help-with-hw-design

[cinder grail]

boost converter?

[cinder grail]

So it wouldnt break anything so long as i could boost the voltage?

[supple pollen]

I think the idea is if you ran a 12V brushed fan on 5V, it would still work, just at a lower speed.

[dry pelican]

No. Unless you overcurrent the esp's 5v traces. But depends on the power of the fan

[balmy tide]

I'm trying to design a pcb that requires some traces to be certain impedance and inductance. Can anyone give me a good starting point?
For context, i am trying to make a cable to extend mipi-csi line, i want it to be approximately 30 cm. I found renesas' design guides that stated resistance should be less than 30mohm, inductance less than 2.8nH. I found that my trace width should be 5mm (that's kinda huge but ok) additionally i think traces should have 100ohm impedance (kinda matches cat6e) and 200MHz data rate per lane. I don't know how toproceed from here.

[vast flume]

Hey guys, I am trying to use this lcd: https://www.digikey.com/en/products/detail/focus-lcds/G126CLGFGLW6WTCCXAL/12691108

It says that it needs a 28 pin 1.27mm ribbon cable but I just cannot find one anywhere

The only one I found is only available for order in multiples of 1000

[dry pelican]

I think that if you're doing impedance matching stuff, you need a 4 layer board so you can have less distance from trace to ground plane and then your traces can be thinner. And maybe you also need to length match. I don't know for sure tho. I've never done anything high speed.

But cables are easily available

Unless it's some custom flex pcb

Most of those flex connectors just have straight traces

[distant raven]

Yup, shorting return to ground reduces trace width requirements for controlled impedance traces

[balmy tide]

I for sure know they need length matching. The 2 way differential lines send 3 way differential signals or something like that. C-phy is pretty complicated. I'm glad I don't have to understand it that well

[balmy tide]

I thought about using rpi camera ribbon but it didn't have enough conductors

The magic is finding width and space between traces

Which renesas' doc says some stuff

But I'm too clueless to understand

[dry pelican]

There are differential impedance calculators

Plug in numbers and it does the magic

[distant raven]

Yup, Digi-Key has one

[balmy tide]

I'll try it

[vast flume]

Where do I get the source voltages for the v0-v4 pins? There is an internal charge pump in the driver chip but the datasheet doesn't explain anything

Or do I just not have to connect anything to those pins?

Actually, ended up looking at the datasheet for the st7565r and it looks like you're just supposed to take the output from the charge pump and feed it into v0-v4

Or am I reading it wrong lol

[rustic linden]

I thought it was talking about an internal regulator/pump but reading it made me understand it less

[vast flume]

I'm so confused the datasheet is terrible

https://focuslcds.com/content/G126CLGFGLW6WTCCXAL_Spec.pdf

Also I cant figure out what pin 27 is supposed to be

Nvm I found that in the chip specific ds

[vast flume]

From this it looks like all 4 just go to ground I guess

This is my final setup

[rustic linden]

Neither of those are showing them pulled down to ground

There's a capacitor between them, at the least

[vast flume]

Oh right yeah sorry

So ig they're internally connected and then the external cap is to just keep it stable

[rustic linden]

it is weird that it said VDD2 or Vss

[vast flume]

Ok one last question, db0-db7 are for a parallel bus. Do I need a pullup/pulldown resistor or does it go directly to my mcu?

[rustic linden]

Because usually those are very different things

[vast flume]

I agree

[cinder grail]

It's a little 40 mm fan

[cinder grail]

It's a little 40 mm fan

[dry pelican]

Probably fine

It shouldn’t blow up any traces

[cinder grail]

ok, thanks

[lost sky]

Might be the wrong channel, but how do guys avoid...this mess in KiCad?
They are R1 - R9 and all 1k - is there a better way than just reorganizing everything?

[supple pollen]

I'd probably make a smaller resistor symbol and text.

Or just hide the labels and add "9x 1k"

[distant raven]

Or fan them out more

[supple pollen]

Oh yeah, offsetting them horizontally would work too

[dry pelican]

Should I length match the traces going to this op amp acting as a CC source?

I'm thinking it might improve accuracy as now the trace resistances are mostly the same

[unreal flax]

I'd be surprised if it would matter. Op-amp inputs are very high impedance, which means there should be essentially no current flowing over that trace, so the resistance will have a minimal effect on the voltage, especially at the milli-ohm level.

[dry pelican]

That's what I would think. But Kelvin connections are used all the time in current sense applications and it's apparently because that small resistance can matter.

