help-with-hw-design

[honest badger]

after jumping it out the current draw is close to 1 amp but never exceeded it

is it possible the fuse was faulty and I should buy another 1 amp fuse? should I just replace it with a 2 amp fuse and hope my house doesn't burn down?

[bright thistle]

Yes
No

[woven bluff]

is floppy disk driver 3.3V compatible ?

[supple pollen]

That would be 100 to 240 watts, depending on the mains voltage in your country: that's pretty high for a soldering iron. Most irons range between 20 and 45 watts.

If it's a self-regulating (PTC) type circuit, it could draw higher current to begin with, in which case a slow-blow fuse might make sense.

[outer adder]

A begginer question:Is this the proper way to connect two mcp9808?I have diffferent breadboards because there is a distance between sensors(around 30cm).Thanks!

[supple pollen]

That looks right, but I2C was not designed to work over long distances, so you may have to slow the clock to get it to work reliably.

[woven bluff]

anyone used acid core solder before ?

I accidentally used once, and the solder point keep spawning sponge-like conductive material that causes shorting everywhere

[inland jungle]

Has anyone had any experience with a MCF8316A? Completely integrated sensorless BLDC driver. Looks like a pretty awesome chip with an I2C interface

[supple pollen]

May want to treat it with industrial strength flux remover and baking soda.

[limpid nest]

How much fiddly analog audio stuff would be involved in making my own board with this hardware on it? https://www.adafruit.com/product/1381

I can always just plop the breakout down but I dun wanna

[ashen fox]

im making an rp2040 pi HAT, wondering which configuration for the RP2040 GPIO looks better/more useful?

[rigid kraken]

Is it possible to run an Led straight from a 3.3 GPIO pin or should i use something in between? im using the Pico

[ember laurel]

Figure out what driving current your LED needs, and what current the pico gpio can deliver - per gpio and total.

[outer adder]

Thanks,in my case how much distance can I2C afford(I know that isnt designed for long distances)

[supple pollen]

It's not that tough, I normally put the audio stuff at one end, so the signal paths to the output pins are short.

[supple pollen]

I don't really know enough to say: it's worth trying it out and seeing if it works. If it does, all is well. If it doesn't, then you can try things like slower clock speeds, fancier cables, and (if it is a severe problem) I2C extenders.

[outer adder]

Thanks,I will do so...

[forest relic]

I couldn't get much to work at the length you said, 30cm. But certainly, your mileage may vary. I found these to be extremely helpful, https://www.adafruit.com/product/4756

[supple pollen]

That's a good idea. If you want to go whole hog with I2C bus extenders, SparkFun offers some: https://www.sparkfun.com/products/16988

[forest relic]

thanks for alerting me to this, I was wanting a couple of these a few months ago, but they were out of stock.

[low anchor]

If I'm using KiCAD and I have a board part for the ESP-32 with a footprint, is there a way I can instruct KiCAD to 'populate' the footprint in the BOM with two sets of pin headers? I figure it'd be easier from a schematic standpoint than using two headers as the components, and I need it to have the correct BOM for an assembly service. Or I maybe I could edit the resulting BOM manually?

[low anchor]

Also this might be a dumb question, but I have an ESP-32 and MAX11612EUA+ chip. What do I need to do to connect them over I2C? Do I need to use pull-up resistors to 3.3v? I didn't realize I needed resistors, but after the chip arrived I realized too late that it seems to be designed for 5V Vcc on I2C - but given that I2C is active low, I'm not sure this makes a difference. The documentation I've seen has been conflicting so far and I haven't been able to suss everything out

(I don't need high frequency I2C either, it's just a ADC for a temperature sensor, which I'll read once every 10 seconds or so)

[supple pollen]

The data sheet specified Vih of 0.7 * Vdd, which works out to 3.5V for a 5V supply. Unfortunately a 3.3V pull-up won't quite manage that (although it may work anyway), and a higher voltage pull-up could damage the ESP chip. There are I2C level converters available (they're very simple, just a transistor and the pull-up resistors for each side).

[woven bluff]

the drv8871 I hot-aired off the adafruit breakout is not working...

and I already soldered it on my board...

I have another breakout, how should I remove the IC ?

first time I blow the opposite side to melt the thermal pad

[edgy jungle]

How does this voltage regulator system work? What is the purpose of the ferrite?

[woven bluff]

it's a LDO, it drops your VIN to 3V3

[supple pollen]

The ferrite bead just helps filter out high frequency noise.

[woven bluff]

it reduce voltage by shorting VIN to ground rapidly, hence the noise

[supple pollen]

The MIC5225 is a linear regulator, not a buck regulator, so it doesn't create switching noise like that.

[edgy jungle]

Ah I see. What does the en line do?

[supple pollen]

That is an input you can use to switch off the regulator

[edgy jungle]

Is that so it can switch between 5v and 3.3v?

[woven bluff]

looks like a switching device

with some sort of active switching noise cancellation

[distant raven]

There might be some ripple during power on but LDO step down voltage through a resistor divider. Which is generally very low noise.

If it’s stepping down voltage from a DCDC or ACDC regulator, you will probably have noise pass through

[knotty tiger]

the shutdown transistor might be used in a switching mode, but LDOs are generally linear devices (the pass transistor is probably operated in its linear region)

switching regulators generally need nontrivial external energy storage elements like capacitors and/or inductors

[woven bluff]

SMPS generally have >90% efficiency

[knotty tiger]

@woven bluff where did you get that screenshot? it doesn't match the MIC5225 datasheet i found

[woven bluff]

it's 5255..

https://ww1.microchip.com/downloads/aemDocuments/documents/OTH/ProductDocuments/DataSheets/20005661A.pdf

[knotty tiger]

so you can shut down the regulator output with a logic signal

that's not the part number that was on the schematic https://ww1.microchip.com/downloads/aemDocuments/documents/APID/ProductDocuments/DataSheets/MIC5225-Ultra-Low-Quiescent-Current-150mA-MicroCap-Low-Dropout-Regulator-DS20006683.pdf

@woven bluff the 5255 is also a linear regulator from the looks of it

[woven bluff]

any advice on removing DRV8871 from breakout board ?

[knotty tiger]

linear regulators can sometimes require inductors (or ferrite beads, which are kinda effectively inductors) for filtering to prevent destructive oscillations in their feedback loops

[low anchor]

is there a design I can use for my own bidirectional i2c level shifter? seems like the chips are either out of stock or very expensive at Digikey right now

[distant raven]

BSS138 is probably okay for 400kHz or slower I2C speeds

Sparkfun has some premade ones and I think even Adafruit does too

[real river]

Putting a bead between the output pin and the bulk capacitor is asking for trouble! At higher frequencies it effectively reduces the amount of bulk capacitance, which opens up the possibility of instability in the regulator. It will certainly degrade the regulator’s transient response.

Better would be to put the 1uF recommended bulk capacitance at the output pin, then a series bead followed by a shunt cap sized for the desired HF filtering. The corner frequency of the HF filter simply needs to be somewhere below the frequency at which the regulator loop runs out of gain (and/or the self-resonance frequency of the 1uF output capacitor). That still makes for a rather psychotic output Z over freq. however.

[supple pollen]

Yes, the product page for this AdaFruit offering (which is the simple transistor based one) includes links at the bottom to the original Philips/NXP application notes explaining the circuit: https://www.adafruit.com/product/757

[outer adder]

I am trying to make 8 sensors(MCP9808 with adafruit board) in a single bus.Each sensor is about 20 cm in distance to another,so bus will be like 1,6 meters.I dont have any requirement about the time,as the datas are temperature and do not change much over time.Will this work without extenders?

[forest relic]

Caveat up front, I am not very knowledgeable, so someone else can probably help more. I had a lot of issues trying to run a couple bme680s across about 12" (~30cm) of shielded cable, although as I do not have much formal training, I am unsure if the cable I used was the problem or not. I was able to use the cheap little dupont rainbow cables in a 8.5" length without much trouble. I was not daisy chaining them, I ran the wire as home runs to each sensor back to the mcu.

If you are daisy chaining, I think the extenders and their qwiic connectors might be ideal for you.

[outer adder]

My idea is smth like this.I dont know much since I am a beginner.

[supple pollen]

That's pretty long for I2C without extenders. Shielded cable may make it worse, because shielded cable adds a lot of capacitance which is a problem for I2C. There are some other signalling methods that work better for distances like that.

[outer adder]

then I will use extenders,thanks

[forest relic]

well, that might explain my issues. I thought shielded was supposedly better?

[supple pollen]

Shielded is better in many circumstances, because it resists interference. The drawback is that it adds capacitance, which is a problem for high speed, high impedance signals (like I2C) in particular.

[knotty tiger]

especially because I2C (at least the lower-speed electrical protocols) relies on pullup resistors to charge the line capacitance instead of active current sources

[real river]

You may need to look into putting additional pullup resistors at the end of the bus chain - the smaller the R value, the faster the rising edge will be. The master will have to work harder to drive lines low (sink more current) but it may result in a cleaner bus without having to resort to active bus extenders.

[broken zenith]

Does anyone know where to find a STEP file for https://www.adafruit.com/product/1201 or similar? I'm trying to get a working 3D render of my board in KiCad, but can't find a model. Thanks

Feel free to ping me

[woven bluff]

I see a lot of space on itsybitsy M4 is unused

you could put an IMU in there

also replace those SOT packages with FN

[glacial gale]

whats the overhead of context switching if most of the registers your using are cached in cpu cache. So that when you have to save the userspace state, its like 32 regs you have to save, which would usually be 32 reads + writes to memory. But are actually 32 reads from cache then writes to cache (write-back scheme). Then if you are using ASID on your TLB entries, you dont have to flush it. Hence the number of cycles for context switching in this case is just 32?

[supple pollen]

That depends a lot on your CPU and bus architecture. If you happen to have a Sparc CPU, the overhead is 1 clock, because the CPU itself maintains multiple copies of its register bank and can switch between them.

[glacial gale]

hmm Im not too familiar with sparc, but that seems like maybe they have a set of "shadow" registers that get used. Do you happen to know the overhead on arm cpus? I see some sources saying 20 banked registers, is that a similar design to sparc?

[supple pollen]

Sparc has both windowed registers (some of the output registers for one instruction become the input registers for the next) and context register banks. It's a little extreme, but makes Sparc very performant when it's switching between a lot of tasks. I don't know of any ARM variants that offer register windowing or per-context banking (although they may exist), but the 20 registers you refer to perform a somewhat-similar function, where some registers are automatically switched to different ones during interrupts, switches to supervisor mode, etc., which will yield a useful speedup (due to not having to save them off) in those situations, but isn't particularly useful for ordinary process/context switching.

For contrast, consider the original Pentium, which, while not even having very many registers, took over 1300 clock cycles to switch contexts (one of many reasons a 20MHz Sparc could handily outrun a 166MHz Pentium)

[glacial gale]

ooh I see

[limpid nest]

is a zener diode with a zener voltage of 6.2 sufficient to block 5V from a adafruit dev board when powering thru the 5V/USB pin?

[supple pollen]

It's a peculiar choice, but might work? Were you going to use it as an ordinary forward-biased diode?

[limpid nest]

I have it on hand

From something else

I need to stop by the shop and get another linear regulator for something else anyways so I can get another diode. What would you recommend?

[woven bluff]

can SSOP3 and SOT-23-3 use the same footprint ?

[limpid nest]

for boosting simple PWM, can I get by with a logic buffer or do I need a signal buffer?

[supple pollen]

I normally use a Schottky diode for the low voltage drop

[supple pollen]

It kind of depends on what you're connecting it to, but a logic buffer is often sufficient.

[limpid nest]

ok thank you

[woven bluff]

there a driver version of logic buffer capable of multiple amps

check the datasheet against your requirement

typical logic buffer have the same drive capacity as your MCU

[real river]

“Boosting” could mean several things. What sort of ‘boost’ do you need? Current? Voltage? Both? What is PWM voltage, frequency, min/max duty cycle?

