help-with-hw-design

[woven bluff]

half bridge

[idle python]

i deleted all traces im gonna start over again

why are 90 degree traces not used often?

[supple pollen]

Amusingly, this came up recently: #help-with-hw-design message

[hushed smelt]

having footprints a little longer than necessary is good for bodge wire fixes too. also provides some leeway for the pick & place machine.

[idle python]

oh yea

acid traps were an old problem now, with modern manufacturing theyre not as often

[idle python]

makng traces should be considered in iq tests

especially when you have to connect 26 nets

in a 2x4cm space

[supple pollen]

Definitely a test of spatial reasoning, problem solving, and optimization

[idle python]

theres one thing i dont understand here

do i route the traces for MOSI MISO and CHIPSELECT coming from the resistor, or from the IC

or does it not matter?

[unreal flax]

It shouldn't matter... the voltage connectivity will be the same. (Which is not always the only consideration, but with just a pullup resistor, it's fine.)

[distant raven]

Acid traps ends up being a far more valid concern than the train of thought about it truncating high speed signals. IBM researchers ended up debunking it with a caveat of speeds above 10GHz actually saw some weird reflections if I am remembering properly

But now a great reason to avoid right angles is issues related to trace adhesion and them getting ripped up when solder mask is applied

[supple pollen]

I didn't know about that one. I just like the look of angled traces, so I normally route them that way, except for things like T intersections and traces coming into rectangular pads.

[dry pelican]

They act as constant current sources. They also limit the current. You need some sort of impedance so the tank isn't connected to high all the time and so the MOSFETs don't blow up

[woven bluff]

any recommendation of online simulator?

[dry pelican]

Ltspice, microcap, falstad. Everycircuit is also a good real time sim

[idle python]

how do i fix these clearance errors? im trying to connect the 2 gnd copper planes together with vias but it gives me this error

[woven bluff]

any idea why my DS3231 breakout slowed almost 1 hour in one day?

the battery still has 3V

[dry pelican]

Edit via properties and change net to gnd

[hushed smelt]

All via's have net's associated with them. By default plopping down a via has no net label.

rename the via net to GND

then what I usually do is copy the gnd via to other places where i need it, because the net label is copied with it. 😉

as for the 2 GND net's still showing rat lines on your chip is because they need to share a common ground, they have to connect to the same gnd plane somehow.

if the top right gnd trace at the chip is connected to a gnd via somewhere else then your idea of placing a gnd via where you did should hopefully work to connect them together.

something like this could also work

[idle python]

finally done but i have a problem

sim datasheet doesnt tell me the max or minimum voltage of net_status

[idle python]

(this is the one from ec25 model but it may not be the same)

this is all there is from eg800

[distant raven]

Probably the same range as other IO given that it’s only driving a transistor

[idle python]

i dont have space for a transistor and 2 other resistors

[distant raven]

What are you doing with it?

[idle python]

with what

[distant raven]

The net_status pin

[idle python]

led indicator

[distant raven]

It’s range is 0.4-1.35V, you’re not going to be able to drive an LED without a transistor

What module is this?

Link to data sheet?

It’s also likely the pin isn’t intended to drive very much current

[idle python]

this is the global version but im using the chinese version cause its the only one i could find in stock

[idle python]

this is the chinese version datasheet but i cant understand anything

[distant raven]

Okay, it’s characterized by VDD_EXT which is 1.8V @ 50mA

You could probably use a red LED rated for 1.8V

https://www.digikey.com/en/products/detail/QBLP601-R5/16499683

Something like this with a low test current

Then you can use like a 10k resistor and drive it really low current

It would still be fairly bright

[idle python]

that doesnt matter, itll be covered up with plastic anyway

[distant raven]

If that’s the case leave it floating

[idle python]

cant, if i dont have those leds i cant tell if the board actually works while testing it

after the entire thing is done they will be covered

do they have other colors?

[distant raven]

Technically yes, but it’s all like orange/red LEDs due to forward voltage characteristics of LED materials

[elder peak]

This may color your opinion of the available options.

[dry pelican]

maybe make it work as a sink rather than a source?

though that might anger the magic pixies

i.e. blow the chip due to you exceeding the tolerances

[distant raven]

It’s output only so that wouldn’t work

[dry pelican]

huh. Never heard of a source only output.

[distant raven]

It’s not characterized for input in the datasheet, just listed for powering external circuitry

[idle python]

finished the board

[supple pollen]

I'm unsure what the silkscreen (I'm guessing the teal is silkscreen) in the lower right corner is, but it looks like it might be over the connection pads for the U.FL connector. You may not want that, as it could interfere with soldering the connector in place.

[hushed smelt]

in EasyEDA silkscreen is yellow on the front (red copper) and dark green on the back (by default but you can change the layer color for anything). Doesn't look like any silkscreen like the silkscreen layer is hidden along with all the component labels.

and even if a silkscreen is overlaying a component, in EasyEDA they automatically ignore silkscreen on components. You could put a huge silkscreen graphic on the entire thing and it'll automatically not print it in the appropriate places so no components or through holes are printed over. You can silkscreen over custom gnd via's though. It's one of the things that makes EasyEDA really easy to work with.

Here's a good example of graphics on the back silkscreen (automatically reverses it for you).

With EasyEDA silkscreen will never interfere with components 😉

[supple pollen]

Not sure what teal is then (I've never used EasyEDA). Doesn't look like copper, paste, or soldermask. Could be a "place" layer, I suppose. Not much on it, other than the corners and pin 1 marks for the big chip/module, the big H-shaped mark (slide switch?), and the concentric circles, dots and line I referred to originally.

[hushed smelt]

There is no teal by default so he must have changed the layer colors.

red is top copper layer, blue is bottom copper layer

and you can set any layer invisible, mostly to make each layer easier to see while working with it but also handy for screenshots in getting help with a layer.

[supple pollen]

Thing to figure out what the sequencer bell actually does. I'm imagining a row of solenoid drivers hitting chimes with rubber tipped plungers or somesuch.

[hushed smelt]

midi sequencer based on TR-808 😉 it's uhhh a long term work in progress.

[supple pollen]

I think I've seen it before, just distracted by the use of the word "Bell". Possibly because I've been watching Anna Lapwood videos about pipe organs with bells.

[hushed smelt]

FoamyGuy played around with it on livestream and actually made it come to life. the most i could do was get it to blink in sequence.

https://www.youtube.com/watch?v=QR1UXm6L_6I

[rustic linden]

I assume it's a play on Cowbell?

[hushed smelt]

Adafruit Raspberry Pi Pico add-on's are called "bells" and if it's a Pico W then "cowbell"

like a shield, featherwing, bff, etc...

[supple pollen]

Ah, that's the missing clue I needed!

[rustic linden]

That's what I thought :3

[hushed smelt]

it's just a huge pico cowbell addon for the pico w. massive yes but just an addon at it's heart.

because the pico and pico w footprint are interchangeable it can be a bell or cowbell depending if you use a pico or picow.

[idle python]

yes it is silkscreen but that looks more like green to me

the silkscreen is intersecting a bit with the gnd pads

i removed it

[idle python]

i didnt change the colors

[rigid fiber]

What is this symbol? It looks like a weird mosfet.

[distant raven]

kind of looks like a buffer

[dry pelican]

I think that's 2 p channel mosfets. Kinda non-standard

[supple pollen]

The little ring is the bit that makes me curious. I'll often see that representing some sort of logic inversion, but it could be because they're P-channel. I sincerely doubt it means they're depletion mode.

[rigid fiber]

This was the entire circuit diagram. It's from the spec sheet examples for a chip that wireless power reception.

In case the context helps

[supple pollen]

It looks like the same sort of back-to-back MOSFET switch used to connect/disconnect LiPo cells from loads

[rigid fiber]

I was toying with the idea of a Qi-powered ESP32 device. 🙂

[supple pollen]

Ah, TI suggests CSD75205W1015 as a possibility for Q1

https://media.digikey.com/pdf/Data Sheets/Texas Instruments PDFs/CSD75205W1015.pdf

I lifted that from the documentation for the evaluation board, which might contain other useful information if you're implementing things with that chip: https://www.ti.com/lit/ug/sluua25c/sluua25c.pdf

[formal sand]

I've seen a similar thing used for load switches

And battery protection on lipos

[elder peak]

Well, @supple pollen should be FETed for having the right answer.

But, yah, back-to-back FET's are a super-useful thing which is why you find a whole world of pre-packaged versions thereof.

[rustic linden]

If you figure it out, I'd be ever so interested

[idle python]

help the mp1482m isnt produced anymore

any alternative that is the exact same as mp1482

[woven bluff]

for BUCKs, I tend to by modules

from XP, Recom, Traco, etc

[formal sand]

You might be able to find a buck with the same footprint if you look in digikey or mouser

After 2020 you basically have to buy your ics as you are designing your board. I've been in situations where I have finished my schematics and half of my components went out of stock lol

[idle python]

someone made that with an esp32, but it charges the battery inside, its not fully powered just by the qi coil

i think theres a video on it

https://youtu.be/hteut2R8eQ0

[dry pelican]

Exactly. I found this TI "simple switcher" buck converter IC. Not sure about part number

https://www.mouser.com/ProductDetail/595-LMR51420XFDDCR

[elder peak]

I've had really good luck finding Diodes Inc's switchers in stock. The problem is the datasheets are awful.

[dry pelican]

Peak chip shortage (this was actually taken yesterday)

Imagine waiting 2 years and 2 weeks for a part

[echo zephyr]

WHAT

2 YEARS

[worldly schooner]

Sounds about right.

[distant raven]

At least some samd chips are back in stock

Digi-Key has a few thousand samd21g18a in stock currently

[dry pelican]

2021 was "um I can't even find an ina219 in stock in the small package"

A lot more are stocked than in 2021

But there is still shortage

[woven bluff]

and the raspberry pi...

[formal sand]

I remember having a client pay $28k for 150 ics that would have costed 1.2k....

Dark times....

[distant raven]

This is getting better

[formal sand]

Its so much better now

[distant raven]

I’ve seen more stock come through the last 2-3 weeks than the last 4-5 months

[formal sand]

Interesting

[distant raven]

Especially 4B

[formal sand]

What kind of ics have you seen that change on? Mcus?

[distant raven]

Samd21

A number of STM32

Though I was referring to the raspberry pi SBCs with my statement about recent stock changes

Speaking of https://www.hackster.io/news/raspberry-pi-s-eben-upton-points-to-stock-level-improvements-predicts-full-availability-in-months-3da382f6176d

[spice turtle]

This is why I bought some dsPIC33CK's for a project I might want to do. Although I feel a bit of guilt knowing they might just sit there being hoarded

[distant raven]

I’ve found things are coming back in stock more often than before

I have been holding on to samd21e18a for a while. They’re pretty well in stock now though

[spice turtle]

You know, I looked into that series and it didnt have what I wanted. Same with STM32. I even looked into the newer PIC32 and they werent at the same price point as even the SAMD chips. I did want to give them a chance though, seeing as I have the programmer (PICkit)

On the subject of drivers:
I just bought this chip for a potential digital buck smps: MCP14628-E/SN
For boost stuff, Ive used two transistors and it seemed to work well.

[dry pelican]

I found some I2C controlled synchronous buck boost converters a while ago

Mpq8875

[rigid fiber]

Is there a circuit that could take either the the D+ or the D- line from a USB port and split it into two, the one way goes to the ESP 32 chip as expected, the other way goes to a LED to make it blink, without corrupting the signal on the first half?

[dry pelican]

Buffer

Or a transistor

Don't really know if that would mess up the impedance somehow

But probably not

[rigid fiber]

By buffer do you mean a voltage follower with an op-amp?

