, but can two devices communicate over USB without sharing a 5v rail?

[hushed smelt]

It's easier to connect pins to netlabels and then have the netlabels be the connection for each pin. Just make sure the netlabel is the same name for both pins you want to connect together.

something like this

the autorouter will do the rest as long as the netlabels have the same name

well technically those are net ports which must always end on a physical pin. netlabels you can attach to a line between net ports.

but you can use netlabels or netports if it's on a pin

[spice zenith]

Pi Pico W would be a good choice. Iit has the WiFi on board, plenty of CPU horsepower, and is relatively inexpensive. It actually achieves all three of "Good Cheap and Fast". 😄

[hushed smelt]

here's an example of netlabels instead of netports.

when you import into the pcb the rat lines will automatically connect to the correct place, all you have to do is either use autorouter or connect the the traces yourself manually.

[spice zenith]

what if I need to plug in USB while it's running to access a serial monitor or something?
I solved this exact problem with a custom soldered USB cable.
Take a cable that would normally connect the MCU to a computer. Cut it in half, and sort out all four wires: red, green, white and black. Repeat for a second cable, but only keep the "A" end, save the other part in your random parts box.
Then from the two parts of the original cable connect black, green and white together, but trim off the red from the cable attached to the "A" end.
Take the other "A" cable, and connect its red to the red on the "B"/"C" , and connect it's black to the common black joint you already have. Trim off the green and white from that.
Shrinkwrap everything to keep it all nicely insulated, and you're all set.
One USB A provides power, plug that into a wall-wart and it'll run fine. Then plug the other USB A into a computer when you need to debug.

[supple pollen]

Yes, they can: they just need to share a zero volt reference.

[hushed smelt]

@supple pollen i've used easyeda exclusively for about 2 years now. if anyone has a use question feel free to ping me. actual electrical engineering i'm not good at but navigating easyeda i am.

[supple pollen]

Unsure why I'm tagged here

[olive sierra]

yes I use net labels in my sketch. I don't use auto routing however

[woven bluff]

how can I find shift register that can drive LEDs? the most powerful 8 bit shift register from TI can only output 64ma, but each bit will have load of 20mA (160mA in total)

[supple pollen]

I think the TPIC6B595 (which is a TI part) can sink up to 500mA total

They also offer shift registers specifically designed to drive LEDs such as the TLC6C598

[crystal ermine]

Does anyone have experience using wemos S3 Mini boards with wemos D1 mini shields? It looks like to me the pins line up on the outer row of the s3 mini for at least power. But the rest of the pins look like they’re general purpose gpio so presumably should be mappable to i2c, spi, and uart as needed?

[quick field]

Hello All! I was hoping that someone can help me here I'm working on a project and using Adafruit ESP32 Feather V2. I'm looking to get a screen, I found this one that seems like it will work perfectly with the uC I have https://www.adafruit.com/product/4650 since I saw them use it with the temp sensor in the uC datasheet. I want to show ECG graphs on the display but I'm afried that it won't be able to show it very well. Do you guys have any other suggestions for a display I could use that isn't bigger than 15x10x5 cm?

[pale umbra]

@quick field do you currently have that MCU because it's overengineered for a 2 colors display so I advise you dont spend on it just for that display or you get a color tft to take advantage of the big memory on the mcu

[quick field]

yeah I do have it ( Not physically but it should arrive today)

[pale umbra]

currently the mcu can easily handle the display (128x64x4)/8 and even store several frames in memory so that's not an issue and your choice of display is not limited by the mcu so you can pick display pixel per pixel or with a frame

so I advise to make sure your mcu can run the display in the pogramming language you want to use instead and to have it as easy as possible for the interface (ie: whatever interface your display use - avoid exotic ones like hdmi or special spi) and that the library you will use in your programming language can display the type of things you want to draw easily (lines/circles/stored pictures/etc) like many circuitpython compatible displays/mcus dont support outputing an image directly from a bmp like file

Im glad you checked with us as well, saw a post recently on reddit with the same questions except a feather m0 and I didnt have the heart to tell them their mcu cant handle their display and they have to get a much better mcu like yours

[quick field]

So far I have been planning to work with Arduino just because it's easier to use for other parts of the project. As far as the connections I haven't looked at it in depth. I'm new to using this kind of stuff so learning along the way XD

[quick field]

Oh god that's terrible! I would cry honestly XD

[pale umbra]

you jknow the display is extremely small right ?

like dice sized ?