[rancid lagoon]

It probably won’t hurt anything to have them length-matched

[unreal flax]

That's because the currents involved are much higher... if you've got 10 amps flowing through your sense resistor, which might be only 1 milli-ohm to begin with, then small differences can matter. But note that a Kelvin connection is not really about length-matching the traces, but about connecting to the sense resistor in places that aren't influenced by the main current path.

[dry pelican]

Ah ok. That makes sense

[woven bluff]

where can I find good quality JST PH battery cable? I don't want to crimp myself..

[woven bluff]

what is the purpose of the yellow tapes on transformers?

[blissful frigate]

It's insulating tape

used to e.g. increase the breakdown voltage between windings

[unique patio]

https://www.adafruit.com/?q=jst+ph+2-pin&sort=BestMatch Note that there is no "standard" about the polarity of the pins. All Adafruit cables and batteries are one way, but be careful of other brand batteries or cables.

[woven bluff]

shouldn't insulating tapes be teal colored? and sometimes the yellow tape only warps around the core, not the windings

[supple pollen]

There are a variety of tapes in use, the yellow (polyester?) tape is commonly used on transformers.

[dry pelican]

I think the yellow tape is usually polyamide

[supple pollen]

Doesn't look like it: https://www.digikey.com/en/products/detail/3m-tc/1350F-1-0-75-X72YD-YELLOW/4866333

[woven bluff]

I think polyamide is the brown transparent one

[supple pollen]

Actually polyimide is the brown transparent one. Polyamide is nylon.

[woven bluff]

yellow tape is the cheapest and leaves nasty adhesive behind

[supple pollen]

I use it anyway when winding my own transformers

[pale umbra]

Can you accidently create a hazard if you wind it too much or if you use a wire gauge that is too large ?

[dry pelican]

I meant polyimide lol. Didn't really know about the spelling difference

[supple pollen]

I'm not sure about a hazard, but the transformer may not work as well if I get the wrong number of turns, and if I use wire that's too thick, the windings may not fit in the space available.

[stoic birch]

got reads i was looking for with a different potentiometer

[hushed smelt]

5" LED boards for my mailbox project came in. No idea if they work yet. 🤞

The interlocking design allows for mounting on curved surfaces, to an extent.

[unreal flax]

Nice design!

[hexed mural]

That’s a lot of LEDs! Outstanding

[balmy tide]

Are they all parallel?

[blissful frigate]

Not necessarily the cheapest. The better "yellow tapes" don't leave any real residue behind

but I've had some cheaper ones that really were quite horrible

[hushed smelt]

Chaining 2 together on a feather results in an error message in Windows I've never seen before. That's pretty cool.

https://tenor.com/view/we-gotta-use-more-power-power-rangers-zeo-exert-more-power-a-little-more-power-just-a-little-more-gif-19564542

[hushed smelt]

They are... whatever this is

[balmy tide]

They are parallel

[hushed smelt]

then yes

😁

[hushed smelt]

This is the flashing light script I came up with for the mailbox notification.

[blissful frigate]

very pleasant, lol

[rain remnant]

https://tenor.com/view/unlimited-power-star-wars-gif-10270127

[woven bluff]

that transformer is huge, are you still using THT parts?

[supple pollen]

Yeah, it's a vacuum tube power supply, through hole is still popular with the bigger parts and higher voltages involved (that particular one as you can see from the silkscreen is a project presented by Elektor magazine)

[woven bluff]

more and more new chips are available only in FN or BGA...

[supple pollen]

Generally not chips working with hundreds of volts...

[vast flume]

Making my own skydiving altimeter, just ordered my pcb 😅

Idk if anyone here remembers but a long time ago I previously attempted but the board was doa

[woven bluff]

I'm not sure rolling your own life-critical device

[vast flume]

Nah I'll jump it with a legit altimeter as well lol

But yeah it'll be super cool. 3.17" ftsn lcd screen and powered by an stm32. Also has 5 super bright neopixels. Will show the time to altitude while in the plane, and then switch to showing altitude while on the way down. Later it would be cool to integrate a gps so it can have a compass rose pointing back to the landing zone but I didn't want to take the chance on scrapping the whole board bc I messed up with an extra feature

The enclosure will be 3d printed with a soft tpu outer shell as well

[woven bluff]

not sure if consumer IMU works at high altitude, the earth magnetic field would be weaker

[vast flume]

Oh no reason to use a imu

Using a normal baro sensor

Dont need 6dof lol just need to know how high I am

[woven bluff]

I mean the compass

[vast flume]

Ah the gps gives you your course

And thats all that really matters

[supple pollen]

Just have to remember direction of motion may be unrelated to the way you're facing

[woven bluff]

my GPS does not work when there's cloud cover

[vast flume]

Yeah but when you're under a parachute you track forwards

[vast flume]

Yeah that would be a potential issue

[supple pollen]

Relative to the air you're in, which may itself be moving at a fair speed

[vast flume]

If you're crabbing more than a couple degrees you shouldn't be skydiving anyways

🙂

[woven bluff]

also remember your hands will be occupied

[vast flume]

Ofc, but theres already skydiving altimeters that do all this

[supple pollen]

This isn't referring to crabbing, it's referring to wind.