[supple pollen]

That's one approach.

[formal sand]

do you want it to blink when there is data transmission happening?

[rigid fiber]

You got it. Blink on data transfer.

[distant raven]

That might be hard.. you’d be better of just doing that in firmware and blinking an LED like say the Adafruit Feather M4 does

[dry pelican]

yeah you would need to do it in software bc USB is differential

and one line might always be high or something

[supple pollen]

I think even when USB is "idle", it's sending clocking data back and forth, so the LED would always be partially lit

[hushed smelt]

Worked out in the end, It does not produce nearly enough light to be used as a workshop lamp. I would have been disappointed anyway.

[supple pollen]

I ended up buying a fluorescent fixture designed for warehouse lighting as a workshop lamp. 31000 lumens is indeed enough light.

[hushed smelt]

Oh I have overhead workshop lighting. I just wanted to fix a closer reticulating arm style work lamp. This would not have been a good replacement for that.

[rustic linden]

I have a bunch of LED ceiling lights I'm considering turning into a workshop lamp

[hushed smelt]

Yeah apparently cree is the way to go for stuff like this.

[supple pollen]

These things are a compact source of lots of light

[rustic linden]

i have a wide range of lightbulbs, but I'm not sure lightbulbs are the best solution

[supple pollen]

LED vs induction: Thunderdome!

[hushed smelt]

That's what I was trying to replace and ended up fixing the power supply issue.

At least the LED's and pcb design works. It's the only thing I didn't screw up with the whole project. :/

[spice turtle]

the challenge there is that most converters are in crazy packages like that. There are very few that arent QFN and are inexpensive.

To do it digitally, it broke down in the following way:
dspic33CK: $1.64 (SSOP)
Driver$2.00.
Total:3.64
(mosfets and passives are excluded in this since they are common to both circuits)

There was one that would fit the bill, and it came in slightly cheaper, which cost $2.84. There was a more expensive one (maybe higher frequency) that comes in at more than 77 cents more expensive than the digital version. There are some advantages of going digital imho 🙂

(

[rigid fiber]

That would be fine. I figured it would be kind of PWM-ish and get brighter and dimmer between idle and actual data transfer.

[supple pollen]

I suppose it depends on the modulation scheme. Perhaps I should get out an oscilloscope or an LED and buffer circuit and have a look.

It's some variant of NRZ (not return to zero) with bit stuffing to maintain synchronization

[lunar cosmos]

Question - I have a 5v/3a supply on my board from a TPS54331. I followed the TI Webench for the design and component selection.

When I pull 3A, the voltage sags to 4.4v. Is there any way to keep it at 5v?

Also, I was trying to measure noise on the 5v rail with my oscilloscope, but I only read 150mv peak-peak, which seems kind of low. Can someone help me with finding the proper settings for measuring noise?

[supple pollen]

It could be your upstream supply is sagging. Curiously, noise measurements are normally limited to 20MHz or so.

[rigid fiber]

Are there any things out there for esp32-family QFN chips that you can just put in the QFN chip and tighten it in place without soldering, and it breaks out the pins, provides a USB port, and so on?

[supple pollen]

There are QFN testing sockets with spring loaded pins. You'd have to mount that socket onto a board that provided the additional connectivity

[rigid fiber]

cool

Somebody at the local maker, their employer is willing to let us tag along on an order to help get it over 10,000, but it means 200 chips for 5-10 people. But it's like half price for an ESP32-S3-WROOM-1-N16R2 compared just getting couple from Digikey or Mouser or LCSC.

[supple pollen]

I wanted a somewhat obscure chip for some breadboarding and I found a reseller that had 22 of them, but I had to buy all of them. I did, so I have spares, I guess.

[hushed smelt]

you would need a custom jig for the chip but that wouldn't be too much different from using a custom pogo pin or stamp setup.

making sure it's flat and making good contact on a flat surface seems like a harder problem to solve

[supple pollen]

That's what the testing sockets do, basically

[paper vale]

Hello I want to measure phase currents of a BLDC motor to use in FOC. I'm finding the best method to adapt the signal for ADC. What is the best method between adding an DC offset and using an active rectifier ?

[supple pollen]

Sometimes it's easier to measure the current drawn by the motor driver, if you're looking to measure overall current. If you want to measure per-winding, it's somewhat more complicated than either an offset or rectifier. However, there are high-side current monitoring devices available such as https://www.adafruit.com/product/904 that might be useful

[dry pelican]

There are also hall effect current monitors

And also current transformers

[supple pollen]

Ah, those are better ideas

[elder peak]

..one might call them more au current

[unreal flax]

That joke is golden!

[tough matrix]

what is everyone's favorite surface-mounted slide switch?
I used this one: https://www.digikey.com/en/products/detail/c-k/PCM12SMTR/1640125
but it is slightly too small for my taste

[distant raven]

I’m a fan of this one: https://www.digikey.com/en/products/detail/AYZ0102AGRLC/1640121

[tough matrix]

seems to be almost same size, 7x3 mm

[distant raven]

Yeah, 0.5mm larger

But it is bigger so.. lol

There’s another one I’ve recommended before.. lemme find it. Do you have a size upper limit?

[tough matrix]

well, I guess anything over 20mm would be hard to fit

[elder peak]

I like this guy: https://www.digikey.com/en/products/detail/c-k/JS102011SAQN/1640095?s=N4IgTCBcDaIFIGUCMAGMKlIQQQIoDkQBdAXyA

I think that's about the biggest you can get in SMT, everything bigger is through-hole.

[dry pelican]

What is a good 5x5mm button? I found the brass ones, but I'm looking for the plastic ones that are on the circuit playground

[supple pollen]

Looking at the schematic on the learn guide, the Circuit Playground switches are marked "EVQQ". A quick web search reveals Panasonic offers an EVQQ switch line that looks like them. Here's one link https://www.newark.com/panasonic/evq-q2w03w/switch-tactile-spst-20ma-15v-pcb/dp/53W8284

[distant raven]

Similarly https://www.digikey.com/en/products/detail/panasonic-electronic-components/EVP-BV2C3L000/16277103

Granted, 6mm

Actually they’re both 6mm

I should have looked closer

I do chuckle though because I’ve had a tab open for those buttons for the last like.. 8 months

I’ve not ordered them but I have plans to eventually

[dry pelican]

Maybe I can find the same type of switch as a 5mm part

[tough matrix]

5 mm is unusual.
common size is 4.5x6mm

[distant raven]

I use these personally https://www.digikey.com/en/products/detail/R-667078/13232037

If you’re curious about how the size compares to say.. 0.1” mounting holes or 0603 LEDs

[dry pelican]

It seems you're right. The only ones I can find on Mouser are these brass buttons: https://www.mouser.com/ProductDetail/CUI-Devices/TS18-5-15-SL-260-SMT-TR?qs=tlsG%2FOw5FFh4WzqzrHrOOQ%3D%3D

[woven bluff]

I have a THT D-Sub connector with 68 pins. Would it be a good idea to cover the pins with solder paste and hot air to solder all pins at once?

[worldly schooner]

For most through hole connectors, you generally avoid this method due to the risk of melting the component you’re trying to solder. I can’t speak for your specific connector, but if you have the option to do it by hand, it might be safer that way.

You could check the data sheet for more detailed information on soldering curves for reflow, if they have any such data?

Or whatever the equivalent process for THT might be. Wave soldering…?

[distant raven]

Under the right conditions, it actually wouldn’t put much more heat than typical THT soldering. But probably some trial/error to see how quickly it would need to be done. I don’t think it would introduce any more heat than wave soldering.

[worldly schooner]

A d-sub would be a lot more tolerant to heat compared to a plastic through hole header for sure. By how much, I wouldn’t know…

[woven bluff]

the idea of hand solder 68 pin frightens me

[rustic linden]

68 pins??

Oh wow

Good luck with that

[formal sand]

68 pins ain't that bad, unless you have to do more than 1 board

If it takes you 5 seconds to do one solder joint, then it would be like 6 mins of soldering

You can use a solder melting pot to do it if don't have anything else on the other side

[upbeat zephyr]

Altium CoDesigner is a thing

[elder maple]

could someone recommend a good LTSpice tutorial?

[dry pelican]

Afrotechmods has one

[rigid plume]

Hi, would anyone recommend an alternative SOC (not module) to the ESP32-S3, that has WIFI and BLE capabilities, and is roughly the same. Thanks.

[unreal flax]

Would you consider other ESP32 variants, or do you want a different manufacturer entirely?

[icy atlas]

Currently usingthese devices with pico pi
https://www.adafruit.com/product/157
https://www.adafruit.com/product/389
but the IR sensor always has inconsistent results. When pressing a button twice it always has like one pulse duration that is waaay different from the other press. Anyone know what could be causing this?

[supple pollen]

Could be the pause between pressings

[icy atlas]

The docs on the site tell you to check for button pressing by comparing to a know pulse for the button with x% fuzziness

How should I do this when I get such varied results

[tidal elbow]

My SMD components keeps failing. The decoupling caps shorted to GND, Pull resistor for reset pin acting funny (no voltage, and resistance value changed everytime pcb is powered). Am I just need to find better components supplier? or my PCB design need another redesign?
Any advice please...

[rigid plume]

Totally different manufacturer, due to security concerns.

[supple pollen]

For the resistor, I would suggest probing the voltage on the component itself to check for good connectivity between the component and the board. If that is bad, it could be a problem with the solder joints themselves, which could be an assembly problem, a board design problem, or a mechanical problem (heat, flexing, or vibration putting stress on the joints). For the capacitor, if it's a tantalum capacitor, this is a known failure mode if the voltage has even a tiny excursion (in time or voltage) above the capacitor's rating (this sometimes happens during power-on while voltages stabilize). If it's a multilayer ceramic capacitor, this is a known failure mode, especially for lower voltage and higher capacitance ones, due to the microscopic thinness of the internal layers. I would also suggest unsoldering and inspecting the capacitor itself: after unsoldering, is it still shorted? If so, you do have a capacitor problem. If not, it could be a soldering or board problem.

[woven bluff]

if board space is limited, would it be better to put a zener at the input or output of a SMPS?

[dry pelican]

Output unless you need to protect input from transients

[woven bluff]

I always put SMPS behind a load switch

I also put a large tantalum at input, if I downgrade to 1815 or 2220 ceramic may be I can stick an input zener there.

[supple pollen]

Be wary of high capacity ceramics, they can have large voltage coefficients

[woven bluff]

I usually use more than twice the voltage rating than the actual working voltage

also I only use x7r or better

it's better to overengineer than deal with problems later

[paper vale]

Hello. I am working on a project to control a brushless motor using field-oriented control. I want to implement the speed and torque control but for that I need to generate the reference speed according to the voltage created by a potentiometer for example. What method can I use to get the reference speed according to the voltage created by the potentiometer?

I want to create the speed and Iq refs

[spice turtle]

By using an analog to digital converter, and then using a microcontroller to adjust the the speed. There are also dedicated IC's that will do the whole thing for you, but are kind of limited in current.

[supple pollen]

By reference speed, do you mean something like a reference frequency? There are voltage to frequency converter chips like the classic LM331 and others.

[paper vale]

Not frequency. I want a way to deduce the speed from the voltage outputs from pot

for example 0V to 0 RPM and the max 5V to the max 370 motor RPM

[supple pollen]

Ah, that's more of the opposite conversion, frequency to voltage

[rigid plume]

What would you recommend that is not manufactured by espressif? Thanks

[dry pelican]

There is already an simpleFOC projecy

[unreal flax]

I don't have a great recommendation offhand, since Espressif is pretty dominant in this space, though Cypress I believe has some WiFi SoCs too.