[quick field]

!!! I did not...

[pale umbra]

I was disappointed the first time I bought a 1.3" diagonal display and immediately replaced it with a 4x bigger tft display because it was too small and only monochrome

"These displays are small, only about 1.3" diagonal" that is very small, as per pythagorus that mean one of the side is smaller than 1 inch

on a computer display that is about the size of a 64x64 icon on windows at 1080p. Not sure what ECG is you want to do is exactly but it sounds like graphs and that is certainly too small for graphs

[quick field]

yeah I saw that.. I for some reason thought it would be the size of my thumb...

[pale umbra]

ah yeah the width sounds right, I was more commenting on the height of it . You do seems to know what to expect then

you will need that library for it for arduino: https://www.arduino.cc/reference/en/libraries/adafruit-sh110x/ . mde by adafruit so all seems fine so far

[quick field]

oh okay I see what you meant nowXD

[quick field]

ohh that's nice thank you:)!

[pale umbra]

only difficulty you might have is to the i2c adress the tutorial dont mention any fixed one but it would be worse without adafruit quality and we can help with that in the arduino channel. Just dont be surprised that you may have to fiddle a bit to find the adresse in thee code at first before you can use the display

btw Im thinking of getting the same thing but with the big tft color screen because I dont want a size / colors limit and Im kinda sick of fiddling with smaller displays on low memory displays so this is a topic I have checked out a lot recently 🙂

[quick field]

oh I heared about that before. Will look into it more once I start coding it!

[quick field]

that's cool! is it for a specific project you've been planning?

[pale umbra]

nah I just think it is a foundation component just like my TV remote/ir receiver and the round TFT I bought from waveshare is hard to use and only has libraries for rpi and arduino(but arduinos also require like an sd card etc). This replace buttons on the breadboard for so I basically have around 50x2 buttons on my remote to operation my project

I plan to use the round tft for a permanent projecvt so I need an easier to use tftf for general diagnosis/showing project informations at runtime and for future proof Id rather have the largest simple most colors screen I can get

I bought many cheaper things before and it made me realize that adafruit is almost plug and play and make projects much more easier

[potent fern]

Usb c pd controller ic's

[willow tartan]

Hey guys
I created a schematic for STM32U575VGT6 and I want to know if am I missing something (especially on the highlighted parts, I saw in a tutorial that VDDA is connected with VDD in the schematic(https://www.youtube.com/watch?v=nkHFoxe0mrU), Do I need the Ferrite Bead? , and the HSE doesn't have capacitors in the schematic but in the datasheet I saw I table that says more details but I don't know what to use). Can someone help me with this schematic?

[latent jungle]

If it’s a ceramic resonator, you don’t need load capacitors.

Ferrite bead is probably there to filter noise from the digital rail.

[fluid dawn]

I am building a circuit and I need to have a switch control if my RP2040 D2 pin is connected to ground or not (This is the short version, it's more complicated than that).

For more information (feel free to skip this) I need to sometimes have my RP2040 store data internally (following [this tutorial]<(https://learn.adafruit.com/circuitpython-essentials/circuitpython-storage>) ) but I only need it in this mode when the entire circuit is running on battery power. So, I'm using the main power switch to toggle this data entry mode (when D2 goes to ground).

Back to the simplified circuit! So I built this here and it WORKS! However, after about 10 seconds, I start smelling some wire burning smell. So I got scared and disconnected everything. Am I missing something here? Like do I need a resistor in series between the 3.7V battery and the transistor emitter? If so, after or before the common grounding wire between the RP2040 and the battery circuit?

[supple pollen]

You need a resistor in series between the battery and transistor base.

[fluid dawn]

Ahh, okay thanks

How do I select a value?

[supple pollen]

You can do it the mathy way or just use 1k

[fluid dawn]

@supple pollen Thank you for the info. Just curious though, what is the math way? I have used Ohm's law for something like finding a resistor needed for an LED when I know the voltage required (from data sheet) - but this is just a circuit through a transistor. So do I just find the max voltage the transistor can handle then calculate resistance value from that?

[supple pollen]

It's current, not voltage. You take the supply voltage (3.7) and subtract the transistor's base-emitter voltage (0.6V or so) to get the voltage drop, then divide by the current you want (1ma or 0.001A or so) to flow into the base (which could be the current you want to flow into the collector divided by the transistor's gain)

[fluid dawn]

Okay, thank you again.