[woven bluff]

FYI, Teensy4.1 is back in stock

[rocky fiber]

Sweet!!!

[limpid nest]

what happens when working with an "isolated" buck converter, if you short the GND (input) pin to the V_OUT- pin?

[unreal flax]

My understanding is that it should be okay, just changing it to effectively now be non-isolated. The converter itself shouldn't pass any current through that path internally, but if you short it externally, I don't think it'll break anything.

[limpid nest]

ok thanks. I can also just use a non-isolated model

[limpid nest]

Is this how you do a footprint for a board that has castellated pads, if you want the option to solder flat or use pin headers?

[bright thistle]

This board has what you're looking for ;)

https://github.com/sparkfun/AT42QT1010_Capacitive_Touch_Breakout/tree/master/Hardware

[hollow steeple]

I'm working on a new board that has a several push buttons (~30) but I haven't done a ton with devices that require actual input yet and so I'm looking into how necessary debouncing circuitry is. Does it make sense to run them all through buffer ICs or is that overkill compared to basic resistor/capacitor?

[hollow steeple]

Although now I'm seeing that the RP2040 has schmitt trigger input built in

[limpid nest]

I'm working on a board to communicate with an unkown-hardware but claims to be DMX machine. The manufacturer claims their machines conform to the EIA485 standard. Does this mean that there is likely a 240ohm resistor across the terminals of the the A/B inputs on the machine and I should match it with a 240 ohm resistor to make the impedance 120 ohms?

or is that 120 on each? I assumed they were in parallel and so you'd need to have double on each end

[distant raven]

From a design perspective, it’s simpler to use the Schmitt triggers on the rp2040. Probably doesn’t incur much effort on the code front to use either

Just make sure you pull up/down your buttons to make sure your input is better defined

[hollow steeple]

Doesn't the RP2040 have built in pulls?

[distant raven]

Pull ups yes

But they are weak pull ups

100kOhm if I recall correctly

[hollow steeple]

Huh I thought it had both

[distant raven]

I think it’s pretty rare that microcontrollers have both. I could be wrong though

[hollow steeple]

Datasheet 2.19.4 seems to show both

[distant raven]

Looks like you’re right

https://forums.raspberrypi.com/viewtopic.php?f=144&t=315020

But internal pull ups are not good for button inputs

Great to keep pins from floating

Not so great for inputs

[hollow steeple]

Is that due to the values?

[distant raven]

Yes

They are intentionally weak

If you need to pull a line high/low, it’s better for you to use external lower value resistors

[hollow steeple]

I suppose it might be worth looking into resistor arrays instead of having to hand place 30 pull downs

[hushed blade]

this mostly matters for external circuitry, but another "issue" with using internal pullup/downs is having to deal with default state during bootup

if you use external ones it's a lot clearer what state things are in without having to look up the datasheet for the mcu

[supple pollen]

You have a few choices. You could use a dedicated switch scanning chip to take care of all of them, or you could add an RC delay and use the Schmitt triggers, or you could debounce them in software.

[pale umbra]

is D. all of these a choice here ?

[supple pollen]

While it's technically possible, it's kind of silly

[hollow steeple]

Oh so I still need the RC bit even with the built in trigger? I didn't even think to look for chips that would handle it.

[supple pollen]

Yes, the Schmitt trigger only changes the switching point of the input to add some hysteresis. However to combat switch bounce, you need a time-based solution. One popular approach is to use an RC circuit to slow everything down to average out the bouncing into a smooth(ish) curve, then a Schmitt trigger to regenerate a clean logic transition from that.

[hollow steeple]

ECE1099 is barely more expensive than the I/O expander I was already using and has a keyboard scan matrix built in.. maybe I should just switch to that

[supple pollen]

I ended up taking a similar approach, although with a different chip.

[hollow steeple]

Similar approach being using a chip that will do the matrix for you? Which chip?

[supple pollen]

I was already driving a bunch of LEDs, so had an HT16K33 chip already hooked up and working, and I saw in its data sheet that it could read a keypad as well, so I added the appropriate components (some diodes) and modified the driver, and it worked like a champ.

[hollow steeple]

Can I ask where you got the HT from? Separate project could use them (can't seem to get the chip I picked to work at a reasonable brightness) but the usual places don't have them.

[supple pollen]

I bought them from the manufacturer's US site: https://holtekusa.com/

[hollow steeple]

Perfect, I had found that site but wasn't sure about it, thanks!

I'd been trying to use the BU97950AFUV but without much luck, figure better to give up and use what adafruit is using for the segmenteds I got from them

[supple pollen]

I bought a couple of tubes of them for my Back to the Future time circuits build: they run all the 16 and 7 segment displays, as well as some auxiliary LEDs as well as the 12-key keypad.

[hollow steeple]

I need them for my own version of the same project 😛

[supple pollen]

Heh, I was also originally using a different chip (MAX6954), also without much luck

[hollow steeple]

3x displays on 3 boards each that hook up to a main ESP32 motherboard.. thought that one isn't really working either. Haven't figured out why it just non-stop restarts so I can't upload to it yet.

[supple pollen]

Amusingly, Desk of Ladyada is discussing the TCA8418, which is yet another keypad scanner option 🙂

[hollow steeple]

I'll probably look at using the holtek since I need some for the other project anyway

[supple pollen]

Aside from the fact that it's not available from the usual sources, I really like that Holtek chip. Running 3 alphanumeric digits along with 10 numeric digits, some bonus LEDs, and the keypad from a single chip worked really well and wasn't hard to get going.

[hollow steeple]

Seems like the bigger ones don't have the keyscan, going to have see if I can cram all 3x 4-digit displays onto the one chip.

[supple pollen]

The bigger ones, like the 28-pin version, or the ones for other displays?

[hollow steeple]

The ones for larger displays like the HT1635D

Although since there's no quantity discount I suppose I don't really need to consolidate to just the one chip type

[woven bluff]

why am I only getting 150mV level on RP2040 ItsyBitsy Pin7 Pin10 ?

I have never seen something like this before

OK, it's just leak from other pins. The arduino-pico core got pin names messed up

[woven bluff]

Great...now I2C is not working

[steep bluff]

What would you guys recommend that's the size of an adafruit trinket m0 , performance of an atmega328, with lowest current draw??
I want something relatively tiny, I want at least an atmega328 as a benchmark and since this is a project for a vehicle I want it to have low current. This vehicle already has speakers / subwoofers so I don't want to suck up more juice than it already is... I've prototyped my entire project on a pico. Thanks a lot! 🙂

[distant raven]

https://www.adafruit.com/product/1501

Might be a bit small storage wise

[pale umbra]

also might not work anymore if they have a modern computer

what about the qt pies though ?

[steep bluff]

I saw that mini trinket before I stumbled upon the m0. The mini trinket page says this "Deprecation Warning: The Trinket bit-bang USB technique it uses doesn't work as well as it did in 2014, many modern computers won't work well. So while we still carry the Trinket so that people can maintain some older projects, we no longer recommend it".
Yeah that would be perfect for my project as size isn't a concern but that disclaimer is my only worry .

[steep bluff]

Those are out of stock 😦

Also I don't need circuitpython.. I'm a real man /s

But in all seriousness my project is in C. And the code itself is small.

[knotty tiger]

if you want something fairly close to an ATmega328P, you could try the Itsy Bitsy 32u4

[steep bluff]

Thank you that looks promising I'm looking at it now. I love the micro usb jack as well.

[pale umbra]

241 in stock on digikey, please don't state they are out of stock when you just don't want to buy from a adafruit reseller or you haven't looked.

[distant raven]

Try not to be rude

[steep bluff]

i haven't looked sorry. That gives me plenty of options to start with then thank you so much all 🙂

nah that's not rude that's fair

[distant raven]

some people could be interpreted that way, I could have been worded better

[pale umbra]

saying it's out of the stock can cause peoples to take the statement as truth and not look and give up electronics because everything they want to do is always out of stock because of the chip shortage.

[steep bluff]

That was my thought process ^

[pale umbra]

I think it's important to be careful with such statements with the chip shortage

[supple pollen]

It would be cool if there was an MSP430FR Trinket, but I think I'd have to go design that myself.

[pale umbra]

that's all. Is there a better way I could have said it while still conveying the same statement ? I'm not perfect and willing to improve ?

[distant raven]

I would have worded it as “hey, Adafruit might not have it but I did see Digi-Key has it in stock”

[knotty tiger]

you also don't have to program a QT Py in Python; many of them have Arduino support

[distant raven]

It’s more cordial in my opinion, plus leaves out any personal call outs. Some people might just not know they are sold elsewhere

[pale umbra]

Thank you I'll try to say something along those lines in the future

[steep bluff]

My understanding is that circuitpython itself would be larger than my C program therefore memory big enough to hold circuit python is not a constraint for my project

[distant raven]

There is Arduino plus microchip studio that can be used

[knotty tiger]

my understanding is that Arduino replaces the CircuitPython image when you upload an Arduino sketch

[steep bluff]

Ah so it's still relatively big

[distant raven]

Thankfully Adafruit keeps the SWD pins useful if you wanted to load a completely custom bootloader, etc… you can

[pale umbra]

qt py have a lot of memory though that's more of an issue on older MCU with circuitpython

[distant raven]

Plus there are a few types QtPy available so you can pick one that suits your needs

I personally like the RP2040 one because of availability

[knotty tiger]

my understanding is that ESP32-series can have quite good low-power sleep modes compared to other 32-bit MCUs, though i haven't tried them yet

[distant raven]

The ESP32 variety are great too

[distant raven]

Yeah, deep sleep is pretty great in that regard both circuitpython and esp-idf

[knotty tiger]

@steep bluff so if you're concerned about power consumption on the M0, ESP32-series chips might work better for you

[pale umbra]

If I may, as I have one, I don't think an esp8266 would be a solution because it does many tradeoffs

I asked a similar question several months ago in here which lead to 4xqt py esp32-s2 and peoples discouraged me from buying several esp8266 because of all the tradeoffs it does

[knotty tiger]

ESP32-S2 is used in the MagTag, and people have reported impressive power-saving performance there

[bitter flare]

I wanted to connect M5stack shield to a feather, got a bit distracted along the way and made this Feather:bit https://github.com/rosmo-robot/Feather-Bit/tree/main/v1 Still quite new to hardware so if anyone can help review before I pay actual money to make some that would be great. The plan is to make some additional daughterboards for them, so I think the design might make more sense then. Anyone got any feedback?

[woven bluff]

Took me a while to replace the charging limit resistor of all my Feather boards... dam*n the 0603..

They should ship out with 1K resistor from the beginning, most people attach way more than 1000mAh battery.

[supple pollen]

It is a pain, but I can understand the design decision when "too much" could yield a lithium cell on fire, and "too little" just means it takes larger cells longer to charge.

[pale umbra]

When I see 3-4 lithium batteries on a project I'm like, shouldn't I use lead acid batteries instead ? Seems less dangerous

[woven bluff]

The best way is to use programmable charging chip

like those on MKR

serial lithium battery needs special protection boards

[supple pollen]

I often opt for lead-acid, NiCd, or NiMh instead of lithium. The first two are good for high current draws.

[woven bluff]

lead-acid can take a lot of abuse, but it's heavy.

[supple pollen]

The main drawbacks of lead acid are the weight, acid sloshing out (except for SLA/gel types), and they'll die if left discharged for long.

[pale umbra]

but cheaper, I mean 3 lithium batteries, an ESC etc, solid enough solid body which is rainproof etc is going to run me 1000$

while a gas scooter that give me even more possibilities, 48V gasoline motor, lead acid battery is like 200$ used

[woven bluff]

if you have space for lead-acid, you'd better off with a bench power supply.

[supple pollen]

I'm going lead-acid for my hovercraft, even though the weight is going to make it harder to hover.

My highest current bench supply can only deliver 45A, if I want more current than that, it's pretty much going to be lead-acid.

[woven bluff]

hovercraft? like ACV?