[rigid plume]

Thanks

[timid dove]

Hey folks. I’m designing a component that needs to be mated via pogo pins. I’d like to use circular contacts (rings) for mating… but I don’t know what they are called, or how to find them. I’ve attached a picture, that better explains what I’m trying to do. Any thoughts on where to find the parts? Note… I’m also thinking designing the “rings” directly on a PCB with ENGR finish.

[supple pollen]

Normally the mating contact would not be circular, just have a contact for each radius, so it will connect with the circular piece at any angle

[timid dove]

Not 100% following; you’re suggesting that I have multiple mating contacts and restrict the mating orientation options (e.g design the “dock” to be a square or triangle or hexagon shape)

[supple pollen]

No, just suggesting two contacts, and not restrict the orientation

Like slip ring contacts https://www.grandslipring.com/wp-content/uploads/slip-ring-contacts.jpg.webp

[timid dove]

Got it…. I’m struggling to find the “rings”. I can’t seem to find them though the normal channels (digikey, mouser, etc). Do you know if there is an smt (or through hole) part that exists? Or is it as simple as designing a circular contact on my pcb (and having it plated)?

Hopefully that question makes sense

[supple pollen]

Oh, I misunderstood, I thought you were trying to design a connector that mated with the rings, you want to replicate the functionality of the rings. Yes, you fabricate that as a PCB (specify "hard gold" for them if you want them to be more durable). You can also just use metal rings and attach them to whatever.

[timid dove]

Sweet! Thanks

[spice turtle]

You could read the voltage, convert it to digital, then use the SCALE function on an arduino and then output it somewhere. But generally to read speed, you need a tach

[hushed smelt]

I need that for my tank turret. I want it.

[supple pollen]

The slip ring assembly? https://www.adafruit.com/product/736

[distant raven]

Using these is a… slippery slope

https://tenor.com/view/spongebob-square-pants-spongebob-cool-finger-guns-you-got-it-gif-3561703

[hushed smelt]

Found some micro slip rings. That should do nicely for adding the pan/tilt mechanism on top of the turret. 360 turret and pan/tilt camera.

[tough matrix]

check mill-max catalog
https://www.mill-max.com/

https://www.mill-max.com/catalog#

in particular, "target connectors/target disks"

most of them are available on digikey

[tidal elbow]

Noted. I found no issue with the solder work, I'll check the caps value next. TBH this issue fixed when the failed components is replaced. But it'll happened again after 1-2 day 😤😩. I source the components from local online market place, that's why I feel like finding better reliable components supplier.

[dry pelican]

Maybe a voltage or overload issue

What is the circuit?

[twilit crane]

Hey folks - using:

• adafruit 64x64 matrix panel
• adafruit rgb matrix hat + rtc for rasp pi
• raspberry pi 4 (running rasp pi Debian 11)

I’ve assembled everything per the instructions and got it connected and running some of the included demos, but am running into an issue where the images are distorted or text just won’t show up (e.g running example-text demo)

Any ideas what might be causing this?

[dry pelican]

Loose connections?

[twilit crane]

The cable to it seems to be well seated; or do you mean the soldering?

Could this be due to power shortage or the lack of clock battery?

[dry pelican]

Not lack of clock battery. But power supply issues can cause weird effects

[hushed smelt]

Attempting to run it from the microcontroller board power only will produce only red LED’s because it doesn’t have enough current for green or blue. Unless your code is specifically only red in color it means you’re not getting external power to the matrix properly. Ask me how I know. 😅

All red led’s instantly equals no external power to me.

[twilit crane]

Thank you! I was wondering about the all red on the default demo

Looks like my power supply is not good enough; will need to order one

[sick geyser]

Pico W or just the wifi chip it uses Infineon CYW43439

[woven bluff]

anyone worked with GM counter? how small is the current transient when a particle hits the GM tube?

I'm thinking about instrumentation amp connected as transimpedance at low side of the GM tube

[tidal elbow]

Pull up resistor for I2C, reset pin resistor, and vcaps capacitor was busted

[dry pelican]

That is very weird and should not happen. Especially those resistors. They're barely passing any current. The resistors short after a few days right? And this is powered on or powered off?

[distant raven]

This bothers me..

I don’t necessarily think this is a good idea

I’d remove the R206 and the LED

[woven bluff]

if you want indicator, better use a driver, BJT, FET, etc

[distant raven]

Yeah, I worry if the pull up is failing that the 100 ohm resistor is going to over draw the pin and do damage.

You’re also making a voltage divider on the data line which is also not good.

[tidal elbow]

You mean the resistor on the top of the optocoupler?

[distant raven]

No the picture

The i2c SDA line

[tidal elbow]

The R205 is a pull up resistor. This is 4 layer PCB. While the R206 is bothersome, I'll try to DNP the indicator all together tomorrow.

Thanks for the suggestion 😊

[tidal elbow]

After 2 days of on and off. It's not running continuesly for 2 days

[supple pollen]

It's pretty small, even after the electron multiplication takes place, you're taking about thousands of electrons, which is not much current at all

[woven bluff]

how does the circuit usually work? a general purpose amp on the low side followed by a comparator?

[timid dove]

thanks for the suggestions. Unfortunately, they don't have the exact shape that I'm looking for. I'm specifically looking for two nested rings (one inside of the other). Gonna follow madbodger's advice and make the ring on the PCB (ideally with Hard Gold - which doesn't seem to be an option for most cheap/prototyping PCB manufacturers).

[hushed smelt]

maybe electroplate it but that seems like a lot of specialized setup for 2 rings.

[supple pollen]

Usually a capacitor to couple the pulse into some sort of amplifier/clipper/limiter

[tough matrix]

PCBWAY has hard gold option

but it is quite expensive

[formal sand]

What about using solder above the copper basically like tinning the whole circle? There might be issues with oxidation of the solder but might be a cheaper solution

[cinder grail]

https://www.adafruit.com/product/1196 more wires

[idle python]

before i buy the pcb and components, can anyone double check if this will work as intended? i dont want to waste 62 dollars on a possibly not working board

min clearance and trace width is 5mil

[native plover]

I want to pulse a solenoid with a Pico. I want to wire it like this and use twelve AA cells (either alkaline or NiMH) to power it. Does that sound reasonable?

I'm thinking that this might draw too much current from the cells or that the power might get a bit unstable for the Pico

Would it make sense to use twice the amount of cells to drop the voltage from higher up? That should reduce the current through the cells.

[dry pelican]

Use a mosfet instead of a bjt

It would probably work

But the current is really high

3.7A

Really high for what I expect from a solenoid

Power instability could be solved with a lot of capacitors

But idk if aa batteries can handle 3.7A

[native plover]

Yeah, it's a fairly chunky solenoid 😉

I tried smaller ones, but they lack the punch I need

Would it suffice to put one large capacitor on the Pico's input?

I was hoping that the relation between Vin/Vout and Iin/Iout on a buck converter is somewhat anti-proportional, i.e. the higher I raise the input voltage, the lower the current that has to go in

[supple pollen]

I'd probably put both a capacitor and a decoupling diode so the solenoid can't pull power from the Pico's capacitor

[native plover]

Where would I put that? Before the 3.3V buck converter, and then the capacitor behind it?

[vast flume]

Hey guys

I think I made (another) oopsie on (another) one of my pcbs, but I think this one is easily fixable

So I have this setup for a buck converter

But I forgot that the input voltage is ~13v, and since I'm feeding that directly into the enable pin (through a resistor to limit the current), I'm still sending around 7v to that pin

But in the datasheet it says the max voltage for that pin is 6v

I tested the pcb and the converter doesn't power on. Wouldn't be supprised if I fried tf out of it when I first plugged this in. I have another one though so I was wondering if anyone had a possible solution in mind?

Could I just use a resistor with a higher value?

[spice turtle]

The datasheet shows a 100k resistor on the enable pin. It looks like you got that

It also shows a voltage of 4.75V to 23V, so your input of 13V is fine

If you have a scope, check the SW pin for a waveform.

Otherwise, id look for any shorted pins. Unless you got the package wrong, I dont see anything wrong with the above schematic

[vast flume]

Hmmm good point

Sadly I don't have a scope with me

No short on the board though, put a continuity tester across the + and - terminals and didn't get anything

[spice turtle]

make sure your pins are actually on the board. With SOIC its easy. Sometimes a pin Looks like it connected but its not

[vast flume]

Honestly that could be it

All the components on this board I pulled off of the previous version of this board

[spice turtle]

You could also just hit it with a air gun to make sure.

[vast flume]

And I didn't add much solder

Yea I'm gonna add more solder

[spice turtle]

not too much though

[vast flume]

Yeah 🤣

[spice turtle]

I have just hit the pins with a hot iron and a tiny amount of solder and it was ok

[vast flume]

Heres how it looks rn

[spice turtle]

is it IC1?

[vast flume]

Yeah

[spice turtle]

I can already tell by looking at it you have unsoldered pins

and Pin 3 (Im assuming its pin 3) needs solder

pins 1 and 2 look iffy at best

Id touch up all the pins

[vast flume]

Laziness strikes again lol

Just added some more solder will report back in a min

We have light 🎊

That fixed it 🙂

Now I just have to figure out whats up with the 3.3v bus lol

[spice turtle]

sweet

[spice turtle]

probably the same. If its a regulator it may not be connected. Those large pads take A LOT of heat

I have to put my iron on its side a tad to get full contact.

[vast flume]

I'm actually getting a short now

My first suspect is U4 bc the pads are inside

[spice turtle]

between 3.3V and gnd?

[vast flume]

Yea

U4 wasn't it

Ok this just got confusing. I cleared the short, but now the inductor in the 5v buck circuit is getting really hot each time I connect it, and the 3.3v linear regulator is just outputting ~0.05v

But theres no longer a short between 3.3v and gnd

[spice turtle]

theres something pulling a lot of current.

Whats your 5V at?

I would disconnect the 3.3V regulator to make sure your 5V line is ok

[vast flume]

5v is at 5.1v

I don't think c17 not being properly connected would cause this right?

[spice turtle]

its more or less for filtering. Some regulators dont like it but you'll know right awya

[vast flume]

Thats true

Idk what to even check at this point

The microcontroller is soldered perfectly

And I don't think u8 and u4 were the cause

But I have no idea what else could cause this bc everything else is so straightforward

[spice turtle]

the regulator might be bad if you are using used parts

[vast flume]

The 3.3v one?

[spice turtle]

yea

[vast flume]

hmm

Is there a way to test it?

[spice turtle]

desolder it and connect wires to it pretty much.

[vast flume]

Ah

Yea I'll just try swapping it

[spice turtle]

basically dead bug it

[vast flume]

Fresh ones on now

No luck 😦

[spice turtle]

you have an issue down stream of the 3.3v taking up a lot of current then

[vast flume]

Yeah

[spice turtle]

If you have the 3.3V removed does the inductor still heat up?

[vast flume]

Good question

One sec

[spice turtle]

it shouldnt heat up with no load unless its a really really awful inductor

[vast flume]

Strangely it does...

Uh oh

The confusing thing is I use this same inductor for the regulation circuit on a different pcb and it works well

But then again this is 14v-5v and the other circuit was stepping up to 5v

[spice turtle]

your using a 10uh inductor, right? Rated for 4A?

[spice turtle]

check C15 and C16 (the output caps)

a scope would really come in handy right now lol

[vast flume]

Yeah that would help a ton 🤣

How should I check them?

Just see their voltage?

Or make sure they're connected

[spice turtle]

see if they are connected

or hit them with some solder

but..hmmm

[vast flume]

I tried both 😭

[dry pelican]

Remove buck and apply power using a variable supply?