[empty hollow]

Hi folks, hope you all are well. I have a sensored brushless dc motor from a ryobi impact driver that is having a strange fault. It doesnt run if its in a specific angular position. I probed the hall effect sensors and found that for both forward and reverse, there are two different set of combinations that make the motor just not spin. All other angular positions make the motor run just fine at low and high speed. I am not sure what could the fault be becuase it seems like such a nuanced problem. Any idea on what could be the root cause?
Happy to share more information as well, just let me know. Any help would be greatly appreciated.

[supple pollen]

Are you getting pulses from all 3 hall sensors? Do all three windings have similar resistances?

[hollow cosmos]

What is the proper forum here to ask about using a J-TAG (ex: J-LINK or SWD) with an ESP32-CAM board (which has most of the GPIO pins hardwired to the camera)? I don’t know if Adafruit sells these J-TAGs, or has learning modules on them, nor which unit to buy — I think I’d need the USB variety because of the lack of GPIO pins. I don’t think any other ESP32 board would help, since it lacks the camera and uses pins differently. Segger sells them for $1000s, but there MUST be a cheap ($15) version, too?

[knotty tiger]

debug probes (the kind of device you’re probably looking for) do tend to be rather expensive. SWD uses fewer pins than JTAG, so that might be a better choice for a board with lots of used GPIO. there is an “Edu” version of the J-Link (that Adafruit sometimes sells) that’s a lot cheaper if you’re a hobbyist, student, or only using it for non-profit purposes

[unique patio]

https://www.adafruit.com/product/3571

also https://www.digikey.com/en/products/detail/espressif-systems/ESP-PROG/10259352

much cheaper, Espressif-specific

[fading escarp]

Any soldering gurus able to answer an awkward question?

[tough matrix]

Not sure if I am guru, but I had to deal with some awkward soldering issues myself, so feel free to ask.

[cursive sentinel]

If it smells like burning flesh, you are holding it wrong.

On a more serious note, what's the question?

[fading escarp]

Right, so...

I have a 2012 Kia Soul that decided to blow apart its door-lock relay (the door lock circuit is shorted and continually trips fuses. ive ruled out everything except the master door lock relay. The problem is that Kia, instead of doing something reasonable, decided to use a through-hole double relay (an EX2-2U1S) so i went on amazon and got one. The problem i have with replacing it is that kia decided to piggyback two boards. the fuses stack on one of the boards, all of the relays and harness connectors are stacked on the other board. so this presents the challengs: There are about 100 pins holding the two boards together, and the bottom side of the board the relay is attached to is UNDERNEATH the piggyback board...

si i need to find a way to desolder this relay, and solder a new one in... and i either need to do that from THIS side of the board, or i need to undo allllllllll of those bottom-side pins...

Was hoping someone with much more experience than Mr could shed some light how I'm supposed to get into this thing, haha

I have a hot air gun reflow tool, and a decent solder iron with a good number of different tips, so tools shouldn't be so much of a problem as much as just the physical reality of the component

[unique patio]

if there is room for the relay to be a little higher, you can cut the pins of the old relay and solder the new relay to the pin stubs. Before doing any surgery like this you should try testing the relay if possible to see if it is really fried or not.

[fading escarp]

Near as I can tell there is a short to ground across the relay from the fuse blade. I spent some time with a multimeter trying to tease that one out, although same problem. The relay legs are pretty well protected from the top side.

Worst-case scenario I can just mount the new relay slightly higher enough to work a fine iron-tip underneath.

[fading escarp]

Yeaaahhh, that will do it. Lol

[hushed smelt]

I would use some snips to destroy the rest of the relay being careful not to nick the PCB in the process. Once down to bare pins use a soldering iron and solder sucker to clean up the pin holes. Oh but resoldering from within the sandwhiched PCB's.. yeah that's going to be a problem.

[fading escarp]

So actually, with the white body scraped off the side of the guts, I was able to patiently heat up the board enough to get it free

And and then also able to patiently coax the new relay back down

[hushed smelt]

Can you reach each pin with a soldering iron?

[supple pollen]

Sounds footery

[hushed smelt]

without damaging any other components in the pcb sandwich?

it's near the edge of the board so you've got that going for you.

[fading escarp]

Unfortunately, no. 😦 I went with "flood the pad with some solder and then seat it back down"

[hushed smelt]

with a heat gun? you'd have to be quick but that sounds doable if you have a rework station.