[pale umbra]

that was my 2nd scneario before using RC

[supple pollen]

I don't know what ACV is, but I'm thinking a plywood disc with a bunch of blowers and a flexible skirt to keep the air on.

[pale umbra]

using an old hovercraft or segway style thing

But I'm too heavy for them

and probably would end up dead/in coma

or I'd have a third leg from falling over it

[woven bluff]

ACV = Air Cushion Vehicle

[pale umbra]

But I'm too old to ride a scooter

so I just cancelled that project

[woven bluff]

you use them for amphibious assault on beach head

[pale umbra]

well you said hovercraft/ACV so I though you meant hoverboard/segway-type vehicles

[supple pollen]

Yeah, same sort of thing, but I'm thinking a single person model. I saw one they threw together on Mythbusters Junior and figured I'd try building one. Some would argue that I'm too old or too heavy but I'm going to try it anyway. I figure if I'm already supporting 100kg of me, a few dozen kg of batteries isn't going to make it that much heavier.

[pale umbra]

I'm 145kg, all the hoverboard/segway i've seen says 110kg / 100kg sometimes 90kg

[supple pollen]

Here's Adam Savage riding it

[pale umbra]

But some scooters go to 150/160kg

There's also the street legal factor(custom vehicles can't drive on roads) and having enough storage to store the electronics safely (ie: bicycle are bad at this)

so when I evaluated all this I thought the best would be a scooter

the storage under the seat can usually be padlocked which solve the last issue

[supple pollen]

One nice thing about these air cushion varieties is I can ride them on the lawn, no need to head out into the street.

[pale umbra]

I don't have one or a backyard 😦

Like the city bus litterally pass 1 feet from my back door which make it feels like a subway sometimes 😄

[woven bluff]

I have seen lawn with pineapples, good luck...

[supple pollen]

Since the skirt is flexible, I can travel over small dents/obstacles/etc. without damage

[pale umbra]

before I buy all of this I'm experimenting with time of flight sensors, ultrasonic etc because the goal is to be warned of imminent collissions or drivers driving too close

[woven bluff]

you'll need radar for that

[pale umbra]

why? some tof sensors have a 25 feet range

[woven bluff]

the same reason cars have radar instead of ultrasound.. more reliable

[pale umbra]

Sucks that I can't have pets though, all of this would be solved by having a cat or a dog(better)

they even provide their own power

[woven bluff]

you can ask psychiatrist to prescribe you a pet

[pale umbra]

I'm allergic that's why. I can pet one once in a while but not live with one anymore.

[bitter flare]

LifeP04 may be a good compromise. Heavier than Lipo, but safe and lighter than lead?

[pale umbra]

or that normally that expensive or it's just the supplier I'm looking at ?

[steep bluff]

I'm trying to decide on which LED strip I need for my project. It needs to be 12V. I have experiecne with WS2812B 5V individually addressable LEDS. Now my only concern is that all the WS2812B 12V 5050s are in 'groups' of 3 so are they not individually addressable? I'm assuming the groups of 3 themselves are individually addressable???

[supple pollen]

Which strip are you looking at with the groups of 3?

[steep bluff]

any 12v led strip comes in groups of 3 they don't have individual ones

https://forum.arduino.cc/t/do-12v-led-strips-with-individually-addressable-leds-exist/538409/2

i read that and it makes sense

[low anchor]

When using this DC/DC converter https://www.mornsun-power.com/html/pdf/K7805-2000R3.html with 12 volts, if I never plan on exceeding 12 volts, can I replace the 22uF/50V capacitor on page 3 with a 22uF/16V capacitor? It makes sense to me given I won't ever give it more than 12 volts, but I want to make sure it's not rated so high for voltage spikes or something like that

[unreal flax]

That should be okay, though I'd recommend something like a 25V capacitor instead of 16V if you plan to run it at 12V, since capacitors get derated close to their upper limit, i.e. a 22uF will act like a 10uF at the higher voltages.

[distant raven]

I can’t think of the term but it’s related to how capacitance changes when different DC voltages are applied and is characterized by the size of the capacitor, given capacitance, and the DC voltage.

[supple pollen]

I think that's more high-K ceramic capacitors than aluminum electrolytics.

[distant raven]

Yes

Electrolytic don’t suffer from DC voltage capacitance variance because they are directional

Among other things

[rustic fulcrum]

ESP32-S2 QY Py, much faster and pretty power efficient, especially if you deep sleep it.

[pale umbra]

the deep sleep is really limited though

It's not like you can operate 20 sensors with it, upload by wifi and display while it's on

[rustic fulcrum]

Yup, though annoyingly deep sleep is currently broken in CircuitPython on the S3

[pale umbra]

it comes with a limited set of instructions (ULP-FSM) which require a custom programming language

[rustic fulcrum]

Well yeah, it's deep sleep, but you can have it wake up on a schedule or have a pin (including via a sensor) wake it up

Certainly would be great if CP had supported for the ULP chip as well, though I'm not sure if even Arduino has support for it

[pale umbra]

well if it doesn't I'll code it in C and make a CP lib

that ain't gonna stop me

[rustic fulcrum]

I am pretty confident #circuitpython-dev would be very happy for you to do that, and I certainly would be too.

[pale umbra]

I'd need help to make it complete and perfect though. No promise that if I do it it won't be odd to use/perfect/complete

and except for the C-family I'm literally starting from 0

[rustic fulcrum]

Ooh, looks like there is a way to do it, though the last commit was more than a year ago: https://github.com/duff2013/ulptool

Oh and I got a response back from Pololu. They said that the boost boards I was looking at have quiescent current of around 200uA and they expect the total draw from a 2.4V input to be around 330mA with a 3.3V, 200mA output, so at least for my projects with lots of deep sleeping, it's probably (?) more efficient to just feed a Feather/QT Py the unregulated output from 3 NiMHs and add a voltage monitoring alert to Adafruit IO. I figure that while the regulator isn't that efficient with higher voltages, as the vast majority of my runtime will be deep sleeping, even if I assume it's only 50% efficient, it'll be wasting far less energy than the boost board would be.

..though I guess if Adafruit ship that ABLIC timer board, maybe I could have that run the interval timing and control the enable pin of the regulator board?

Of course for you @pale umbra, you need the 5V for your display, so it's a bit of a moot point, but I did find this the other night, and while it doesn't have low voltage cutoff so you'd need to figure out a way to do that yourself/change batteries a bit preemptively, they don't have power numbers for the whole thing, and apparently they stuck an LED on it you'd probably want to solder off, the datasheet for the TI chip they're using shows it as having absurdly low quiescent (20μA): https://www.dfrobot.com/product-991.html

[pale umbra]

well 😦

I was going to test with 5V but I forgot I switched to circuitpython 😦

[rustic fulcrum]

How would that change things?

[pale umbra]

well all I have is an arduino sketch and like I said I start from zero in circuitpython

and I use libs that probably don't even exist in python so i'd have to rewrite it

[rustic fulcrum]

Ah

[pale umbra]

to be honest I'm just overloaded with electronics and RL stuff to do and don't know what to start on

since wemos stuff is not from adafruit and circuitpython big backer is adafruit I don't expect much already made for it

[rustic fulcrum]

Start with something that'll be useful for you and not too difficult.

[supple pollen]

I doubt it's the directionality as much as the insulating medium (aluminum oxide instead of barium titanate)

[distant raven]

True, voltage and design are also factors listed in this cool article from Digi-Key I found 🙂 https://www.digikey.com/Site/Global/Layouts/DownloadPdf.ashx?pdfUrl=863968494F2E4E13BBEA65B55A358443

[rustic fulcrum]

Huh, so some more digging and it looks like the S2 and S3 have a core that doesn't require a custom programming language: https://docs.espressif.com/projects/esp-idf/en/latest/esp32s2/api-reference/system/ulp-risc-v.html

(asking about this in #circuitpython-dev where this arguably belongs)

[north quest]

Hi. I'm making my first DotStar LED strip art project, and I've got a question about wires. I've got 180 LEDs, 2 strips, 3m each, 30 LED/m. So, I don't expect more than 5A.

The distance from the power source to the LED strips closest edge is 4 feet. What should the gauge be of the GND and V wires be? I plugged things into some calculator and got 16 AWG. Does that sound right? Does it matter if it is stranded or solid core? And, do the Data and Clock wires also need to be the same kind of wire?

(The strips are in a heart-shape, and I'll feed power to both ends of the strip as well as the middle.)

[supple pollen]

Solid or stranded wire is fine. 16ga should be sufficient if you're running all 8 wires to the power supply. If you're combining power and ground and then splitting it up nearer the strips, the combined wire could be carrying over 10A and you'll want something heavier. Traditionally, data and clock are the same kind of wire, but it's not a requirement, however, you'll want to have a separate ground reference wire from the LED strips to the controller instead of depending on the same ground as the power is flowing through. @north quest

[lost patio]

What type of connector do I require if any to connect lipo battery to a feather m4?

Using breadboard

And is there any wiring diagram or reference for connecting two cc1101's to a feather m4?

[north quest]

Thanks for the help! I wonder if maybe I didn't make myself clear. There are 180 LEDs in total, and I don't think they would draw more than 5A in total. (more like 180 * 20mA = 3.6A)

[north quest]

So to get a non-current ground reference wire for the microcontroller, would I do something like this? Run the 16 AWG wires from the power adapter straight to the LEDs, and then splice off some smaller 22 AWG that goes to my perma-proto board housing the microcontroller?

[woven bluff]

SEN55 data sheet said it can run I2C at 100KHz, but it only works when I lowered I2C clock to 50KHz,

any ideas?

not enough pull-up? too much pull-up?

[distant raven]

What value are you using for pull ups?

[pale umbra]

could it be cause by the MCU clock not being able to run at 100khz or ic2 bitbanging on the MCU ?

[distant raven]

Which MCU?

I believe bitbanging i2c limits you to 100kHz anyway. I wonder how consistent the clocking is when doing bitbang

[pale umbra]

I don't know which MCU AcThPaU is using

I just didn't want to raise potential causes that wouldn't help solve the problem

but personally these are the first two things I would check if I was in that situation

[knotty tiger]

if the stuff you're talking to can tolerate a little bit of clock stretching, the I2C spec isn't very picky about exact clock pulse consistency, other than maintaining reasonable setup and hold times

[inland jungle]

how long of a cable run for the I2C signals?

[minor axle]

About a meter at the default baudrate of 100K

[granite steppe]

Hello all,

I have a project where I need to control 4 servo motors. Power is not an issue, all are micro plastic gear ones. I have no pins left on my microcontroller, so my options are either SPI or I2C bus. So I looked in to how Adafruit makes their 16 channel 12 bit servo controller, and I want to add something similar to my PCB.

What characters should be I looking in to when selecting a LED driver to driver servo motors instead?

[inland jungle]

Sorry, I should have replied to AcThPaU, I was wondering if cable length was causing his clock issues

[short river]

I am searching for a cheap IC / circut for wireless power receiving, compatible with qi chargers but not certified as it's too expensive and has a whole cpu in it, any suggestions?

[unreal flax]

I think you're going to run into trouble with the requirement of "Qi-compatible but without all the stuff that Qi requires".

[distant raven]

There are a few IC options but it looks like they are $5 or more + PCB and applicable externals

Texas Instruments appears to be the major player in this market

https://www.digikey.com/en/products/detail/BQ51052BYFPT/5218173

[pale umbra]

I'm surprised they never made a Ti-83duino you can bring at an exam

[distant raven]

That charger IC is Qi compatible so that’s a plus

[regal lodge]

I'm still not clear on how caps work, does the cap on this PCBs VCC line only really affect the thing it's near or anything on that VCC line?

I think/assume it's anything on the line but wanted to make sure I understood it correctly

[distant raven]

Think of it like a line of people doing the wave

The wave can exist at any point on the line as it moves down.