And see what current is

[vast flume]

I don't have a variable supply 😦

[dry pelican]

Current is the important part

[vast flume]

Yeah

[dry pelican]

Multimeter can also

[vast flume]

Oh yeah true

One min

[dry pelican]

Do you know how much current the buck uses unloaded as well?

Do any caps heat up?

[vast flume]

Not sure

[vast flume]

I'll check

[dry pelican]

And you may be able to also use your multimeter to measure AC current through tge inductor

Well since it's heating, it's obviously drawing too much current

And what buck IC?

[vast flume]

0.25 amps is the current draw that I just measured

[vast flume]

MP2307

Also just checked and at least one of the caps is heating up

C15

[dry pelican]

Ooh

That might be a lead

[vast flume]

Yeah

[dry pelican]

What is the value of C15?

The solution in this case might be moar caps

[vast flume]

22uF

[dry pelican]

Add some smaller ceramics in parallel

Like 0.1uF

Larger values usually have more esr and esl

[vast flume]

Ok

in parallel with the current one?

Added a 0.1uF in parallel to both c15 and c16 (the output caps)

Still gets incredibly hot 😦

[dry pelican]

That's weird

Look at datasheet and see if you are exceeding anything?

[supple pollen]

I don't have the circuit in front of me, but as a guess, supply to diode, then capacitor, then buck converter

[formal sand]

Have you tried desoldering the cap?

[native plover]

I see, thanks!

[balmy tide]

Can i use cat6 ethernet cable to transfer mipi signals? I am using MIPI retimer IC from TI and I need to move and swap some pairs in in between devices/connectors. Mipi requires 100 ohm differential impedance which cat6 cable has and with cat 6 I believe I can swap how rj45 jack wired to swap pairs around without needing to use vias on the board

[cursive sentinel]

Should be able to, although it's difficult to say how far you can run it.

[balmy tide]

Retimer thing tested with 30cm on fr4 board as long as I can match that it's golden

[neat raft]

Hello, I have built a circuit with the QT PY and wanted to see how I could replace the dev board with an RP2040 on a PCB to replace the MCP23017s with the RP2040s GPIO pins. Using the "Hardware design with RP2040" I have built the minimal circuit required and hooked up the GPIO pins to the rest of my circuit.

Long story short: I have seen that the GPIO pins were connected to header pins as well as being connected to ground for a general purpose dev board design. Should they be connected to ground in addition to my other circuitry which mainly consists of a 32 pin header for game cartridges?

[supple pollen]

I'm unsure why you would ever hardwire GPIO pins to ground. It seems like asking for trouble.

[exotic oak]

I'm wondering if with the circuit playground express, is it safe to have it plugged into USB at the same time a AAA (alkaline) battery pack is attached or if I need to make sure to always turn off the battery pack first so it doesn't try to charge it?

[neat raft]

It also struck me as a bit weird. btw I am referring to this particular part

[supple pollen]

Ah, that has them going to two-row headers, where one row is the grounds, and the other row is the signals, so you can pick up a signal and ground from two adjacent pins.

[supple pollen]

Yes, it's safe. The Circuit Playground Express does not include a battery charging circuit, just steering diodes, so it can be powered from either USB or the battery connector, without backfeeding power from one to the other.

[neat raft]

Thank you for clearing that up. Now it makes much more sense.

[exotic oak]

Ahh thanks. I had been using feathers that had charging and had just assumed the express did too apparently. Oops.

[rustic linden]

i thought that was the case

[broken zenith]

Is there a simple way to toggle a relay's state (on to off, off to on) every 200ms? (I have a very specific need for a relay and not a transistor of any kind) I plan to use the EE2-24NUH if that makes a difference. I'm looking for something involving cheap timer ICs or an RC circuit, so no microcontrollers ideally. Thanks!

[unique patio]

NE555 timer is classic and can sink up to 200mA at up to 15V. But the EE2-24NUH uses 24V coil voltage. Can you pick a relay with a lower coil voltage?

[broken zenith]

Oh, I didn't realize that. I'm just looking for a physically small SMD signal relay, can you recommend any? Thanks for the help

[unique patio]

what voltage and current are you switching? What supply voltage is available/convenient?

[broken zenith]

This is gonna sound crazy but the voltage and current switching capacity don't matter. 3.3v is the planned supply voltage, but I could increase that if need be.

[unique patio]

don't matter, do you mean you just want a click noise?

are you not switching any voltage? Is it an input, like a touch input or something?

[broken zenith]

The click and the tactile feeling, essentially. This is the E-Fidget Lite, so I'm just using the relay for haptic feedback.

[unique patio]

ne555 is 4.5V minimum. It could drive any small relay. But you want to pick a relay based on clickiness, then. Did you try the relay in question already?

[broken zenith]

Nope, but that might be a good idea. I've got a relay with a good click to it, let me check what it is

[unique patio]

you could use a latching relay and reduce the power consumption

https://www.adafruit.com/product/1201

[not a relay, but might feel satisfying with a pulse]

[broken zenith]

Hmm, I hadn't thought of that! thanks. The relay I've got is a SRD-05VDC-SL-C.

I used vibration motors in the last version, and they were pretty good, but I want to try something new.

[unique patio]

more modern 555-style ic: https://www.adafruit.com/product/5222

digikey sells 1000's of small relays, buy a few maybe and see how they work otu

[broken zenith]

As for the 555 - What's a good resource for learning how to get that to do what I want? I saw the Adafruit learn guide but it seems like more of an overview than an in-depth tutorial.

[unique patio]

The 555 is so common that there are thousands of tutorials. Books with examples include Make:Electronics and the Art of Electronics

this seems pretty thorough: https://learn.adafruit.com/getting-to-know-the-555

is that the one you mean?

[broken zenith]

Yes, but I guess I thought wrong! Taking a second look it seems pretty good. I would want it in astable mode right?

[unique patio]

yes

[broken zenith]

Alright, I'll do some reading. Thanks for the help!

[unique patio]

yw!

[raw thorn]

Relays can be wired as a buzzer & timing altered with capacitors & resistors.

[idle python]

can anyone help me with this pls?

[distant raven]

It would be easier if you shared your files zipped up so everything can be evaluated

[dry pelican]

I would say it looks fine. A good tip for final verification is to print it out (only soldermask and silkscreen) on a paper printer to see how it will look irl

[left citrus]

hey guys i want to interface a 1.8V SPI Device (ICM20948) and use a level shifter with my MCU (3.3V) : commonly BSS138 is used, but due to footprint restrictions i want to find alternative; i believe RE1L002SN might be the right equivalent, can anyone help me out to confirm ?

[supple pollen]

BSS138 is normally used for I2C style communication, not SPI. You'll want a different kind of level shifter.

There are a bunch of choices out there, such as the SN74AUP2G34, which includes two level converters and is available in an SC70-6 package (two of them would handle SCLK, MISO, MOSI, and CS)

[left citrus]

i assumed it would BSS138 as it has been used on Adafruit ICM20948 module:
https://learn.adafruit.com/assets/93864

[supple pollen]

It looks like that chip supports both SPI and I2C. The AdaFruit board appears to use it in I2C mode. Since you specified SPI, I figured you wanted to use it in SPI mode.

[left citrus]

The adafruit board does support SPI, but i think i'll stick to dedicated level shifters, thanks alot

[cursive sentinel]

The BSS138 shifter is usable for I2C without hurting performance too much (as long as you stay below 400kHz) since it is an open drain interface. SPI is entirely push-pull and typically run at significantly higher clock rates than I2C and no bidirectional signals, so you really do want an IC level translator.

[left citrus]

I figured NTS0104 might be the IC with 4 channels

on the datasheet it says: Push-pull: 50 Mbps

[cursive sentinel]

That should work assuming it supports the voltages you need.

Although keep in mind that one of your channels needs to go in the opposite direction of the others.

[left citrus]

i have no clue what mosfet sparkfun is using, but thier breakout works on my breadboard prototype

[cursive sentinel]

Pretty sure it's the BSS138

[supple pollen]

It's a lot better than the popular 2N7000 for 3.3V logic, and acceptable even for 1.8V logic

[cursive sentinel]

The 2N7000 is unsuitable for 3.3V logic.

[supple pollen]

I agree, but it's commonly used as if it were suitable, unfortunately

[woven bluff]

I often avoid old parts... like xNxxxx

[grand fossil]

I think this is the right place to ask -- for Eagle if I wanted to place a 2.0mm 4 pin JST connector (like the stemma ones) what should I look under for the part? I was able to find the 3 pin ones in the adafruit eagle libraries but having trouble finding the 4. My first time with eagle so a bit of a novice.

[hushed smelt]

The smaller ones are JST-SH and the bigger 2mm ones are JST-PH

@grand fossil https://www.adafruit.com/product/4390

[grand fossil]

I mean for the eagle design component placement for PCB design -- trying my hand at a custom board. Was able to find the 2 pin and 3 pin ones there but I may be missing something with the 4.

[hushed smelt]

Are you asking for a footprint?

[grand fossil]

Yeah -- wasn't sure if there was a specific component I should use when placing a part for a jst 4 pin 2mm.

[hushed smelt]

The product page has a link to the manufactures website where the blueprints are for creating your own footprint. If you want something premade that you can drop in I recommend finding a product that already uses it, go to that Adafruit's product page, find the downloads for .brd and .sch and snag it from there.

[grand fossil]

Sounds good, will give that a go, thank you.

[hushed smelt]

Do you know of an Adafruit product that uses the connector you want to use?

[grand fossil]

Not sure offhand but will look around.

Thanks for the assist.

[hushed smelt]

The .brd and .sch are usually in the learn guide for a product. So you'll want to go to the product page, scroll to the bottom, find the learn guide associated with it, find the downloads link in the learn guide.

stemma is JST-PH, stemma qt is JST-SH.

@grand fossil The stemma soil sensor has one. https://github.com/adafruit/Adafruit-STEMMA-Soil-Sensor-PCB

It uses a JST-PH connector. How you'll separate the component in the .brd and .sch file and get them into your project is up to you. 👍 I don't use Eagle so I can't offer more specific advice.

[cursive sentinel]

don't use eagle
You included the advice in saying you can't offer advice 🤣

[grand fossil]

Sounds good, it's good advice for the future re: digging into the product page there for the brd -- will definitely do that. Was able to find a library with a 2mm one so should be set.

[compact frigate]

Which pins of the RP2040 can sdioio use? Are some pin combinations (including their order) slower than others or even impossible? Or might sdcardio be faster than 4-pin sdioio on the RP2040?

[cursive sentinel]

Generally speaking on any microcontroller there won't be pins that are slower than others unless you are doing something you shouldn't (such as remapping the reset pin as a GPIO).

[compact frigate]

Often SPI is faster than bitbang, but requires specific pin combinations. So if sdioio is faster than sdcardio with SPI, then maybe sdioio also does something special under the hood and requires specific pins for reaching maximum speed.

[cursive sentinel]

Those are both hardware peripherals, so neither would rely on bit banging. I'd have to look at the datasheet for specifics which I can't do at the moment.

[opal rune]

I believe it's because those are the QSPI pins that the flash memory uses, so I don't think it's usable for sdioio, unless they're free. Though according to https://learn.sparkfun.com/tutorials/rp2040-thing-plus-hookup-guide/all and the datasheet, the pins are:
Pin 52: CLK
Pin 56: CS
Pin 53: DATA 0/DI
Pin 55: DATA 1/DO
Pin 54: DATA 2/WP
Pin 51: DATA 3/HOLD

[grand fossil]

(Just deleted and reposted after fixing an issue I noticed, apologies).
Would it be possible if someone has a chance to get a second set of eyes on my pcb I was hoping to have printed out? No rush of course or expectation. I'm relatively new with electronics and pcb design in general so in case I've done something terrible.