[fading escarp]

To give more spatial context:

[hushed smelt]

and not burn your fingers

[fading escarp]

That's what needle nose are for :p

[hushed smelt]

oof that's much closer than i thought it would be. yikes.

[fading escarp]

Yeah, this design is the dumbest thing I've ever seen

Lol

Thanks kia

[hushed smelt]

does it check out?

that's one heck of a repair

[fading escarp]

So the only two legs with a short to ground now are the ground legs of the coil windings, which makes sense.

[hushed smelt]

if they're ground legs then not really a short to ground that's just continuity?

[fading escarp]

I would expect a leg of the coil goes to ground, and the other leg goes to the trigger pin from the wiring harness that leads off to the Body Control Unit

[hushed smelt]

relays come in all shapes and sizes, can't guess about what a pin does without a diagram. typically the gnd legs are a different color metal but not always. making assumptions based on working with other relays is never a good idea. always get a diagram and confirmation by testing it.

great job with the soldering though, impressive and well done.

[woven bluff]

that part (and family) is 5V only and vendors generally don't include driving capacity for shift regs in the listing.

maybe a 7-seg driver is better, they generally have resistor programmable current limiting and PWM dimming

[supple pollen]

If you're driving LEDs, an LED driver can be a better bet (especially the ones that regulate the current for you). Vendors generally DO include driving capacity. I didn't know 5V was a problem for your use case.

[woven bluff]

they do list in manufacturer sites but not retailers such as digikey

and digikey does not list bus clock speed for LED drivers

[knotty tiger]

DigiKey often has manufacturer datasheets available

but yeah, DigiKey also often doesn't make all relevant specs available for search

[woven bluff]

I checked 10 of those and they all are like (absolute max) rated total output +-20mA

maybe LED driver is the way to go, I just need clock fast enough to make POV display

[supple pollen]

You want more than 20mA and a POV display?

[woven bluff]

20mA is total load, each chip has 8 leds

[knotty tiger]

a shift register might also be too slow for some POV use cases

[woven bluff]

I have 32 RGB LEDs so 96 bits

unmuxed for max brightness

I'm thinking about 12x8bit SPI-in parallel-out LED driver

st also has some 16bit 90mA 30MHz drivers for 1$

or I could go for a dot matrix driver with more than 96bits for minimum component count

[supple pollen]

20mA per LED or 20mA total load?

Shift registers can be plenty fast for POV, you can run many of them at 10MHz over SPI

[fading escarp]

🫡 i do have a diagram for the EX2-2U1S. its basically two independant relays that sit oriented opposite of eachother.

[woven bluff]

IF 20-25MA, peak IF 60-100mA, per bit, normal shift reg typically have 20 mA total load capacity, so not remotely enough

[knotty tiger]

so… TPIC6B595 with level shifters on SPI?

[woven bluff]

certainly could work, but I like to use newer parts when possible, like STP16CPC26PTR

[supple pollen]

That's a good choice, it's a current regulated LED driver

[woven bluff]

I hope my time on KiCAD and wallet on digikey/JLC will pay off with my impression on the girl..

perhaps I should make something fancier instead, like a working Divergence Meter that measures the fine structure constant of our universe

there has been advances in optical chips, so it might be possible to measure fine structure constant to eight digit accuracy on a chip.

[supple pollen]

That would be awesome to the right person.

[woven bluff]

it a shame Makise Kurisu does not exist in our world line (0.0072973α). Our world line is too close to 0.000000α so CERN will probably take over the world in 10 year🧐

[supple pollen]

The solution to that is obvious: switch lines. There's a reason I label my PCBs with "FG" and a number...

[fading escarp]

Update, it's fixed!

[minor flax]

hello all I am on a stupid mission to turn this pedal into a smd board. The KiCad schematic is the original trace and then the vero board is basically the same without the digital switching. I have a few questions about the other variances.

1. There is a Diode added between the 9V battery and R1 I assume this is just for protection?
2. D5 is missing I don't know why this wouldn't be included?
3. R19 is missing, is this because the digital switching was removed?

[supple pollen]

D5 is a parallel reverse voltage protection, the added diode is series reverse voltage protection. They both protect against reverse voltage, but in different ways. When you say R19 is missing, is it replaced with an open or a short?

[minor flax]

It's open as far as I can tell there should be 100k resistor between the 100nF at D:10 and Level 1