If you put your capacitor too far from the power pin of the individual IC, it will have some effect on reducing the impact of the wave but it would be more effective to keep it from touching the IC power pin if say another wave was introduced after the capacitor that is further away

[regal lodge]

the VCC line there goes 2 directions on either side, either to the pad there or to the through hole, the through hole goes to another board, only 1 connection would ever be used at any given time

but since the cap is "after" the through hole I wasn't sure if that mattered in regards to something connected via that hole

[distant raven]

Well, traces will act like an antenna and could pick up on EMI introduced at other points in the design

Which is something we also want to account for

[regal lodge]

I'm honestly not 100% clear on the purpose of this cap in this design, rather that was actually something I'm trying to understand

[distant raven]

In this case it’s probably to buffer power of the display

The display probably draws quite a bit of current on startup

So the capacitor helps smooth those spikes

[regal lodge]

well that's not a display connector it's near but same principle I guess

[distant raven]

Or I’m assuming that’s a display on those thin pads

But either way

If there’s a lot of power drawn on power on, you want to have bulk capacitance to buffer spikes in current draw

[regal lodge]

yeah so I'm just trying to understand is the cap there to buffer for only the VCC after it or for the VCC right before it as well

[distant raven]

In this case I don’t think it necessarily matters

[regal lodge]

also would the cap interfere with a multimeter continuity test between the through hole and pad?

[knotty tiger]

probably not, if they're the same net

[regal lodge]

I know I've heard about caps messing with multimeter readings in some situations because they start trying to charge off of the multimeter

[knotty tiger]

if you're testing for open between VCC and GND, though, the capacitor would read as a short circuit briefly until it charges up

[regal lodge]

oh another question, if I need to communicate with multiple devices over the same set of output pins on an arduino and they require different voltages (3.3V & 5V), what's the most optimal way to handle that?

it's quite a lot of IO too (pretty much all of those through holes...)

[rustic fulcrum]

I assume probably a logic level shifter?

[regal lodge]

ideally it'd be something that can be a fixed voltage on one side (5V) and then variable on the other

[pale umbra]

QUESTION: I need a timer that work in the freezer to unlock it, freezer temperature is about -40oF is this impossible even if I don't use an MCU ?(the batteries would be freezing too, the wires, etc)

[regal lodge]

would a 555 work?

[pale umbra]

probably, I'd like to limit myself when I buy a dessert and I can't help it. So I think sealing it , say an ice cream, for 24h would help me keep it reasonable

I only saw 1 product that does this so I'd like to attempt this

[regal lodge]

are you looking for a timer IC or an RTC

[pale umbra]

sometime that going to tell a servo to turn a lock (tab type probably) every 24h or so

doesn't need to be super precise +/- 1 hour would be fine

[regal lodge]

maybe this, https://www.adafruit.com/product/3296

[pale umbra]

I wish they were humorous in their FAQ especially when they says no 😦 "In short: no. While cold environments help maintain battery life, refrigerators and freezers are not safe to put batteries in. The moist environment will cause condensation on the batteries. This in turn will lead to rust or other damage. Avoid putting batteries under extreme temperatures at all times"

[regal lodge]

I would not use a rechargeable in a freezer

lithium metal batteries should be alright though

[pale umbra]

what about coin?

[regal lodge]

most coin cells are lithium metal

[pale umbra]

I was going to say that surely you they can use batteries in very cold weather wheter for climbing or for arctic research but google says lithium

[regal lodge]

I think the issue you'll have is that -40 is literally at the extreme edge of most components range

[pale umbra]

yeah but I've been reading a lot of datasheets and it says the min is always in the -40 to -50 region

aand I'd be willing to maybe the freezer hotter by 5-6oF for this

[regal lodge]

the DS1307 I linked only goes to 0C, you can use it for prototyping but you'll need to swap the IC to the DS1307N(+) to get the -40 minimum

[pale umbra]

of course the alternative if nothing works would be to get a timer box with the key

[inland jungle]

are you sure your freezer is set to -40F? Is this a laboratory freezer or something?

[pale umbra]

and padlock the box with the icecream since padlock have less issues with cold

[regal lodge]

yeah tbh -40 is kind of nuts for a consumer fridge/freezer

[inland jungle]

0 F is typical

[pale umbra]

well it's -30 -35oC not sure what it is in oF

243 kelvins or so I think

[regal lodge]

-40 is the same in C and F

[inland jungle]

-18 C is typical for consumer freezer

[pale umbra]

-22oF to -32oF is -30oC to -35oC

[regal lodge]

lol

[pale umbra]

mine is set colder because I don't like meat continuing to degrade while it freeze

and I want room-temperature store-bought sugar-less soft drinks and juice to get cold quickly

[regal lodge]

lithium metal...

[pale umbra]

if it condense thought I'm in big troubles 🔥 🎆 🧑‍🚒 🚒 ☀️

[regal lodge]

have you actually measured the freezer?

[pale umbra]

yes but I dont have a thermometer in it anymore, it is at the coldest setting

[supple pollen]

They're RGB LEDs, so each DotStar has 3 LEDs in it, drawing a total of 60mA max per LED.

[pale umbra]

and from what I remember the acohol ter ... oh I have an ir thermoemeter gun give me a second

[supple pollen]

You don't need to run V+ between the LED strip and the controller, unless you're powering the controller from the same supply as the LED strip

[regal lodge]

this will always be an issue even if you powered it with external wires the stuff inside could still condensate and short, lithium metal is not as temperamental as lithium ion though because you aren't trying to charge it

[pale umbra]

well that's disappointing, -2oF

[regal lodge]

yeah I suspected as much

allowing you to set a very low temp and reaching that temp are not the same thing

[pale umbra]

why so hot 😦 alcohol melt at this temperature

[regal lodge]

I mean pure ethanol doesn't freeze until -173F lol

[pale umbra]

surely I'm not the only one who freeze alchol to concentrate it then removing the water ice right ?

[regal lodge]

as for freezing meat and stuff quickly, just use LN2 :3

I mean you can do that sure but it's easier to just buy some everclear or something

if it's home made alcohol that's a good way to go blind (depending on the kind it is and how you use it)

[supple pollen]

Nope, in the old days, people would ferment apple juice, freeze it, then pour off the now-more-concentrated lower melting point fraction: this was known as "applejack".

[north quest]

Ah right, but I won't ever run them at full brightness and pure white.

[north quest]

Don't I need to supply the logic level shifter with the 5V reference voltage?

[supple pollen]

Then, naturally, they'll draw less average current (although the peak current can be high, which is why a capacitor can be useful)

[supple pollen]

If you're using a level shifter, that would make sense, yes.

[north quest]

I'm also using the 5V to power the USB power source for the controller. That way I don't need to have a lipo battery involved.

[supple pollen]

You have a few options then. The ideal (electrically) one is to run separate power leads from the supply to the controller, to avoid "dirty" power that has interference on it from the LED strip. However, in the real world, sometimes a topographically ideal setup is preferable to an electrically ideal one. Since it sounds like you're using a 3V controller, it will have a local regulator, which will isolate its power from the "dirty" 5V from the LED strip.

[north quest]

Yep. I've got the capacitor. Wasn't sure if I should put it before or after I splice the power and ground to go to the controller (via the built in 5V to 3V regulator) and logic level shifter. Does it matter?

[regal lodge]

here's a thing I did with LED strips :3

that's an arduino though, powered the arduino with USB 5V and for the LED strip I used the power supply 5V directly :3

[north quest]

Does it matter where the capacitor goes? (Blue question mark)

[supple pollen]

That picture doesn't match your other one, nor your description. Basically, a capacitor near the LED strip will help smooth out power to the LED strip, and a capacitor near your controller will help smooth power to your controller.

[regal lodge]

if your LEDs are WS2812B-based, some flavors of WS2812B are fine with 3.3V data btw :3

[north quest]

Oh, so the capacitor should go close to the LED strips?

[north quest]

They are DotStars, so think that makes them SK9822? I think. said so on the datasheet anwyay

[regal lodge]

datasheet seems to indicate that it'd be fine with 3.3V as well

since you are undervolting I'd say it's unlikely to cause damage and might be worth seeing if yours are fine with it

[pale umbra]

When you get a box of 350 leds and all the power in your house isn't enough voltage to run them off mains

but they are just leds

probably the highest voltage thing in my house, would need about 1090 volts total

[north quest]

Sorry for asking beginner questions (but... you know, I'm a beginner 😀) but what do you mean by undervolting, and how can you tell that I am 'undervolting'? I thought I was supplying the LEDs with 5v?

[regal lodge]

I thought I saw you say that your controller is outputting a 3.3V signal

[north quest]

Ah, yes, but I'm running it through a logic level shifter to get it back up to 5V.

[regal lodge]

I was saying you might not need to do that

[north quest]

Got it. But there is no harm in doing it right? I've already got the circuit with the logic level already in there so ... no more work really.

[regal lodge]

you don't need to but it could simplify the circuit by omitting it

also those LEDs may also be able to be driven by 3.3V (though they'd be dimmer, not a bad thing if you weren't going to run them at full brightness anyway)

[pale umbra]

how are you powering the 5V side of the level shifter ?

[regal lodge]

undervolting is just running something under the recommended/"default" voltage

[north quest]

I splice a 22 AWG wire from the power line that goes to the LED strip.

[regal lodge]

it can save power and in the case of LEDs (and some ICs) it can increase their lifespan

[north quest]

And that power line is hooked up to a 5V 10A power adapter which is the source of all the power.

[regal lodge]

it should at least tolerate 3.3V, up to you of course

if it's also a learning experience then using the level shifter has the advantage there even if it's not absolutely needed ^^

[north quest]

Interesting... I wonder why the Adafruit DotStar guide (https://learn.adafruit.com/adafruit-dotstar-leds/power-and-connections) says different.

"DotStars are 5 Volt devices. They may respond to 3.3V signals, but this is not a guaranteed thing. If using a 3.3V controller (Feather, Raspberry Pi, etc.), add a logic level shifter to boost 3V logic to 5V…something like a 74AHCT125 on the data and clock pins."

[regal lodge]

because 5V is nominal

[north quest]

I've just been following their advice since I have little experience to draw from myself.

[pale umbra]

on the 8266 it explicitely states the pins can't take more 3.3V

if you connect them to a 5V logic device they are toast

[regal lodge]

3.3V won't always work, it'd be an issue if this was a commercial product but if it's just for you then it only really matters if it works this one time or not

[supple pollen]

It kind of depends on what problem you're trying to solve. You could, of course, put capacitors in several locations.

[regal lodge]

ok?

[regal lodge]

the main reason I suggested it was so you could avoid a custom PCB

[north quest]

This is for an art project out at Burning Man (Black Rock Desert in Nevada) and so once I'm out there, I won't be able to head back to civilization to get more stuff. (and I am making the electronics away from the LED strips, so can't test it until I'm out there. Got a 1 m strip to test on here locally)

[regal lodge]

if 3.3V worked directly then you could omit the level shifter and wire directly to the controller board, avoiding the need to get a PCB for the level shifter

[north quest]

Thank you, that is kind of you. I kinda enjoy making my first PCB soldering my first perf board. 🙂

[regal lodge]

ah I see

then yeah that wouldn't be optimal :3

[north quest]

If I can do that, then maybe I will put one close to the PCB and a second one close to the LEDs. Which problem am I trying to solve? I don't know! I'm just trying to follow the Adafruit guide (which btw agrees with you; they said to place the capacitor close to the LED strip. I misread it!)

I'm sure something will go horribly wrong but at least I can say I gave it my best try.

[regal lodge]

caps are fun

I didn't use one for my LED strips but I was powering them with a computer power supply, already got pretty big caps and the cable to it was pretty short so probably why I never had issues

[short river]

I meant as in when I slap it on a qi charger it will work, I think a normal receiver IC would work, but I can't find any cheap ones

I thought about using pogo pins instead but they are even more expensive

I'll probably have to stick with USB-C, which unfortunately will remove the waterproofnes from my project

There are some waterproof usb c inputs but they are too expensive as well

[supple pollen]

Does it have to be Qi or would a simpler wireless inductive solution work?