An attempt to make my dumb RGB mouse bungee smarter.

Elements:

• Emulates the form factor of the existing board such that it'll still fit (I had to extend the top a bit so the WS2812B-mini would still clear the weights which sit under the board on the bungee
• Add the ability to extend the bungee with i2c in the future to support additional sensors
• Have the ability to control additional light strips in the future from the same d1 mini (small light bars, nothing large given the current needs)
• Support the existing button 6x6x15mm from the bungee because the cutout in the frame exists for it anyhow

Considerations taken:

• Each light strip and the light square have a 1000 uF capacitor across their VCC to GND prior to their connections to smooth out the voltage from spikes and associated noise
• The LED data lines also have a 470 Ohm resistor
• There's a 10k Ohm pull up resistor for the button
• The button also has a 1uF debounce capacitor
• Have mosfets setup for the stemma sda and scl lines and a 4.7k Ohm pull up resistor for the 3.3v to 5v signal lines

Apologies for the rough nature of this - was my first eagle project so still learning. I haven't sorted out how to make it look nice with all the different sections yet. I did sort out the way to map up the different net lines though so it shouldn't be too hard to parse.

Main elements are:

• On the right I have the WS2812B-minis
• Connected to those other two GPIOs on the right (located above and below on the schematic) I have the JST connectors setup for those LED strips
• On the bottom left I have the stemma connector setup for i2c
• Button on left
• JST 2 pin connector for power top left

[supple pollen]

You may have the level shifter MOSFETs in backwards, depending on what the expected voltage of the external I2C bus is

[unreal flax]

The capacitor on the button looks problematic, because it's blocking the path to GND instead of being in parallel with the signal. Also your MCU would probably appreciate a local small decoupling capacitor on its power pin.

[grand fossil]

Thank you for taking a look. Will take a look into those issues.

[grand fossil]

Apologies, could someone take a once over of my logic here? I'm going to order some mosfets to try things out with a proto board prior to ordering a custom pcb just want to make sure I understand the concept fully.

The long line at the bottom there is the gate and then the other goes from source to drain (more speaking out loud in case my thinking or understanding is off here), the 4.7k ohm resistor is a pull up resistor on the MCU's pin keeping it in the high state normally (as needed for i2c), the gnd here is the source, drain is the SCL load. The MCU's pin's signal acting as the gate in this case causes the 3.3v signal to be emulated by the 5v line (high when high, low shutting off the flow when low) so I think my logic is sound with my latest change here attached. Is that correct?

Really appreciate the assistance.

[grand fossil]

Adjusted those other issues, ty. I had read about adding the button debounce when looking into potential gotchas I should address, added the capacitor, and completely blanked putting it in series vs in parallel.

[unreal flax]

The button cap is still a bit weird, since you have both sides of it shorted together to ground. You want it on the other side of the switch, like this sort of arrangement:

[grand fossil]

Ah I see, that makes sense. So the button debounce is when the button is depressed it doesn't send additional input signals during that period of time when things are returning to the normal state. When the button is pressed it becomes low but once it comes up it's high again. Without the capacitor as the mechanical aspects move it would flow to low potentially for a brief moment but by having the capacitor it ensures once the button is depressed it immediately reads high again as it smooths the voltage avoiding those random fluctuations. Is that right? That's why it's on that side as if it was on the other side it wouldn't do anything in this case?

[unreal flax]

Yeah, the capacitor "holds" the voltage steady for a short time even if the switch is bouncing open and closed, so you get a smoother, slower transition which looks like only one low-high or high-low change to the input pin.

[grand fossil]

Awesome, ty. Makes sense. Will give this a try with a breadboard once I get the parts and give it a go. Really grateful for the assist.

[cursive sentinel]

In the datasheet I'm not seeing any interfaces dedicated to SDIO, unless it is implemented with PIO. Where are you seeing info about it?

[woven bluff]

hardware debounce is costy

you need RC filter and Schmitt trigger

[waxen halo]

Hi I'm taking my first attempt at creating a DC-DC power supply and I'm really struggling trying to be sure of my mosfet selection and what the thermal limits are. I'm using the MIC2169BYMME synchronous step down switching regulator. The datasheet states:

For applications where Vin < 5V, the internal Vdd regulator operates in dropout mode, and it is neccessary that the power MOSFETs used are sub-logic level and are in full conduction mode for Vgs of 2.5V. For applications when Vin > 5V; logic-level MOSFETs, whose operation is specified at Vgs = 4.5V must be used.
I'm struggling to interpret this. Does that mean that for my 12Vin, Vgs th must between 4.5V and 5V? Or just that it's at-most 4.5V?

[woven bluff]

the chip operates at 5V, if you want to minimize heat, you'll need a FET with Vgs(Rds->min) < 5V

[waxen halo]

As in it doesn't matter what the upper limit of Vgs is, just that I need to judge it by the specifications for Vgs being 4.5V? So like if a mosfet says something like:
Rds(on) max:
0.0087 at Vgs = 10V
0.0135 at Vgs = 4.5V
I might want to look for something else?

I think I'd been thinking I needed to be thinking of these mosfets at 12V, but that seems clearly wrong now. Vds would be 12V, Vgs would be 5V.

[compact frigate]

Thank you for looking into this. https://datasheets.raspberrypi.com/rp2040/hardware-design-with-rp2040.pdf says "The constraints our SD PIO software puts upon us is that the four data signals must be connected to contiguous GPIOs. The CLK and CMD signals can go anywhere." Does sdioio use PIO with the same constraints? Would some pin combinations (within these constraints, or beyond them by changing the PIO implementation) work faster than others?

[woven bluff]

https://github.com/raspberrypi/pico-extras/tree/master/src/rp2_common/pico_sd_card

[cursive sentinel]

The reason that constraint exists is likely due to the fact that they are using side-set, no pins will be noticeably slower.

Regarding sdcardio and sdioio: those are Adafruit libraries. It looks like sdcardio uses SPI while sdioio uses SDIO.

SDIO will be faster.

[compact frigate]

Does sdioio also use PIO and specifically PIO "side set" on the RP2040? If so, does it have the same constraint (contiguous GPIOs)? If not, PIO "side set" wouldn't work noticeably faster, right?

If sdioio requires using contiguous GPIOs on the RP2040, can their order be inverted, for example GPIO 22,21,20,19 instead of 19,20,21,22?

[cursive sentinel]

You could probably change the order by editing the PIO assembly, but side-set always requires contiguous pins. I'm not familiar with Adafruit's implementation, but as a general rule SDIO is the 'native' interface for SD cards. SPI is really just the 'fallback' interface. It's slow in comparison, but everything can support it.

[compact frigate]

So sdioio might be implemented with PIO or run on the main processor, we don't know, right?

[cursive sentinel]

Ok let me find the code...

[compact frigate]

Regarding "side-set always requires contiguous pins": can their order be inverted, for example GPIO 22,21,20,19 instead of 19,20,21,22, when assigning the four data lines of the SD card?

Ah, that's what you wrote, sorry

[cursive sentinel]

You could probably change the order by editing the PIO assembly, but side-set always requires contiguous pins.
^ Yes, but only if you edit the PIO code. I'm still trying to figure out what implementation they are using. CircuitPython has a lot of abstraction to navigate.

[compact frigate]

Thank you for looking into this! 🙂

[cursive sentinel]

Looks like they haven't implemented it for the RP2040 yet.
https://github.com/adafruit/circuitpython/issues/7671

[woven bluff]

thought on this thing?https://flipperzero.one/

looks useful but a bit pricey

[twin hedge]

I'm making a board to use a ESP32-C3-WROOM-02-N4, but I've never soldered surface-mounted components. Where should I start?

[twin hedge]

I have one. I'm not exactly sure how to explain my thoughts on it, but I can say that it is fun to have, and it can actually be useful.

[unreal flax]

You might consider getting it professionally assembled by the PCB house.

[twin hedge]

I'll probably do that. I'd just mess up the build anyways.

[twin hedge]

I could use some help with parts. I'll attach the photos of my hand-drawn schematics. My LDO is a LD1117A. Should the resistors be 1/4W resistors? Also, what type of capacitors should I use?

[cursive sentinel]

Those resistors are used for feedback, so they'll generally draw a couple mA at most. For caps I would recommend a mixture of electrolytic for the larger ones and ceramic for the smaller.

[twin hedge]

Alright. What kind of switches should I use for for the boot pin and the reset pin?

[cursive sentinel]

Push buttons with RC filters to debounce should be alright.

[twin hedge]

I've seen KMR2 switches used in other circuits for ESP32 modules. In the schematic for the ESP32-S2 and ESP32-S3 Feather boards, an RC filter is only used on the reset button. I'll use the same resistor and capacitor values in my schematic. However, I'm wondering if I can just use any KMR2 switch, or if I have to choose based on other factors. Also, would the capacitor in the RC filter be ceramic?

[woven bluff]

without Schmitt trigger, you'll need a big cap, bounce can be quite long

[unique patio]

I don't think the cap is for debouncing, but is to prevent stray resets from noise, etc. See 2.2.2 Reset in https://www.espressif.com/sites/default/files/documentation/esp32-s2_hardware_design_guidelines_en.pdf

[twin hedge]

That's what I was thinking. I've noticed that I can only find switches for 32V@0.5A, so do I just use one of those?

[idle python]

was gone for a few days

[distant raven]

can you include a pdf of your schematic?

[tough matrix]

what is a good MCU for most basic projects (blinking some LEDs)?
requirement:
cheap
Arduino compatible
easy to hand solder

I used to use attiny85.
is there something better nowadays? or is it still the best choice?

[cursive sentinel]

The "best" choice is whatever meets your requirements.

[elder peak]

I think there's a few of the new-series AVR in DIP format.

But, dono, I think everything's moved towards that which is hard to hand solder.

[unique patio]

those are the maximum ratings, so they are perfectly fine.

[tough matrix]

I am fine soldering SOIC packages, but not TQFP

[distant raven]

Samd11 has Arduino support

16kB flash and 4kB SRAM

[elder peak]

Is there a SAMD11 core that's maintained?

[distant raven]

Mattair compiles

[elder peak]

There's some STM32 parts that are TSSOP-20 or SOIC-8 that are covered by the STM32Duino core.

And there's the new-series of AVR's

[tough matrix]

sure, but they are unlikely to be cheap

[distant raven]

Samd11 is $1.65 for a single

Like 4 other components for basic ops

[elder peak]

Yah, the TSSOP-20 STM32's are about the same.

[distant raven]

https://www.digikey.com/en/products/detail/ATSAMD11C14A-SSUT/5226470

Not terribly many IO though

Just 12 in that package

[elder peak]

Which reminds me, I've got a STM32L041F6P7 that I should really actually screw with.

Like, I found this guy: http://www.technoblogy.com/ and he's been doing a lot with the new-series AVR for cute lil cheap projects

[distant raven]

I made my samd11 blink from Microchip studio lol

That’s about as far as I got with it. I got distracted trying to fix a logic analyzer Arduino sketch

[elder peak]

Well, it still means that your effort was flashier than mine.

[twin hedge]

I made a mistake. It's 0.05A, not 0.5A. Would it still work?

[unique patio]

Yes, the reset current is quite low

[formal sand]

I'd go with the STM32 families. The G0 and L0 are cheap and easy to get.

Also the tooling around the STM32 stuff is much better than what arduino can offer if you are thinking into moving into more serious work.