[distant raven]

@short river this is the only in stock one I’m finding

It’s Qi compatible

But the BGA chip is going to be interesting to work with

Actually doesn’t look too bad

[bright thistle]

You know how some hifi have a knob that is spring loaded, acting like a button when rotated to one of the ends?

Does that have a specific name?

[fervent lance]

It's just a switch.

[woven bluff]

cannot even find a charging chip... every thing left are FN or BGA...

The world has fallen

[distant raven]

https://www.digikey.com/en/products/detail/MCP73831T-5ACI%2FOT/7065591

There’s a few variants of that

[woven bluff]

I need at least 1A

[supple pollen]

A charge rate of at least 1A? Do you have a limited charge time available?

[distant raven]

Here https://www.digikey.com/en/products/detail/MAX1874ETE+/1512794

1A, single cell Li-ion charger

In stock but it’s 5x more expensive

Here’s another option https://www.digikey.com/en/products/detail/BQ25050DQCR/2704589

Cheaper too, but the package is slightly harder to solder

[woven bluff]

I don't think I can solder FN packages...

[distant raven]

Get a cheap hot plate, solder paste, order a stencil with your PCB

[supple pollen]

You make it sound easier than it is...

[woven bluff]

I have reflow at the university, but I'm not confident that I can work with FN.

[distant raven]

Sure it takes practice to get a good stencil technique.

But the only way to get decent at something is to try

Practice on bare boards without placing parts first

[supple pollen]

Practice, a jig, a good squeegee, the right particle size solder paste, and perhaps a heat spreader on that cheap hot plate

[distant raven]

I was kind of thinking that small one that Adafruit sells MP320T or something like that

Oh that went up in price

[vast flume]

Hey, would anyone be able to look over one of my pcb designs to see if there are any possible issues?

[low anchor]

What kind of issues?

[vast flume]

It is the first one I have designed so I just wanted to make sure I didn't make and dumb mistakes

[limpid nest]

post it! Folks are usually willing to take a look

[vast flume]

Sounds good! What format should I post in? @limpid nest

[limpid nest]

what format are the files in?

[vast flume]

Its a kicad project

[limpid nest]

I don't have kicad but lots of folks here do. If you're ok with it, just post the kicad files 🙂

zip would work well

[vast flume]

Will do 😎

Actually I just remembered I have some custom libraries

How should I handle that with a zip upload?

[limpid nest]

unsure

I don't use kicad

[vast flume]

Ah ok all good

[eternal fox]

Sup peeps!

So, im trying to figure out how to make a usb-hub and make it ready for sending to a manufacturer.
Now ive singled out the chip i want to use and im studying the documentation for it.
I understand the part by part nature of it, but i cant really see it as a functional design.
Is the idea that this refference design is sent with the gerber files to the manufacturer so that they understand what the parts do, or is it JUST for me to be able to connect the traces correctly between the different sub parts?`

https://assets.maxlinear.com/web/documents/xr22417_rev10.pdf
Sent the docs if that helps 🙂

This is my first time trying to do something like this and any input is more than welcome ❤️

If im reading the above page properly, the first page is the controller, then you route the ports of the second page to the controller as shown i page 1?

[unreal flax]

The reference design is just for your benefit in creating the board. The actual PCB fabricator will only be going from your gerbers and bill of materials, etc. since they have no idea what you might have wanted to change or add to the original design.

For connections between pages in the schematic, you go by the signal name. That is, if "EXT_RST#" appears on two different pages, it's really the same net and those wires should be connected together somehow on your board.

(The PCB CAD tool will help to enforce this, since it'll know whether you have signals in your schematic that are supposed to be connected on the board.)

[vast flume]

@limpid nest So I ended up finishing my pcb design, but now I am getting a ERC violation that I cant resolve. It is saying that the input power pin is not being driven by any output power pins, but it is connected to the power output of my voltage regulator so it should be working

Is my decoupling cap (C2) in the correct configuration?

[eternal fox]

Thank you Ed, thats wonderful ❤️

[unreal flax]

It's possible that the regulator library part just isn't fully annotated with the functions of its pins, so Kicad doesn't know that one of them is a power output.

[vast flume]

Ah ok cool

The decoupling cap is in the right config though?

[unreal flax]

Yep.

[vast flume]

Awesome. And yeah just checked and you're right about the library being incorrect

And then the connections on the right side of this accelerometer look ok? @unreal flax

[hushed blade]

I often find that I still have one of those "input not driven by output" because it doesn't know what connector I'm using to get (in my case DC) power from off-board. You have to annotate a line to tell it that it's a power source

but if the libraries are all good you should ONLY need to do it once for the external power source

[limpid nest]

how does it work to have any number of shift registers, say '595s, hooked up to a spi bus AND another chip that uses SPI? I thought that the RCLK line on the '595s acted like chip select?

So I just specify a chip select for my other chip(s) that use spi and it's fine?

[unreal flax]

Yep, looks good to me.

[eternal fox]

Am i silly or can the HS8836A USB IC really deliver usb 3.0 speeds to a single port? The ad information says it should, and theres a full complement of cables from the USB C upsrtream connector.

The fourth connector, which is labled as usb 3.0 uses the DM4 and DP4 pins on the HS8836A chip, but has 4 pins going straight to the usb C upstream connector.

[pale umbra]

what does the datasheet says?

because I can see two chips with that name and one is no-brand and 2.0 and the other one is a reputable brand and 3.0

[eternal fox]

2.0, 4 ports only.

oh

[pale umbra]

so no if it says 2.0 it's 2.0 not 3.0

[eternal fox]

hs8836A Pz042471 Thats what i can find on the IC.

[distant raven]

Fpga are hard, let that be said

I’m trying to us a SPI flash chip with an ICE5LP4K. The flash chip programs just fine, but after reset, the fpga doesn’t load the program off.. not sure if I messed up wiring but it’s interesting.

I can do normal cram programming but loading from SPI flash is illusive so far

[unreal flax]

Some sort of strap-pin setting to tell it to boot from flash? Do you see any activity on the SPI bus?

[distant raven]

I need to sniff it somehow

I was smart and didn’t include probe points 🙂

Due to space constraints I used a 8-uson chip too 😶

[unreal flax]

How are you programming the flash chip, then?

[distant raven]

Ft232h

[unreal flax]

Ah, gotcha.

[distant raven]

It says it successfully programs the flash

But a trigger of creset doesn’t seem to complete the chain

[unreal flax]

Any chance the FT is still holding MOSI?

[distant raven]

It’s possible, though I’m not sure what other boards do to overcome that

[vast flume]

Hey guys, what do I need to be able to solder this IC onto my pcb? I've never done smd soldering so I don't currently have any of the hardware needed for that, just want to make sure I get the right stuff https://www.digikey.com/en/products/detail/memsic-inc/MC3479/15292802

[unreal flax]

I'd presume that after a power cycle the FT wouldn't be asserting any lines until the driver tells it to.

[vast flume]

The package is a 12-VFLGA

And the PCB I'll be putting it on will already be partially assembled from JLC, they just didn't have this part in stock

But digikey does so I bought it from there

But since it'll be partially assembled I wont be able to use a smt stencil

[unreal flax]

In theory you can do it with an iron if you extend the pads out from under the chip when making your board. Other than that I'd probably dab down solder paste with a toothpick/pin and try your luck with hot air.

[distant raven]

I'm wondering if I just have the lines backwards

I probably need:
FL_MISO --> FP_MOSI
FL_MOSI --> FP_MISO

which I guess would explain why it works to program the FPGA in CRAM mode

realistically, you probably shouldn't be able to do that if you're programming SPI FLASH

[vast flume]

Sounds good. If I use a heat gun how do I make sure I'm not too hot?

[distant raven]

this will be an interesting bodge job hehe...

hopefully I can pull this off with 0.4mm pitch QFN

[unreal flax]

I'm not sure, I'm afraid. Most of my heat-gun use is for desoldering, so I know it's hot enough when the part pops off. 😁

[vast flume]

Ah ok 🤣

I'll do some research, thanks for the starting point

[distant raven]

@unreal flax my hunch was right

This was the most difficult bodge of my life

0.4mm QFN to a 8-uson flash chip

[unreal flax]

Ooof, impressive!

[distant raven]

The QFN was the easy part lol

[steep bluff]

are these bright enough to be used on a drone at night https://www.adafruit.com/product/754

[supple pollen]

It depends on what you're trying to do with them. They tend to concentrate their light in a narrow angle, so if one of them is aimed at you, it'll look pretty bright, but if not, it will be considerably less so. That said, LEDs are normally pretty easy to see at night.

[bright thistle]

big OLED sucks ~200mA, the tiny OLEDs on the five buttons pull ~100mA each, add the (small) load from the I2c mux... guess I should bump the AZ1117 (1A) to an AZ1084 (5A) just to be sure XD

[supple pollen]

Or go with a switching regulator

[bright thistle]

as in an external one? or are there on-chip solutions I could use on an HAT?

[bright thistle]

would this do? 🙂
https://www.mouser.it/datasheet/2/115/DIOD_S_A0007089856_1-2542907.pdf

(gotta love how Diodes Inc. seems to be always using the same two caps for Vin and Vout 😄 )

[supple pollen]

You'd have to build the rest of the regulator, but that would get you 2A instead of 1A. I'm lazy and tend to buy already built switchers from Pololu, etc.

[steep bluff]

hm I'm trying to build a programmable night strobe solution for my drone. People use these https://www.amazon.com/Lume-Cube-Strobe-Anti-Collision-Lighting/dp/B07TDNL1KV I've only ever used WS2812bs and WS2811s on projects. I was trying to find similar white LEDs to those in the images on digikey but no luck im not sure what voltage and amperage they are. the one's i've looked at seem to be too weak

[supple pollen]

That looks like a quartet of ordinary 1-watt white LEDs. These are probably similar performance: https://www.adafruit.com/product/518

Er, these look like a closer match to the Lume Cube ones (less than 1 watt): https://www.adafruit.com/product/682

[orchid phoenix]

Hi, I’m a noob when it comes to all this stuff so I’m trying to understand the transfer from breadboard to a protoboard. The picture on breadboard is based off my prototype in real life. I’m trying to understand how to ground everything properly as well as the voltage connections to the external power supply.

Are the ground and vin connections in second photo correct?

[steep bluff]

You are a genius

ty

would these be way brighter?? https://www.digikey.com/en/products/detail/samsung-semiconductor,-inc./SPMWHT541MP5WAVUS5/6052694

[supple pollen]

Looks like it. You can either solder a bunch of holes together like the bus bars in your breadboard, or you can daisy-chain them like in your picture.

[orchid phoenix]

Ohh okay so for the bus bars I just solder the ground wire to a specific hole and then solder a bunch of holes in a straight line connecting to that, and then I can just connect each component to a specific hole

[supple pollen]

Yes, that's a popular way to do it.

[bright thistle]

uh-oh... kicad doesn't have tsot-26 package 😐

ah. it's just an alias for 23-6 😐

[woven bluff]

is it OK to tie SA and SB to ground ?

[bright thistle]

Ok, found a suitable one.
How would you use it? Would you integrate it into the HAT design, plugging it in with a pin socket/header combo, or would you connect it with a loose wire to destination?

[supple pollen]

I think that would disable current sensing

[supple pollen]

Good question. It's basically a substitute for a linear regulator, so I'm guessing whatever you had in mind for that?

[bright thistle]

I did try the former option, and i managed to route the board, but not without difficulties 😅

Had to move the EEPROM section to the backside, among other things

[supple pollen]

That makes sense, there's limited room, and a switching regulator does tend to take up more room. Mounting it as a daughterboard might help, or allow an either/or approach. I took a similar approach with one of my boards (switching regulator is at the right end, mounted vertically)

[bright thistle]

Were I to try the latter, I'm afraid it'd result in a cobweb of cables...

Aaand I'm an idiot.

Never thought about mounting it vertically, I just laid it flat!😅

[supple pollen]

Come to think of it, there are two switching regulators on that board: the little black box next to the orange resonator is another switcher, that mounts in the same TO-220 hole pattern as a 7805 but thicker.

[bright thistle]

Drew a 0.7"x0.7" square, placed the socket inside at the right distance and avoided placing anything in that space

[supple pollen]

That sounds workable. Oh, as for "cobweb of wires", here's the prototype:

[bright thistle]

Nixies 😍

[supple pollen]

Yeah, the bigger switching regulator is the +170V supply 🙂

[bright thistle]

Gonna place it vertically next time I work on it.

This way I might get back some precious space, wasted enough already because I put two 1x12 headers 0.6" apart for the I2C mux

And at this point I might just place it outside and use a simple 1x4 for power and signaling

[supple pollen]

Oh right, that's a bunch of I2C pins to route. I've had to reluctantly grow some of my boards just to make room for connectors.

[bright thistle]

Nah, not many to route actually, the only ones going to Pi are power, and an SDA/SCL pair, all other pins go downstream, but take precious PCB space even not routed

https://www.adafruit.com/product/2717

Imagine this as a daughterboard

Now imagine being constrained to the HAT form factor, what a mess it'd be to place and route among a sea of unconnected pins 😅

(btw, Ada should consider updating its wiki so to include the option of the mux overlay, much easier to use that manually selecting the channel at every data transfer)

[supple pollen]

I'd consider some of those high density surface mount connectors or something funky like TagConnect.

[bright thistle]

Uhm? TagConnect? Was is das?
.
..
...
C|N>K

[supple pollen]

https://www.tag-connect.com/

[bright thistle]

Saw it, that's what the C|N>K was for 😉

[supple pollen]

I wasn't sure what that meant

[woven bluff]

Divergence meter! A man of culture here.

[supple pollen]

Well spotted!

[bright thistle]

http://www.catb.org/jargon/html/C/CNK.html. 😉

[woven bluff]

I made one myself, direct concept to PCB

I don't want to prototype with matrix board anymore

[supple pollen]

Why not?

[woven bluff]

takes too much time

while you can order PCB for a few $

also, you don't have to use soviet shift register. There are modern HV shift register for plasma display

I use 3x HV5530 to drive 8x IN14

[bright thistle]

scribbles: ask for help here when it'll be time to design the PCB for the tube theremin

[supple pollen]

Yes. Do that.

[supple pollen]

At the time, the modern ones wanted a 12V signal, and I didn't want to fight ghosting + multiplexing + shift register issues, so I went with the non shift register ones.

[woven bluff]

use shift register without multiplexing

assign each pin to dedicate shift register bit

[supple pollen]

That wasn't going to work with my design without a lot of effort. Amusingly, the reason I breadboard is the same: PCBs take too much time. I can have a breadboard in a few minutes, it takes days for a PCB.

[woven bluff]

now I have a IV18 project

where multiplexing cannot be avoided..

my current solution is to have the second core of RP2040 drive the multiplexing shift register

[supple pollen]

Not the PIO?

[woven bluff]

I'm bad at PIO...

[supple pollen]

Oh yeah, the HV5530 does need 12V signalling, which is inconvenient.

[woven bluff]

no, 5V works

[supple pollen]

It's out of spec though, so it might work but I'd be unhappy to have to count on it

[woven bluff]

even 3v3 works...

if I'm going to spend time learning PIO, I'd better off with FPGA

[orchid phoenix]

One more question, I plan to use a 9V battery to power the arduino nano. The sensors I plan to connect need 5v, is it okay to connect them to the same voltage line for the arduino or is there a way around it?

[unreal flax]

There should be a 5V output from the Nano's regulator you can use. It will have some current limit (and the 9V battery is lowish current to begin with). Sensors are generally not too power-hungry, but don't hook up a motor that way.

[eternal fox]

Anyone knows the maximum units a single i2c controller can handle? Pi pico to be exact. Speed isnt needed really and there will never be any large amount of data sent.
I assume theres a data saturation limit and a device/voltage limit, right?

[knotty tiger]

only 110-ish available addresses (for 7-bit addressing), in practice, far fewer, because devices in the same category often use the same address. there's bus capacitance limits, which can be partially overcome by decreasing pullup resistance

but yeah, it really depends on the details. what specific sorts of things are you trying to do?

[eternal fox]

Im just Going through the various constraints placed on me by using different MCU's for a keyboard project. Trying to see what will be viable and what wont, if i can go crazy and use 10 rotary encoders in one keeb or if 2 rotenc's and 2 oleds are too much 🙂

[knotty tiger]

yeah, i'd suggest looking for rotary encoders with enough address selection lines if you want to put that many on the same I2C bus. OLED matrices can take a lot of bandwidth, and you might find them updating too slowly if you have large ones or multiples

if you run out of I2C addresses, you can use multiplexers, but they add complexity to your code to use

[eternal fox]

Yeah ive looked at multiplexers, roughly what do they do? Do they pretend to be just one device and hiding 5+ behind each of its built in ports?

[knotty tiger]

they're I2C bus devices of their own accord, and you issue commands asking them to connect different back end I2C buses to the front end one that you use to talk to the multiplexer. they can help you deal with I2C address conflicts that you can't resolve any other way (like multiples of a sensor that doesn't have address selection options)

[eternal fox]

Thanks m8 🙂

[vast flume]

Hey guys, how should I learn how to build an ESP-12 based board with support for programming?

My current board design requires it to be plugged into a separate board with a usb to serial converter for programming

I just cant figure out how to automatically set the pins in the correct configuration for it to go into programming mode, without having a separate board

I don't know if my question makes sense, sorry if it doesn't

I can share more details about my current setup if needed

[distant raven]

Not to discourage you from using the esp-12, but why not use an ESP32-s2?

Native usb means no special programmer needed

Huge upsides performance wise too

[fervent lance]

I compiled for esp-32-s2 first time last night (I don't own any hardware). I bothered to split the compiler's output to reasonable length lines and saw how many libraries it compiled in.

[gusty aspen]

Noob question 🙂 I want to put temp/humidity sensor in my attic, and connect to a wifi board. Do I need one of the "enclosed" or "encased" type sensors? If I got a simple breakout board version, would it still work if inside an an acrylic case, assuming it had some holes for airflow?

[unreal flax]

AFAIK there's no reason it has to be in an enclosure at all, if you don't want one.

[gusty aspen]

Well I want to protect the electronics from dust, rodents, etc (we live in an area having huge issues with mice especially)

I guess if we mounted it so the exposed breakout board was facing down, it wouldn't get super dusty at least

[unreal flax]

Gotcha. A box with air holes should be just fine, then.

[gusty aspen]

Awesome, thanks 🙂

[vast flume]

Omg wait

So I can just connect it to a usb port?

I wish I knew this existed before I spent $100 with jlcpcb to manufacture my current boards 🤣

[distant raven]

Yeahhh, ESP32-s2 is great plus sdk, Arduino, circuitpython, and MicroPython support are all very stable

[distant raven]

There’s always V2 🙂

[vast flume]

Very true haha

[vast flume]

Do you have any links to documentation for how to wire it to be able to program via usb?

[distant raven]

ESP-12 should have two UART pins TXD and RXD, those would be connected to the usb serial converter

ESP-TXD —> UtS-RXD
ESP-RXD —> UtS-TXD

I don’t recall if there are external buttons required to put the ESP-12 into download mode though

https://www.hackster.io/brian-lough/3-simple-ways-of-programming-an-esp8266-12x-module-c514ee

But this guide should help

[vast flume]

Thx!

Uh oh I seriously messed up then 😅

Jeez

Thats not good

I wish I saw this previously

I guess I now have $50 worth of paperweights arriving in 5-8 buisness days lmao

[distant raven]

You can program the modules before placing them

Look up “flexy pins” from Solder Party

You can make a jig to program the modules before placing them

[vast flume]

The problem was moreso that I didn't realize I needed to tie en, rst, and gpio high, and gpio15 low

Programming should work though, I was just stupid and looked up the circuit for programming and not also the circuit for normal operation

I guess v1.2 is happening sooner than I thought lol. This was the crazy solution I had for entering programming mode before

But on this version I'll either just use an esp32, or put a usb port & usb to serial chip on my board

[rigid kraken]

do i need a resistor between the led and the transistor? i breadboarded it and it worked but i want to make sure

[supple pollen]

That emitter resistor will give some degeneration, I'm not sure if it's intended as a current regulator or what.

[heavy jasper]

At the moment you have no ground connection listed, so it might be good to get that defined. If your emitter of Q1 is running to ground and just not listed, then R2 isn't doing much for you.

[rigid kraken]

the ground is on the other side of R2. the reason i have it there is when i start up the board, the signal registers as high so i have to pull it down to prevent the led from lighting prematurely

ive bread board tested it and im only running 5 mA through the led so i feel like a current limiting resistor will dim the led

this will all be run off a pico so the max current given to the led at any one time will be low

[obtuse sky]

So ground is in the bottom right and the "+3.3V" is the signal output from the Pico?

The max current from the 5V pin is higher. I'd have thought you do need a resistor. Not sure why you're only getting 5mA at the moment.

Hi all, I noticed that some Adafruit boards have a neopixel powered from 3.3V. This seems to go against the datasheet for the chip used, and also Adafruit's own guides about how to use neopixels. But presumably it's reliable if they're manufacturing PCBs with that arrangement. What's happening here? If I wanted to make something similar, how to guarantee that it's going to work?

[supple pollen]

It is outside of the data sheet specifications, but seems to work well. The chip drives the LEDs with a current regulator, so the LED current and brightness aren't greatly affected. It also allows the LEDs to be controlled with a 3.3V signal without a level shifter. There isn't a lot of headroom, if the voltage were to drop more, the higher forward voltage LEDs would not be able to light as effectively.

[obtuse sky]

Thanks for the info. I guess it depends on the LEDs as well as the chip. I saw one where the datasheet discussed the LEDs too and it said the forward voltage of the blue LED could go up to 3.4V. I guess Adafruit have a particular neopixel that from experience is reliable at 3.3V, and put it on everything.

[supple pollen]

Right: the blue LED is marginal in that application, I suspect the forward voltage varies with temperature so the 3.4V figure may be for a worst-case LED at the worst-case temperature, and most LEDs at room temperature would basically work.

[obtuse sky]

If it doesn't work, is the consequence that you don't get blue, but the other colours are still fine?

Or might it damage something?

[supple pollen]

Exactly, it's the same effect people get with long strings of LEDs where there's voltage drop: the LEDs at the far end tend to be more red as the blue and green chips run out of drive voltage. Nothing is damaged, the higher voltage LEDs just don't light as well (or at all, in more extreme cases)

[obtuse sky]

I see, so even if making a board with an uncertain part, if "not using blue" is an option then that seems safe.

It would be prettier with blue but not essential for functionality.

[supple pollen]

Exactly: when the worst case outcome is "color is different" instead of "parts are destroyed", being experimental is less worrisome.

[obtuse sky]

Thanks!

[distant raven]

I actually measured the voltage going to the sk6812 that I use when powered from the microcontroller pin, it actually pulls the voltage of the Micro up to to 3.6V if the regulator and the micro allows it to be over drawn voltage wise which is interesting.

[obtuse sky]

Hmm, that could be bad though?

[distant raven]

I know some MCU clamp pin output voltage to 3.3V though so this is probably more of an exception rather than a rule

[distant raven]

If the micro can operate at 3.6V, no. Might run down the life of your regulator faster though

[supple pollen]

Oh right, on many boards, the LED chip is powered from the battery supply instead of the 3.3V regulator, so it could be a somewhat higher voltage. And yes, many chips include protection diodes for the I/O pins to limit the voltage excursion.

[obtuse sky]

If you power it from a higher voltage and use 3.3V data, it shouldn't be pulling anything up. Sounds like @distant raven is talking about powering it from 3.3V and it going up to 3.6V on a board that can go up to 3.6V?

If you have a definitely-3.3V board with an exact 3.3V regulator, that regulator won't be putting out more than 3.3V.