[tough matrix]

no, this is toy project to teach kids

for serious work I would indeed use stm32 or esp32

[dry pelican]

ATTiny probably meets those requirements pretty well

They don't make any new DIP uC designs (not 100% sure), so the older ones are definitely some of your only options

castellated modules are also hand-solderable if your kids are up for a challenge

so an ESP32-S2 module maybe

it is circuitpython compatible and native USB, so that's good

[formal sand]

Yeah I feel something like an ATtiny would be good

I have some DIP attiny25s those were neat

[woven bluff]

lol, kids need to start with STM32, not obsolete tech

[tough matrix]

I feel it is a waste to use stm32 to blink 3 LEDs :)

[formal sand]

There are some that are meant for applications like that, usually the super cheap ones with very little ram and flash

[woven bluff]

make them do native SDIO, and HID 🙂

[supple pollen]

Some of the MSP430 chips come close. Normally they use in-circuit programming which is easy to wire but needs a programmer (not expensive, but it is another part). However there is the MSPBoot project that offers a bootloader. I don't know if there's direct Arduino hardware support, but Energia is available, which is an Arduino port for the MSP parts.

[woven bluff]

also old MCU require 5V rail

[tough matrix]

not attiny. those can work from 3.3

[supple pollen]

Here's a quick test I did a while back. Programming one using their "Spi-By-Wire" protocol takes 4 wires, two of which are power and ground. The inexpensive Launchpad boards include a programmer, including jumpers you can use like this to program offboard chips instead of physically moving chips into the socket to program them.

[flat vigil]

I'm interested in the G0 too. It is programmable over I2C. JLC has some for 0.52 in singles

[formal sand]

Wow that's really cheap

[flat vigil]

I've made a stemma qt board with it 🙂

the C0 is supposedly cheaper too

[formal sand]

Oh yeah the C0 is the new one they released, it is very interesting

[flat vigil]

ya, my guess is that it'll be cheaper in the long run but now it isn't very available

[formal sand]

Yeah they just released them not too long ago

Maybe it needs more time for them to build up stock

[inland jungle]

are there any advantages/disadvantages to getting rid of the thermal breaks between copper pours and SMD pads? Since I'm using solder paste and a hotplate, it's easy enough to solder without the thermal breaks, so is there another reason to keep them in?

[cursive sentinel]

Well, better heat transfer for the thermal pads definitely. Otherwise I don't think there will be anything noticeable.

[bold basin]

Will the "Adafruit LiIon or LiPoly Charger BFF Add-On for QT Py"(https://www.adafruit.com/product/5397) work with a regular Xaio(https://www.seeedstudio.com/XIAO-ESP32S3-Sense-p-5639.html) board? The QT py boards show A0-A3 which would imply Analog ports whereas the Xaio boards have them marked as D0-D3 which would imply Digital input. I'm guessing the charger board only really interacts with 5v and ground by default, but I wanted to make sure.

[pale umbra]

are you allowed to sell electronics after you pcb them / 3d print them or everything sold require license 3d printer / extra charge when making the pcb / CE or FCC markings etc ?

[dry pelican]

Ch554 or ch552 seems like a good one

[dry pelican]

Should work

[dry pelican]

Yes. Probably. Eagle's free license is non commercial, but FOSS programs like KiCAD allow you to sell what you make. JLC or any other manufacturers don't care if you sell or use for personal

And CE and FCC are only if you want to make a real product that needs to be compliant. Otherwise I think it just needs to be part 15 compliant which is automatically free. (I'm not sure. Not a lawyer)

[dry pelican]

The xiao's d0-d3 pins are also adc pins

[tough matrix]

thanks for the suggestion! do you have a link for programming them over i2c?

[supple pollen]

You'll need to get FCC certification before you can put FCC markings on your product.

[pale umbra]

I guess that's even outside the money of adafruit

since they only asked for it once

[supple pollen]

Who asked for what once?

[pale umbra]

adafruit once asked for and got a FCC certification for a shield

https://fccid.io/S6OBLUEFRUIT

for some reason 4 of the fillings including the schematics are confidential despite adafruit normally publishing schematics/eagle files but this might have been before they became open

[dry pelican]

The filings are probably confidential. But adafruit always releases schems

They were always open

Sparkfun got fcc cert for artemis module

[tough matrix]

you are certainly allowed to sell electronics - just look at tindie.com! or check the work of Unexpected Maker.
Sure, if your design is based on someone else's work (e. g. Adafruit), you need to check the license of the original work.
FCC certification is out of reach of almost all hobbyists, but you rarely need it - most of us, when we need wifi/Bluetooth/LoRa connectivity, use already certified boards or modules

depending on your local laws, if you generate enough revenue, you may need to register as a business - @distant raven can share his experience

[cursive sentinel]

Yeah you'd only need the certification if you were designing the RF board. Using an already-certified wireless module is fine assuming you don't modify it.

Although selling it in Europe might be more difficult

[tough matrix]

I sell some of my electronics, but my sale numbers are so low that I never bothered to make any official registration (do not tell this to IRS!)

[cursive sentinel]

As long as it's under like $10k/year I don't think the IRS cares.

[distant raven]

Though your state department of revenue might for unpaid sales tax

[cursive sentinel]

No sales tax in Oregon (although that's not necessarily a good thing).

[supple pollen]

The USB wiring looks dodgy to me. It looks like there's no USB connector, and the USB data pins on the LTE module and the USB/SPI module are connected together. Normally each would go to a USB connector.

[worldly schooner]

Ch55x are dirt cheap and soic, and I’m pretty sure someone has a working arduino core for it.

[idle python]

the usb connector is just 2 pin holes

but on the pcb the traces from ft220x go directly to the lte module

and the usb connector gets usb traces close to where the usb points on the ft220x are

the blue pads where the traces go is the lte module

[supple pollen]

I'm not sure that's what you want to do.

[idle python]

then what do i do

i still want access to the usb pins in case the spi pins dont work for whatever reason

[worldly schooner]

Even though it’s not super high speed signaling, it’s usually good practice to have length matching and impedance matching for usb, as it is a differential signal. Also, you probably want separate pins for each USB, as multiple usb drivers can damage each other if not interfere.

Ideally, you give them some space to have their own micro-b connector.

Or wire them to a usb hub/switch.

[idle python]

the length mismatch is not more than 3mm and the impedance is exactly 90 ohms

even if i gave the lte pads their own pins and the ft220x their own pins id still have to solder them both together

[idle python]

actually i can see why this would be wrong

if i wire the spi to a device and i also connect the usb pins to another device

the usb data incoming from both devices would be mixed

like if i send some data on device 1 then the lte will send the data requested by device 1 to both devices

[idle python]

if this is the only problem with the connection i made then i dont really care about it since obviously im not gonna connect both devices to spi and usb, pllus im the only one whos gonna be using this pcb

[flat vigil]

I haven't found a great reference for the protocol. I know STM AN2606 lists what pins it is active on for all of the different boards.

ah, found it. AN4221

https://www.st.com/resource/en/application_note/dm00072315-i2c-protocol-used-in-the-stm32-bootloader-stmicroelectronics.pdf

[tough matrix]

Nice.
I'd expect that someone has written an arduino script for flashing the bootloader to STM32 over i2c

[elder peak]

Heya, if I've got a switching power supply that's happy with Vin at 12V but the enable pin is only safe for 6V, if I just want a SPST switch to turn it on and off, the right answer is to make a voltage divider instead of just a pull-up resistor, right?

[unreal flax]

Yes, that seems right. It's a little bit unusual for an enable to not be happy with the input voltage, but if that's the case, a divider is the way to go. You can have the pulldown portion of the divider connected to EN, and the pullup portion of the divider on the other side of the switch.

[elder peak]

Yeah, I had three different step-down switching power supplies in my electronics hoard and 2/3 of them are happy with the enable line at the input voltage and this one... wasn't.

I haven't let the magic smoke out in years.

[paper vale]

Hello. I need some clarification please. I read an article about space vector modulation in the 60° coordinate system. I understood some parts but I have some incomprehension about the calculation of duty cycles. As for VLLg and VLLh, I don't quite understand what they refer to. According to the image VLL correspond to is when Vrg and Vrh are 0, 0 for example.
Does it mean that VLL correponds to the switching time 000 and Vuu to the switching time 111?
And what do VLLg, VLLh, VUUg and VUUh correspond to ?

[graceful sparrow]

Seeing as I have done a very similar thing (I think), with an "Ai thinker ESP32-CAM" board, I would like to throw that out there as an idea, I used a nRF9160 feather (in AT command mode). I will post a link to the blog page that I wrote about it, and hope that is ok, a lot of info is there.
Feel free to use any of the information there if you would like to help your project.
All source code, etc is published there.
https://www.drassal.net/wp/nrf9160-esp32-cam-jpeg-image-uploader-through-lte/

Adafruit also sells a version of the nRF9160 board that is almost the same, just to throw out an Adafruit link too https://www.adafruit.com/product/4753. A wonderful little board, if you want to experiment with LTE connectivity.

[vast flume]

Hey guys, I am in need of help again 😦

I cannot, for the life of me, connect to a bmp581 over i2c. I am using a pcb that is a modified version of a pcb where the sensor worked perfectly so I am so confused as to what could cause this

This is the footprint for the sensor that I made

And this is the pinout from the datasheet

I have the sensor oriented so that the port hole is on the top right, which should be positioning it in the correct orientation

[supple pollen]

It looks like your footprint has the marker at the other corner, which is confusing. There's an interior layer trace that goes under pad 8 and just stops without connecting. Is the I2C bus working otherwise?

[vast flume]

Yeah I messed up with the marker on the silkscreen, since the only visible marking would be the port hole which should be in the top right corner

One min I took a picture

Also here is the datasheet I've been following https://www.bosch-sensortec.com/media/boschsensortec/downloads/datasheets/bst-bmp581-ds004.pdf

To make things even more confusing however, is this is also in the datasheet:

[idle python]

were you able to do more complex things like send video files or make websockets with the nRF9160 ? im trying to make the esp32 cam livestream the camera feed with the lte module but the uart interface is not beefy enough for such task

[pale umbra]

Thanks. The part about registring business/tax etc is very familiar to me since I worked in that before for peoples that wanted to do this so Im very familiar with my state laws on this

[rigid plume]

What do you folks think of this ESP32 S3 soc based dev board (breadboard comp.)

500ma battery charging ic, batt fuel guage, rgb led, rf switch for ipex or pcb antenna and extra psram footprint on the bottom side of the board and cp2102n UART

[unreal flax]

I'd be a little bit worried about the proximity of the header pads right at the side of the PCB antenna.

[rigid plume]

Yeap me too, but I ordered some of the pcbs for a reference and they seem to be at an safe distance away

I don't think too much interference will be caused anyways with that.

[frosty scaffold]

Can I get some feedback on my schematic? It's supposed to be an ESP32 CANBUS reader/datalogger that uses 5V. I plan to use SPIFFs for datalogging.

[rigid plume]

Can you show your whole circuit incl usb?

Also connect the EP pins to GND

[frosty scaffold]

No USB. I would be writing the firmware to the ESP32 directly using one of these and use ElegantOTA to push firmware updates. https://www.amazon.com/gp/product/B08K2JKNRK/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&psc=1

[rigid plume]

Oh sure, that's great

It seems you leave alot of the GND unconnected?

[frosty scaffold]

Tell me more?

[frosty scaffold]

Ah one moment

Better? Do the capacitors look good? Should I add a capacitor to the input as well?

[rigid plume]

A circuit like this for the LDO would be good

Caps at the input and output

[pale umbra]

what are the caps for here ?

[frosty scaffold]

Filtering high voltage spikes

[rigid plume]

decoupling caps

They maintain stability

[rigid plume]

Great

[frosty scaffold]

One more thing: Should terminal 4 and terminal 2 of the AMS1117 be connected instead of terminal 4 going to GND? I saw some conflicting documentation in some schematics online

[rigid plume]

Term 4 is for heat dissipation, if you connect it to the gnd plane. Term 2 is the voltage out, if you connect them both it would short

[frosty scaffold]

Gotcha, thanks for the clarification

[rigid plume]

Also this is an TAC for the SN65HVD230D, you can put a 0.1uf decoup cap on vcc

You can add the act45b-101-2p-tl003 to supress noise

Will it only be for datalogging?

[frosty scaffold]

Yep, and sending some CAN signals if necessary

[rigid plume]

Oh, well then you can follow this. This is probably the best one

The diodes can be added which are the DFN1006-2L-BI-1 ( https://www.lcsc.com/product-detail/ESD-Protection-Devices_WILLSEMI-Will-Semicon-ESD5611N-2-TR_C2939834.html)

After doing this your fine

[frosty scaffold]

Hmm what are R18 and R21? 10 ohms respectively?

10R is a bit vague

Also, what direction are the diodes facing?

[rigid plume]

Nope they are not polarised

[rigid plume]

I think they are using them to prevent some filtering or interference.

The 120R termination resistor is already there so I don't see any problems with adding them

Also one last thing with your esp32 power input you should add an 0.1uf cap series to the 10k resistor to gnd from en.

[frosty scaffold]

Ah thank you

[rigid plume]

That covers everything, I hope I could help

[frosty scaffold]

You were a great help! PCB designing time.

[rigid plume]

Great

Just put NC on the pins you won't be using

Also you forgot a GND on the esp32 module

[frosty scaffold]

Ah i see it

[rigid plume]

On pcb design, make sure the decoupling caps are close to the ic inputs, they should not be far away

For the 5v input use 0.8mm on the traces, then on the input to the esp32 0.5mm is fine

0.254mm for everything else

[rigid plume]

Looks great!

[frosty scaffold]

Dope! Can I come back for more feedback when I have the PCB designed?

[rigid plume]

Yeap, I work with easyeda alot so that should be fine with me.

Also did you want power efficiency?

The AMS1117 is always overkill for most simple circuits

I recommend https://www.lcsc.com/product-detail/Linear-Voltage-Regulators-LDO_Torex-Semicon-XC6220B331MR-G_C86534.html

It's small and can still produce 1A, and works great with battery power circuits aswell

[pale umbra]

so about this pic, I checked that linear regular and it seems to convert 100 uA @ 5V to 1A @ 3.3V which kinda surprise me. I thought this was something to use usb as a power supply ?

but the datasheet says the max input current is 100 uA and supply voltage is 1.1V to 6.8V

[rigid plume]

the 100 µA figure you've mentioned refers to the quiescent current of the device.

It is the amount of current consumed by the device when it is not actively regulating eg when no load is connected or the output is not enabled.

The 1A is the max the load can draw from the LDO @ 3v3

[pale umbra]

Thanks for taking the time to answer. I like to learn from actual choices of peoples more knowledgeable than me and understand why those choices were made!

[frosty scaffold]

I actually could use a recommendation. I am currently stepping the power down from 55V to 5V and then again from 5V to 3.3V. Do you have a recommendation for something that could step down from 55V to 3.3V? I found a few regulators that could work, but I'm a bit skeptical since ESP32 could draw a lot of current when in Wifi, specifically for OTA updates.

[rigid plume]

The max the esp32 module your using should be around 600ma, your fine

That's steep from 55v

Checkout this https://www.lcsc.com/product-detail/DC-DC-Converters_Texas-Instruments-LMR36520FADDAR_C1858395.html

You won't get a decent efficiency, but it can output 2A aswell as support 5v upto 65v

[frosty scaffold]

Yeah, it's a battery pack for a ~48 volt Lipo battery. The BMS on that battery has a LM5017 buck regulator and was thinking of using one of those, but the rated max current is 600ma, which is really dang close

https://www.ti.com/lit/ds/symlink/lm5017.pdf?&ts=1589157738344

[frosty scaffold]

2A is hecka decent

[rigid plume]

This is the circuit for 3v3 output using that https://www.lcsc.com/product-detail/DC-DC-Converters_Texas-Instruments-LMR36520FADDAR_C1858395.html

[pale umbra]

Honestly it's just the esp8256 that has current issues with wifi That was sorta solved in esp32s. From what I can see the wifi uses about 200mA in "active mode" (ie: when not just maintaining connection to the AP)

[frosty scaffold]

Efficiency isn't that important considering that the battery is huge

Oh I should mention that the ESP32 is in AP mode

[pale umbra]

where as on the esp8256 sometimes the wifi will draw almost to the max current of the MCU

[rigid plume]

Yeap, just the IC is a bit expensive

[frosty scaffold]

$2 is alright

[pale umbra]

A shame that the esp32 doesnt have a low-power sleep mode for wifi it would be almost perfect with that 😄

[rigid plume]

I reccommend you this tool for checking what ICs you need https://webench.ti.com/power-designer/switching-regulator

[frosty scaffold]

Ah thank you!

[rigid plume]

I think it does

It is on the newer versions (s3)

[pale umbra]

hmmm I checked the S3 manual to say this statement and it says: "In Deep-sleep and Light-sleep modes, the wireless peripherals are powered down. Before entering Deep-sleep or Light-sleep modes, the application must disable Wi-Fi and Bluetooth using the appropriate calls (i.e., esp_bluedroid_disable(), esp_bt_controller_disable(), esp_wifi_stop()). Wi-Fi and Bluetooth connections will not be maintained in Deep-sleep or Light-sleep mode, even if these functions are not called."

That is why I was saying it doesn't have a wifi sleep / low-power mode as it doesn't seems like it will even maintain the AP connection

[rigid plume]

Well I don't think low power mode with wifi on is necessary for some applications

Deep sleep is enough for most, where the s3 will power on and send a signal or something with wifi then go back to sleep

[pale umbra]

I had hoped for a wifi wake-ip when I got my S2/S3s but it doesn't seems possible since you have to call methods to turn it off

For a small IoT device that did data gathering of a single sensor and I was hoping that adafruit-io/rpi base station could ping it every 10 mins by wifi. Had to change my approach because of this and it's annoying because it requires precise timing and checking time on both ends

[frosty scaffold]

@rigid plume Ready for feedback 🙂

[rigid plume]

Let me check it in easyeda

I'll send you the updated file soon

[rigid plume]

Also will you use JLCPCB to manufacture the pcb?

Will you assemble it yourself?

[frosty scaffold]

Probably yes. I’m thinking of getting the boards made by JLCPCB and maybe partial SMD assembly so that way I can flash the ESP32 modules myself and solder them on by hand?

Open to ideas of course

If it’s not evident, this is definitely my first time doing this lmao

[supple pollen]

I like the placement of C9, but I'm less fond of the long trace coming from it and going the long way around the board where it's going to impact the antenna's field

[rigid plume]

That's definitely fine, I'll remove the esp32 s3 from the BOM. You will just have to buy it from lcsc separately

assembly from jlcpcb will save you alot of time with small components

Is it fine if I put the antenna off the board?

You will get better signal

@frosty scaffold

[supple pollen]

Yes, I've seen many designs where the antenna hangs off the edge of the board. This works well and saves board space, but it does add an opportunity for the antenna to get impacted

[rigid plume]

Yeap definitely

[frosty scaffold]

Yep no problem. This entire board is likely going to be potted for water resistance anyway

[rigid plume]

I added a GND plane for simplicity, I then also made sure the traces were not right angled (not good for signal integrity)

For a simple board like this I think this is definitely enough

[frosty scaffold]

Ah, thank you!

[rigid plume]

If you solder on the esp32 , I recommend an hot airgun aswell as an soldering iron to correct some joints

[supple pollen]

And flux. Flux is your friend.

[frosty scaffold]

Dumb question, how do I view the front of the board?

[frosty scaffold]

I bought a lifetime supply of no clean liquid flux lmao

[rigid plume]

Also if you want to learn more about easyeda and routing and design in general of esp32 boards watch Robert Ferranec. https://www.youtube.com/watch?v=S_p0YV-JlfU

[frosty scaffold]

I was also following that video!

[frosty scaffold]

Found it!

[frosty scaffold]

@rigid plume Another dumb question: My 22 uF capacitors are out of stock. How do I replace them in the boardview and go back to schematic mode?

[graceful sparrow]

I don't think LTE-M or NB-IoT (the IoT lightweight version of LTE protocol which this nRF9160 does) has the bandwidth to do live streaming. It takes several seconds to upload an image. If you would like to live stream the camera I think you will have to go with a full blown LTE connection. But for doing things like IoT devices (MQTT connections, Azure, etc) the nRF9160 works great.

Personally I would try this project with a Rapsberry Pi Zero and a LTE modem (laptop version (mini PCIexpress with USB adapter). Most of the laptop cards work off a USB interface, then it is a matter to get a working driver setup on the Rapsberry Pi, then just treat it almost like a WiFi connection. Just make sure the adapter has a SIM socket, some do not.
For instance, these two below products might work well with a Raspberry Pi.
https://www.amazon.com/dp/B07YY5967K
https://www.amazon.com/dp/B079QG22NS

The EM7455 might need the firmware rewritten to enumerate correctly, internet searches should reveal that process.

[graceful sparrow]

If the footprint (package) is the same, there is no need to change that, just change your BOM file that you will send to the board house. But, yes, edit the component information in the schematic view, if you need to change the footprint, change it in the schematic view, then reimport the changes into the boardview (KiCad is it?). It will update only change components.

If the board layout is complete, I would recommend to find a replacement component that fits the original componenet's footprint, since this is the option that requires the least amount of work (and changes).

[graceful sparrow]

For pre-flashing, this is a cool solution.
https://www.amazon.com/dp/B08K2JKNRK
I have one for each type of ESP32 module.

Make sure to get the one that matches your part.

[frosty scaffold]

I’m using EasyEDA

[graceful sparrow]

Ah I'm sorry I have really only experience with KiCad, but, the same should still apply. As long as you choose a part that is the same size you will not have to modify the board layout (recommended). If you search for the specific part on Digikey there will be a specification that shows size (0402, 0603, etc). Match this package size up with the replacement part. Sometimes Digikey will even offer replacement suggestions at the bottom of the page that might be compatible.

In this case, you will only need to modify the part's BOM (parts list) data, which is most likely in the schematic view (if you edit the part).

[frosty scaffold]

I’m trying to get it fabricated by JLCPCB using EasyEDA’s built in utility, how would I swap it?

[graceful sparrow]

Looks like this might be the way to do it, with just a quick internet search.
https://easyeda.com/forum/topic/Replace_part_in_schematic-aEasKEVdv

[frosty scaffold]

Dang, it looks like I gotta modify the original schematic, and I don't have the original schematic. Guess I gotta do it the long way

[graceful sparrow]

Ah, the BOM parts data must be stored somewhere (see if you can get access to that and edit it). But honestly the PCB layout should not change and just requires a part substitution (part should be the same size). I'm sorry that is the extent I can assist with this as I am not too familiar with EasyEDA but hope you get it sorted out.

[graceful sparrow]

Here is one more link, try this editing method. https://easyeda.com/forum/topic/BOM_edit-APlfxrJDV

You may be able to open the file with a text editor (KiCad files can be edited this way). And do a find/replace on the part information.

If the data is not stored binary.

[frosty scaffold]

@rigid plume I made another revision! Would you be able to give feedback?

[woven bluff]

can flyback or boost regulator drive non-flyback transformer?

[rigid plume]

Yeap will check it soon

[rigid plume]

@frosty scaffold It looks fine and everything will probably work. Just consider these points when designing a pcb.