[distant raven]

Some regulators allow voltages over their “fixed” rating

And if a pin doesn’t clamp the voltage to 3.3V and allows the pin to work at 3.6V then it’s likely the microcontroller can operate at 3.6V and pulls this from the regulator if it can. Otherwise it will be 3.3V. It’s all very confusing and microcontroller dependent

I’ve only ever seen it happen on samd21

[obtuse sky]

So maybe I don't quite understand, are you powering your neopixel from "3.3V" or 5V or battery?

[distant raven]

USB to a 3.3V LDO

[obtuse sky]

A separate regulator from the one on the board?

[distant raven]

Yes, one I designed

It does have some voltage headroom above 3.3V, many low dropout regulators do

[obtuse sky]

And the neopixel is pulling up its input?

[distant raven]

In my case it was

I honestly don’t think it’s very common

It could have easily just been because of the LDO I use has more headroom than what Adafruit uses

But generally, this won’t really be a big issue.

[obtuse sky]

Were you powering the microcontroller from the same regulator?

[distant raven]

The neopixel was powered from GPIO on the samd21, the samd21 was powered by the 3.3V LDO

[obtuse sky]

Ah OK, so just an oddity with your regulator I guess. Premade boards with an on-board regulator will be set up so the regulator doesn't output a voltage that's too high for the chip.

[distant raven]

The samd21 can operate at 3.6V, you will just find that most designers opt for 3.3V since it’s a far more common regulator to find

[obtuse sky]

And better compatibility with everything else

[distant raven]

Right

[unique patio]

i have never seen RGB LEDs we use not give reasonable colors at 3.3v

[vast flume]

If I am using an ESP12f, do I need to add pullup resistors to the i2c bus? I saw online that it should have internal pullups but just wanted to make sure

Also what is the difference between self powered and bus powered for the CP2102N USB -> UART chip?

[knotty tiger]

i think internal pullups on a MCU are rarely strong enough for plausible I2C scenarios. internal pullups tend to be 20k-60kOhm, which is a lot weaker than normally recommended for I2C

[vast flume]

Interesting. In that case, do I just need one set of pullup resistors for the entire bus?

[knotty tiger]

i think a single set of pullups is usually sufficient for short I2C buses (e.g., contained on a single small board)

[supple pollen]

It's a shared bus, it only needs one set of pullups (the pullups are shared too)

[vast flume]

ok

And then for using the CP2102N chip, should I use the bus powered mode or self powered mode?

[distant raven]

Either is probably fine. In self powered mode it utilizes a 3.3V internal regulator where as bus powered would need a 3.3V supply available.

I personally like bus powered mode with an external 3.3V LDO

[pale umbra]

Looking for a DIP level shifter for I2C. I give up on trying to solder the normal ones even adafruit version since it still have those extremely tiny resistors

Would be perfect if it could also buck-up the voltage

[distant raven]

https://www.adafruit.com/product/1787

If you operate at 100kHz i2c bus which is pretty standard, that should work in theory

Or this would be fine too: https://www.adafruit.com/product/1875

[vast flume]

Ah ok, makes sense

[pale umbra]

my voltage is 3.3V though, not 3V, does it need a particular voltage? one I currently have just have LV and HV no particular voltages

[pale umbra]

yeah that's the kind of the ones I couldn't solder without desoldering the tiny resistors on it 😦

[distant raven]

That has 3 other parts that shouldn’t be an issue

[pale umbra]

ah 2nd one doesn't work for i2c: "nly thing that doesn't work well with this chip is i2c (because it uses strong pullups which confuse auto-direction sensor)" same version for i2c would be fine, also not restricting myself to adafruit (as I mentionned before I can only order in Canada)

If else fails I guess I'm going to have to get 2022 3.3V LCDs / OLEDs 😦

[distant raven]

Oh whoops didn’t see that part 🫣

[pale umbra]

thanks even if I don't completely understand it the 1st one seems to be good

[vast flume]

One more question actually, on my pcb footprint there are pads SH1-SH4

Are these just to improve the physical connection to the pcb or do they need to be routed somewhere?

[knotty tiger]

if "SH" stands for "shield", making sure they're connected to a solid ground plane is probably a good plan (though thermal relief might also be needed to make it easier to solder)

[vast flume]

Ah ok thx!

Its going to be assembled by jlc so I don't have to worry about soldering

It'll be attached to the ground plane though

[knotty tiger]

oh, ok. in that case, hopefully their DRC should flag your design if it needs thermal relief there

[vast flume]

Should the ID pin be left floating?

@knotty tiger

[knotty tiger]

it's used for USB OTG (On-The-Go)… i'd have to look up what the spec on that is

[vast flume]

Ah ok, so if I don't want to use that protocol then I don't need to worry about it?

[knotty tiger]

https://en.wikipedia.org/wiki/USB_On-The-Go#OTG_micro_plugs
probably floating for a device (not a host), unless maybe you're doing high-current charging?

[vast flume]

Yeah floating should work then

Am I correctly reading that I just put a 1k resistor between RSTb and VIO?

[unreal flax]

Yep, that's how I read it too.

[distant raven]

And it works too 🙂 I recently used the cp2102n on my board here:

Don’t mind the red circle, I was showing how I used a truck I learned to get a resistor value close to what I needed lol

[vast flume]

Thanks!

For the 4.7uF and 0.1uF bypass caps does it matter which order they're in?

They aren't labeled in the datasheet schematic

[supple pollen]

It doesn't matter much, but my thinking is that the 0.1µF is for the higher frequency noise, so I tend to put it closer to either the part (the part that is either the noise source, or one I'm protecting from noise).

[vast flume]

Ok, that makes sense

[vast flume]

Does this seem like the correct setup?

[distant raven]

Seems right

[vast flume]

Thanks for the help today! 0% chance I would've gotten this right without everyones help

[distant raven]

That’s what we’re here for 🙂

[pale umbra]

how did you produce that schematic ? Been looking for a replacement for fritzing

[gusty aspen]

Does anyone have ideas for adding wings to a feather, without stacking headers?

I've got a great working prototype here for a project for my house, with my Huzzah ESP8266, the radio wing on top, then I decided to add the oled wing to help monitor/debug. I'm ready to move it into a permanent case - but the radio wing would be in the middle and it has normal headers, not the stacking ones.

I considered desoldering the headers and putting stacking ones on, but that's so tedious 😬 I have them in the "tripler" right now, so I can use them, but it's a bit expensive to devote to a permanent project, I'd like to reuse it for other prototyping. (Even the doubler feels pricy for that solution)

[gusty aspen]

I guess another option might be to grab a cheaper OLED module and hook it up, and save the OLED wing for future projects

[pale umbra]

not sure what is the problem here, that seem super expandable and you can also use standoffs or 3d print small rods or enclose stuff

[gusty aspen]

I don't want to devote the $8.50 prototyping board to a permanent project and have to buy another one I guess

[supple pollen]

I figure it's cheaper than the radio wing, and cheaper than my time to unsolder/resolder headers. However, that equation is both different for everybody and nonlinear (for example, you can only spend the amount of time/money you actually have to spend)

[gusty aspen]

Right, the soldering time is a big factor. 🤔 hmmm

I was just hoping there might be a solution I wasn't seeing 😄

[supple pollen]

One possibility for stuff you're building into something is just attach them to perfboard and run wires to connect the signals you need (for I2C Wings, it is often just a few wires)

[gusty aspen]

Oh actually I think the OLED wing even has a stemma QT connector, which would make it even more simple to do that, I just need to wire up the connector to the main feather

there we go

Thanks for the hint there 😄

[vast flume]

KiCad

Does it matter if I put the pullup resistors for the I2C bus before the first slave device or can they be positioned anywhere along the bus?

[distant raven]

I think it’s common practice to put it either next to the host pins or the peripheral pins

[vast flume]

Ok

[pale umbra]

speaking of which, is it possible to sort of (tiny production)mass produce something in the form of proto board instead of a pcb so someone can say easily desolder something and re-add it back etc ?

ie: like this but ordering say 10 of them, I swear that i've seen a radio electronics kit for children once that came with a circuit in that board pre-assembled in a clear case to learn

[knotty tiger]

i think you'd want to make it as an unassembled kit, so they can use leaded solder if they want. desoldering ROHS-soldered boards isn't very beginner-friendly

[pale umbra]

oh it's not for that, just for reuse for myself

[limpid nest]

having a goof moment. An "off-mom" switch is open circuit until you press it right?

as in, it's SPST-NO, no?

[limpid nest]

Unrelated, does this qt joint look OK and does anyone think the pins are mess up here?

I'm having issues with the circuit this connects to and I've tested it a bunch and all I can come up with is: it's the connector

[inland jungle]

pins look a bit messed up, try and poke the right 2 pins to center them a bit better

[twilit mango]

They aren't centered

The left two need to be closer to the side of the connector.

Joint looks ok to me.

[inland jungle]

that's not a stemma connector?

[limpid nest]

I think it was the cable actually

I just tried a new cable and tests are passing that weren't passing before

yeah this is a JST SH connector, I got it from adafruit

[twilit mango]

Ah yeah. Cable could do it.

[limpid nest]

tsk tsk, should have gone with adafruit for the cable

[twilit mango]

Still, careful with those slightly shifted pins

[limpid nest]

(you were sold out of the length I needed)

[twilit mango]

A cable could cause them to get smushed.

[inland jungle]

ah yeah, pins should be in line with the notch

I smooshed a pin once, but was actually able to recenter it

[limpid nest]

yeah

I really want this to work without replacing that connector

So we'll see

[twilit mango]

Should work. Simply watch for the pins. 🙂

[limpid nest]

it's in a sensitive spot

I am going to make a new revision at some point that doesn't use small pitch connectors for things that don't need them!

[vast flume]

Finally finished routing V2 of my flight computer, with all of the help that y'all have provided! Thank you guys so much

[distant raven]

Maybe it’s the image quality but the ground pad of U6 looks connected to all the pads 😳

[knotty tiger]

it might be a poor choice of layer colors

[vast flume]

Heres a closeup of U6

[distant raven]

Ah good, looks good 👍🏻

[vast flume]

Is the way I connected the gnd pad to one of the gnd pins ok?

[distant raven]

I’d drop a via or two in the ground pad for better thermal conductivity

[supple pollen]

SparkFun took another approach with their Protosnap, which had an Arduino and several peripherals that you could snap off and then re-use by attaching wires/headers. The 1BitSquared iCEBreaker has a similar idea with snap-off LEDs, switches, etc.

https://www.sparkfun.com/products/retired/10817

[vast flume]

Good point

[obtuse sky]

Hi all. The Pi Pico datasheet says, "One thing to note is that the ADC capable GPIO26-29 have an internal reverse diode to the VDDIO (3V3) rail and so the input voltage must not exceed VDDIO plus about 300mV. Also, if the RP2040 is unpowered, applying a voltage to these GPIO pins will 'leak' through the diode into the VDDIO rail. Normal digital GPIO pins 0-25 (and also the debug pins) do not have this restriction and therefore voltage can safely be applied to these pins when RP2040 is unpowered." However, the RP2040 datasheet, Pin Specifications, Absolute Maximum Ratings say "Voltage at IO: Maximum: IOVDD + 0.5V". How does that add up?

[distant raven]

I’m confused by what you’re asking. Asking how the analog can be VDD + 0.3V vs standard GPIO being VDD + 0.5V?

[obtuse sky]

Pi Pico datasheet says it's safe to apply voltage to digital GPIO when unpowered, but I interpreted RP2040 datasheet as saying it's not.

[distant raven]

I don’t think it would necessarily be a good thing to apply voltage to the pins of an unpowered device in the first place. There’s always a chance for voltage leakage on any of the pins.

Are you planning to apply voltage to GPIO when the rp2040 is unpowered?

[knotty tiger]

the RP2040 datasheet lists GPIO0-GPIO25 (among others) as fault-tolerant, which have different electrical characteristics than the others

Fault Tolerant Digital. These pins are described as Fault Tolerant, which in this case means that very little current flows into the pin whilst it is below 3.63V and IOVDD is 0V. There is also enhanced ESD protection on these pins. Programmable Pull-Up, Pull-Down, Slew Rate, Schmitt Trigger and Drive Strength. Default Drive Strength is 4mA.

[distant raven]

That said, the “just because you can, doesn’t mean you should” concept applies here in my opinion.

But generally, I don’t think most people are setting their devices up in a manner in which this would be an issue

[knotty tiger]

powered-off devices still connected to an active I2C bus seems to be a common situation where this can happen

[distant raven]

Sure, that’s fair.

[obtuse sky]

Yes it's something like that though not exactly

[vast flume]

Does anybody no where I can find a library for the HP203B for kicad?

I've looked on snapeda but they don't have it

[knotty tiger]

yeah, i think that technically it's outside of the specified "absolute maximums" to have a pin voltage greater than 0.5V when IOVDD is 0V. i read the fault-tolerant pin annotation to mean it's less likely to damage the chip, but still out of spec @obtuse sky

[distant raven]

Try ultra librarian?

[vast flume]

No luck 😦

I might have to choose a different part then sadly. This one was really nice bc it is only $1 and it is able to calculate altitude itself, instead of just providing pressure data

[distant raven]

You could make the footprint based on the data sheet

[vast flume]

Ah true yeah I might have to do that

[distant raven]

It’s a good skill to have 🙂

[obtuse sky]

Yeah, that sounds right. So I'm puzzled why the Pi Pico datasheet says it can be done "safely". But clearly if somebody is going to insist on doing it, they shouldn't use an analog pin.

[distant raven]

I think they are probably attempting to convey the same thing, it was probably lost in translation between the documents

[knotty tiger]

i think that specs like "absolute max VDD+0.5V" really mean something more like "forward-bias voltage of protection diodes at 'safe'/non-damaging currents is 0.5V; exceeding that safe current will increase forward-bias voltage to more than 0.5V", but phrasing it in terms of voltage means not having to explicitly specify that max fault current, and the user can figure out whether the impedance of such voltage sources is high enough to avoid problems

[obtuse sky]

Usually yes, but I think these digital pins don't have diodes?
And on the analog pins, not sure what it means if you get leakage onto VDDIO. I guess then VDDIO>VDD and that's what the problem is?

[distant raven]

Back powering a bus can cause damage. And usually analog capable pins are more closely coupled with VDDIO than standard GPIO

[knotty tiger]

even if there aren't explicit protection diodes, CMOS output stages might have MOSFET body diodes as a result of details their construction, as i recall

[obtuse sky]

I see, though it was saying that the current would be very small in this case

[distant raven]

But going in the wrong direction if it’s back powering.

Which is where damage can happen, even at low voltages

[knotty tiger]

unfortunately, the RP2040 datasheet doesn't give much in the way of equivalent circuit models for its I/O pins, which might be helpful to have

[obtuse sky]

As you pointed out earlier, it says "very little current flows into the pin whilst it is below 3.63V and IOVDD is 0V" so it really does sound like it's voltage in this case.

[knotty tiger]

yeah, that would tend to imply a CMOS output stage design that didn't have body diodes; i'm not sure how they would arrange the biasing so that none of the junctions ends up forward-biased when given a voltage higher than IOVDD

[obtuse sky]

On simple chips with clamp diodes to VDD, it's more about current. Though also, inputting from a higher voltage using series resistors can make VDD go up and hence the output, which can be an issue even if it's not harmful to this chip.

[vast flume]

When designing a footprint in kicad, is it from the "top" or the "bottom"?

Like should I envision that I have the chip flipped over and am looking at its pads, or that I am looking at the chip from the top?

[obtuse sky]

Top

And normally you draw your board from the top, but if you have a two-sided board you flip some of the chips over

I mean on a one-sided through-hole board the tracks normally go on the back.

[vast flume]

Ok

I've also noticed that some other ic footprints will have a dot on their silkscreen layer. Is this so the manufacutrer knows how to orient the chip?

[distant raven]

Usually, yes

[vast flume]

If so, in my case would I position it off of pin4?

[distant raven]

I think that’s actually the hole for the barometer

But you could use it to help remember alignment during placement

Usually the dot on the silkscreen is for indicating the index pin, or pin 1

[knotty tiger]

yeah, using similar index mark positioning as used in the manufacturer's package markings will minimize confusion during assembly

[limpid nest]

when powering a feather board thru it's USB port, do you need a diode or does it have one? It seems from the schematic that you do need one

[distant raven]

The USB pin?

[limpid nest]

Yeah pin sorry

[distant raven]

Are you planning to power from a USB port at the same time as powering from the usb pin?

If not, you could probably skip using a diode (schottky)

If you are powering through other means as well, it might be worth adding a Schottky diode

[limpid nest]

I am not the only user

So safety takes priority

[distant raven]

Wouldn’t hurt to use a mosfet in this case if you anticipate there being power already on the USB bus

[limpid nest]

What does that look like?

[distant raven]

I’d have to think about it

[knotty tiger]

there's a p-MOSFET polarity protection circuit between VBUS and VCC on many Adafruit dev boards, if you want to look at the schematics for examples

i think a Schottky diode will be good enough, if losing 0.2V or so of voltage to the diode drop doesn't cause problems for you

[limpid nest]

Oh I've done polarity protection before

ok I'll look at it, thanks

on a P channel symbol, the body diode points towards the source, right?

[limpid nest]

I'm not seeing how that can be adapted for this, given that I'm off board? Maybe I'm just not seeing it

I can see how drain and source would be set up: put your external 5V on drain, and source to the USB pin, but then what happens with the gate? connect to USB pin as well? that seems wrong

[obtuse sky]

The Pi Pico datasheet has schematics and explanations for doing it on a board that doesn't have one already

[limpid nest]

related, I'm trying to model this in Falstad but I can't get a single terminal 5V supply to attach to anything. No matter what I do, it just gives me "1 bad connection"

[distant raven]

It would be easier if the feathers had both VUSB and 5V to the LDO broken out

[limpid nest]

let me check that out. Do you recall off hand the area of the datasheet?

[obtuse sky]

"powering pico" I think

[distant raven]

Then you could drive right to the 5V pin if vusb is 0

[limpid nest]

indeed

I may just live with a diode. What's the benefit of this method?

[distant raven]

Reverse current protection

[limpid nest]

oh I can always add that up stream

[obtuse sky]

What's your 5V source?

[limpid nest]

the diode protects well enough, no?

[limpid nest]

Some undetermined voltage regulator

[obtuse sky]

I mean do you care about saving a little bit of power?

[limpid nest]

not in the least

not a battery application

[obtuse sky]

Diode is fine then

[limpid nest]

that's what I thought but I wanted to know

[obtuse sky]

Mosfet better for battery applications

[limpid nest]

but you need more voltage pins than just the USB pin

I guess you could use the 3v3 regulator

since that's strung to the 5V

[distant raven]

Regulate down to the 3.3V pin before applying voltage though

[obtuse sky]

Oh yeah, the Pi Pico does, so if this board doesn't, Pi Pico instructions might not be relevant

[limpid nest]

say that again?

[distant raven]

Sorry typing with one hand

You can regulate your source down to 3.3V and apply it right to the 3.3V pin of the feather

Avoid the usb pin all together

Then you can use a p-chan mosfet in line with the 3.3V pin that toggles on when vUSB is 0V

[limpid nest]

oh interesting

I didn't realize

I need a 5V line as well but I can just have two regulators

[distant raven]

Yeah, that’s pretty typical

Have to source regulators

[limpid nest]

are you aware of any feathers where you can't "back drive" the 3v3 line?

[distant raven]

Not that I’m aware of

Regulators usually have reverse voltage/reverse current protection to keep the device safe. As long as you apply 3.3V to the pin, should be more than safe

[limpid nest]

so I need a P channel mosfet that's turned on at -3.3 G-S?

[distant raven]

Basically follow the orientation of the P channel mosfet in the battery voltage selection on the feather

[limpid nest]

that's what I thought, maybe I can even use theirs

[distant raven]

Treat the 3.3V regulated output as the battery voltage, and the 3.3V feather pin as the regulator input on the feather. VUSB goes to gate

The same mosfet should work too as you were thinking

[limpid nest]

that's what I was thinking, thanks

[knotty tiger]

i think basically all DC regulators have to tolerate their outputs being near their set point voltage even with no input, because otherwise, shutdown wouldn't work if there were any capacitance on the regulator output

[limpid nest]

hm, I can only find one suitable pfet

actually it's BGA

[vast flume]

Oof this is expensive 🤣

[distant raven]

Orientation of the esp12 looks off 🙂

Good ole JLCPCB renders

[vast flume]

Yup 🤣

Same thing happened with my last order but I contacted their support and they said not to worry

[limpid nest]

I need to fit together N of something like this: https://www.te.com/commerce/DocumentDelivery/DDEController?Action=srchrtrv&DocNm=796692&DocType=Customer+Drawing&DocLang=English on a board. Rather than deal with a manufacturer CAD part, I feel like it makes more sense to make a custom footprint that's N x 1Length long. thoughts?

That should just clip together and fit in, no?

[tough matrix]

After a really long absence, stopped by to check what is new here.
Am I right that this is the new name for "help with pcb design" channel?

[distant raven]

Yup! It’s been renamed to be more broadly focused on electronics hardware design

[limpid nest]

Welcome back!

Glad to have you

[tough matrix]

thanks

[limpid nest]

Eventually I'll get around to my motor driver board and bug you

[vast flume]

So I made a mistake

Again

I didn't know tantalum caps were polarized

So I put one of these as a decoupling cap on my pcb, not knowing it was polarized: https://jlcpcb.com/partdetail/140521-CA45_A_16V_4_7uFK/C129235

Is it possible that there would be a ceramic cap with the same footprint? Or am I screwed again

[distant raven]

Nah, not screwed. 4.7uF is a common ceramic value

But it should be fine if it’s polarized.

So long as it’s oriented the right way

[vast flume]

Ah

Thats good to hear

In theory would it work if I had them replace them with this? https://jlcpcb.com/partdetail/30577-1206B475K500NT/C29823

[knotty tiger]

i vaguely recall that tantalums have lower ESR than aluminum or "wet" electrolytics, so you might be OK anyway

[vast flume]

It looks like it should be the same footprint, just defined in imperial vs metric

[distant raven]

When coupled with a 0.1uF or other small value caps for input capacitances, it’s usually better for larger value to be electrolytic. Since it’s just bulk capacitance more than filtering noise.

[vast flume]

I see

Ok so if I do go with keeping the tantalum cap, I just need to make sure that the gnd side on the cap is connected to the ground pad?

[distant raven]

Is that on the CP2102n?

[vast flume]

Yeah

[distant raven]

Okay, Yeah should be more than fine

[vast flume]

Ok

Good to hear, thx

Glad I chose "verify part placement" this time lol

[distant raven]

I use tantalum for 4.7uF, ceramics also work if you rather avoid toxic fumes from failure

There are also aluminum polymer which failure wise tend to be a little safer

[knotty tiger]

if, as i recall, this is going inside a high-power model rocket, fumes from possible capacitor failure are probably not going to be significant in comparison to other emissions…

[vast flume]

🤣

[distant raven]

Very true

I know Harry recommends ceramic where possible

There are some ethical concerns with tantalum caps too

[heavy jasper]

The tantalum is rotationally symmetric, I think, so you could just mount them “backwards” and call it a day.

[vast flume]

JLC is assembling them for me

[heavy jasper]

As skerr mentions - tantalum is a conflict mineral so I try to avoid it, especially from lesser-known/vetted suppliers

[distant raven]

I’ve blown tantalum by backwards mounting lol..

A nice fireball on failure

[vast flume]

All should be good now though, I thought their email was saying that I chose the wrong part but really it was just verifying the polarity that they chose for assembling the caps. The polarity was correct so it should be good hopefully

I got scared and thought I wasted another $100 lol