1. Route at 45* Angles, not at 90* (You will lose signal integrity)
2. Use the grid lines as a way to neatly route your connections

Avoid sharp traces like this

[supple pollen]

Not very efficiently. Flyback transformers have a gap in the magnetic path which stabilizes the inductance and stores energy. Without that gap, the core will saturate more easily (which is hard on the switching transistor) and without the additional energy storage of the gap, each pulse out of the transformer will contain less energy, reducing both the efficiency and power capability of the regulator.

For that trace coming in from the left at the bottom, going to the same pad as the trace from the upper left, I'd have the traces join to the left of the pad. This will also avoid cutting through the ground plane for the bottom pin.

[idle python]

lte-m has a speed of 1mbps on both up and down, b ut the ec800 is an lte cat 1 module with 5mbps upload and 10mbps download
with the esp32 cam having only MJPEG format, i could send over a 480p video on 30fps with low quality mjpeg, its not much but its something

i think i should just remove the ft220x chip, and add a raspberry pi zero in the project so it could convert the video feed from mjpeg to h264 and i connect the ec800 to the pi instead, no need for spi pins at all and no need for custom driver

[supple pollen]

Note, those are ideal speeds, and you'll often get less (sometimes much less) under real world conditions

[graceful sparrow]

Yes, for occasional sending of images I think NB-IoT or LTE-M would be ok but wouldn't recommend it for streaming. Also consider the interface between your MCU and the nRF9160. If you use UART you might be able to do 921600 but I could not get that speed to work reliable. At those speeds the length of wires and noise will start to become important. 115200 is ideally what I have been able to get to work reliably (with an ESP32).

I would look down the Rapsberry Pi Zero path, as it also has a built in camera interface, and access to a fast USB interface to an LTE modem.

[idle python]

yea i think im gonna use a rpi for the networking

but ill still use the esp32 as the camera cause i cant buy a csi camera right now

[supple pollen]

I agree, trying to squirt video over an asynchronous serial link seems iffy

[graceful sparrow]

You might find a bottleneck with the connection between the ESP32 and the Raspberry Pi (UART).
I believe the camera interface operates around 2M, I would have to verify that, but you might have trouble getting your data transfered fast enough. You will also have to devise a protocol to pass the data.

[cursive sentinel]

DisplayPort does this

[supple pollen]

I'm pretty sure DisplayPort is synchronous

[cursive sentinel]

It's asychronous

SERDES architecture

[graceful sparrow]

Best choice is to just take advantage of the Raspberry Pi camera interface (when you can acquire a Raspberry Pi camera).
The work has already be done and all you have to worry about is the software (personally I might use Python).

[cursive sentinel]

The minimum data rate is also 1.62Gbps though.

[supple pollen]

A little reading shows that DisplayPort is neither, it's isochronous...

[cursive sentinel]

But yeah you're not going to be able to send video data through a UART.

[idle python]

i wont need uart cause the esp32 makes its own wifi network and i can just make the pi grab the video from the web server esp32 is hosting

[graceful sparrow]

Ok yes, I forgot, that is an option, but adds a little complexity to the system.

[cursive sentinel]

Architecturally it is an asychronous interface though. Self-clocking.

Note: I do device characterization for video SERDES parts.

[supple pollen]

Right, self-clocking differential signal, not the scheme with start bits and stop bits like the sort of single ended asynchronous serial you get out of a UART.

[cursive sentinel]

Correct

[graceful sparrow]

If you do this, watch the work load on the ESP32, I have read the camera (or ESP32) can overheat if heat is not managed properly.
They do get hot when you stream, I made a WiFi streamer (streamed to a websocket on a server running Python) with the Ai Thinker ESP32-CAM (as a proof of concept) and it did get rather warm.

[supple pollen]

To me, "self clocking" is distinguished from (UART style) "asynchronous", even though neither of them have a separate clock signal like synchronous serial. The main difference being the number of bits you have to wait to resynchronize.

The timing benefits of a differential (versus single ended) signal are important too.

[idle python]

great

now i just have to remove the spi bridge chip

from the pcb i made and get a rpi zero 2

[graceful sparrow]

With that thought, if you want to try the streaming route, and want to direct the ESP32 to a Raspberry Pi running a Python script I might be able to give you a head start (the ESP32 code and some Python script to work with). The ESP32 sends a stream of data to a websocket (opened by a Python script on a server) which then accepts connections from browsers to watch the stream. My stream was simple though, it was a stream of JPEG images, not an actual stream, but it might be a starting point.

[supple pollen]

Amusingly, it's probably easier these days to get a Pi compatible camera than a Pi Zero 2.

[graceful sparrow]

They should be fairly common off Amazon (made in China) which I don't mind at times when it is just for a proof of concept.

[idle python]

i can get one from aliexpress for just 10 dollars but i dont know about the quality

[supple pollen]

I avoid sketchy sources like amazon for proof of concept, as spending extra days debugging isn't appealing at all. Aliexpress is a better bet.

[graceful sparrow]

But I hear the Raspberry Pi should be returning down to realistic prices by the end of this year, or early next as supply chains are reestablished.

[supple pollen]

When qq_ said "right now", I figured "by the end of this year" may not be in scope

[graceful sparrow]

Does Adafruit have any inexpensive Raspberry Pi cameras? I should take a look, they might.

$35 is not bad https://www.adafruit.com/product/5658
$25 too https://www.adafruit.com/product/5659

I had to edit that post, an auto bot kept rejecting my post with a link...

[cursive sentinel]

You are correct in saying it is isochronous. That's the term used in the DP spec. The fact it is differential has more to do with signal integrity than timing though, aside from the fact that the differential signals are AC coupled. The actual self-clocking mechanism is handled by the data scrambler and line encoding (8b10b for DP1.x, 128/132b split over four lanes for 2.0).

[graceful sparrow]

Adafruit is out of stock on all Raspberry Pi variants now it seems, but I remember hearing stock is occasionally put up so just click that "notify me" button and try to snag one as soon as stock appears. I would love to get my hands on a Zero 2 when I can.

[supple pollen]

The problem with a single-ended signal is the need to differentiate high and low levels using a threshold. A little noise might make the transition earlier or later. A differential signal usually uses a crossover threshold with hysteresis, which minimizes that jitter.

[idle python]

maybe i can get a banana pi m2 instead or another radxa zero? i have this radxa zero but its broken cause i tried to desolder the headers

[graceful sparrow]

Yes that works too, not sure on the camera interface compatibility, but if using the ESP32 for the camera then that can be a starting point, assuming you have the SD card prepared for it.

[cursive sentinel]

Yes, but that's still independent of the timing. There are multiple reasons you might want to use a differential signal, but the interface timing is not one of them.

[idle python]

which one would be the fastest at re encoding mjpeg to h264 tho

banana pi or radxa zero

[graceful sparrow]

Sorry not my area of expertise, but you could do some tests (using Python or such) if you have both of them functional.

[graceful sparrow]

That encoding might be CPU intensive unless you have hardware support (I think Raspberry Pi has hardware support?).

[cursive sentinel]

It does.

And hardware encoding is the only way you're going to be able to get a processor like that to encode H.264 fast enough.

[graceful sparrow]

↑↑↑

[idle python]

pi zero 2 isnt in stock anywhere unless you wanna buy it for 90 dollars

should i just buy a pi zero or a radxa zero

wrong channel

[woven bluff]

all pi SOCs have been out of stock for years

[pale umbra]

When I finally got my pi 4 b 8gb I opened a bottle of champagne. Also I waited for when I actually had it in my hands because the shipping trucks could have been raided like those bak trucks carrying gold ingots 🤣

[idle python]

well my only choice is buying a board that has little software support or buying a board thats weaker but great software support

i dont know if a raspberry pi zero with hardware encoding support on ffmpeg would be better than boards like radxa zero, mango pi, banana pi which have better hardware but no hardware encoding support (not that i know of, maybe there is hardware encoding support)

im not gonna be re encoding 1080p 60fps videos, esp32cam or a csi camera couldnt even dream of reaching those resolutions and fps at the same time

most im gonna be doing is 720p at max 30fps which shouldnt really be heavy on any of the boards i just mentioned

[idle python]

what does this mean

@supple pollen what do u think about this one

[twilit mango]

Hey folks. I have a weird question. I already talked to one of my folks who is relatively familiar with EagleCAD, and he didn't know. I also googled for an answer, and I'm finding only solutions for AutoCAD 2015-2018, and the one post I found for EagleCAD was unanswered. I have a kludgey workaround. However, before I accept it and give up, I figured I would ask here on the off chance one of you lovely people happens to know the answer.

I would like to move the dimension value text to the left side of the measurement from the right, to where it very adamantly defaults.

That way it can be next to the measurement to its right without the text overlapping.

[pale umbra]

can you show a bit more of the right side ?

[twilit mango]

Sure.

Hold on.

[pale umbra]

It might just be conforming to engineering rules about aligning measures

But Id need to see if it has a continuous line going to center of says an hole to conform to engineering standards for plans

[pale umbra]

so you want the 0.10 on the other side because it's showing the size of the bottom-left hole ?

[twilit mango]

I want it on the other side so I can make it closer to the board, and not have it overlap like this.

Smaller measurements are to the right. Larger ones are between the lines. If that wasn't already obvious.

I'm thinking the reason all I was getting were AutoCAD results is because it is perhaps not possible in EagleCAD.

[pale umbra]

Im still not sure what it is measuring 🤣 so confusing having it on the right side

[twilit mango]

Hah!

The hole above it, on the left side of the board.

[pale umbra]

Like I dont know if its measuring the big hole on bottom-left or the distance between the green IN hole and the S- one

[twilit mango]

Not sure myself if it's a mounting hole or not, but we like to have thorough dimensions.

Oh!

[twilit mango]

It's way more obvious when they are not overlapping

I did that to show my frustration.

That's not how it's going to be in final.

Here is my kludgey workaround. I changed the size of the text. Is that clearer to you?

[unreal flax]

Does applying the rotate or mirror operation to the measurement object do anything?

[twilit mango]

Oooooh. Not sure, trying that.

Mirror won't apply to it. Refuses to even accept it as a mirrorable object.

Same for rotate.

[idle python]

new product?

[twilit mango]

NeoKey Key Switch Socket Breakout for CHOC key switches.

It's in the shop. 🙂

I'm writing the guide.

[pale umbra]

all you seem able to do is to measure it upside down so it show on the other side vs left-right

[twilit mango]

Actually, perhaps better workaround, I'll move the 0.33 measurement to the top of the board. Both sets of pads are available there too. It will fit there inside the full board width measurement.

[pale umbra]

like with the 0.10 atop the lines instead of at the bottom like in your pic

[twilit mango]

Yeah.

That's better.

Usually it's routing the board that turns into a puzzle, not adding the dang measurements. Uff!

[pale umbra]

what layer does adafruit usually put the measurements on ?

47. measures ?

[twilit mango]

Measures

[pale umbra]

dont know if that is planned but are diameter/measure of the mounting holes usually included?

[twilit mango]

Thanks for looking into this with me, folks! I appreciate it. I feel sort of better that no one else can figure it out either, and at this point feel like it's simply not possible.

[twilit mango]

They're included in the fabrication print image that is on the Downloads page in the guide.

The PCB files are always there to check yourself, but we include images of the board and schematic so folks don't necessarily need Eagle to look at them.

[pale umbra]

I tried many tools on the measures layer and nothing seems to let me adjust where the dimension is/alignment/style etc

[twilit mango]

Thank you

[idle python]

did adafruit ever produce white pcb versions of the powerboost 1000c

[pale umbra]

I miss autocad dimension styles now 😦