[vast flume]

Right but the wind causes a crab angle (depending on your direction relative to the wind)

[supple pollen]

Yes and no...

[woven bluff]

it would be nice to have AR glass

[vast flume]

Wdym?

[woven bluff]

is there an open source or fully hacked AR glass?

[vast flume]

I've seen some, and theres actually a company working on an ar altimeter for skydiving lol

But yeah pretty much just making this:

With a bigger display and brigher leds

[woven bluff]

lol price tag

maybe you can get it under $100 if you roll your own

[vast flume]

Yup thats exactly how much I spent on mine

[woven bluff]

also you'll need anti-reflection coated glass

[vast flume]

$22 for pcb shipping and $80 in parts

[vast flume]

My display is a transflective ftsn so it shouldn't be glossy or anything

[blissful frigate]

If you call that huge you should see the ones I design at work, lol

Stacked E100 cores

[rustic linden]

What's wrong with THT?

[woven bluff]

most new chips are not available in THT anymore

[woven bluff]

I think it's large for vacuum tube power supply. It's small compared with those the oil-filled transformers that sit on electricity pole.

[rustic linden]

Unfortunately, tbh

[fringe jacinth]

Hello dear people I'm currently designing my own USB 3.0 Hub. I've found all of the parts that I need, however I'm running into one issue at the moment:
I use an USB Type C (3.0 for SuperSpeed) as the power/data source which has to be connected to the USB Hub IC. The "problem" is, that the Hub IC only needs TX+,TX-,RX+,RX- (all of the superspeed signals) - however type c provides these "pins" two times because it's reversible. Now the thing is, I apparently can't just connect TX1+ to the TX2+ but rather need a USB 2:1 Mux which "switches" depending on what data line actually gets used. A mux however needs one pin "select" to be pulled up or down depending on what of the 2 inputs lines I want to output: How do I determine which of the ones of the Type C Port are used tho?

[unique patio]

I thought there was a way to determine the orientation of the C plug: https://acroname.com/blog/why-usb-c-connections-sometimes-dont-work#:~:text=The indication of orientation of,be used as “CC".

[fringe jacinth]

There is! It seems to be done/able with the CC Pins (CC1 and/or CC2). But I just don't quite get how I could implement that into my circuit.

This is all I have right now. With this schematic (of an 2:1 mux) I already kinda know what I have to do: Detect the orientation with the CC Pins and depending on one of these values (or maybe compare both?) I can input a high or low voltage into the SELECT Pin of the mux which will then switch to the used data lines

[unique patio]

there seem to be a number of reference designs. I searched for usb c hub schematic and got a number of interesting things, including youtube overviews and mfr datasheet designs

i don't see that you need an external mux

[fringe jacinth]

Hm most of them are usb 2.0 (or actually all of the ones I found). Some USB 3.0 ones are there, but then they use micro usb as "power/data" input which isn't reversible so it only has TX1 (no TX2 and such)

[unique patio]

https://www.microtype.io/open-source-usb-hub-design-with-kicad/

[fringe jacinth]

uh but I found a video where a guy uses type c and 3.0! I'll take a look at that. He also uses a mux.
In case anyone else is interest, I'll share it here https://www.youtube.com/watch?v=m3twt3aY2nQ

[unique patio]

have you taken apart an existing hub?

[fringe jacinth]

nope, never 😅 I dont have one but I assume they'll also use some sort of Mux or an HUB that has a mux integrated

[unique patio]

https://www.ti.com/lit/ds/symlink/hd3ss3220.pdf?ts=168086049986 says it has an "integrated 2:1 SuperSpeed Mux"

[fringe jacinth]

Apart from the USB C to the Hub IC connection (because of the reversible port) everything else seems to be clear and easy to understand from the sources online. It's just that small thing thats difficult for me right now. And I think most of the USB Hubs available have a cable soldered on

[unique patio]

https://electronics.stackexchange.com/questions/587526/usb-c-to-c-receptacles-with-a-multiplexer-in-between

https://community.infineon.com/t5/USB-superspeed-peripherals/USB-type-C-without-mux/td-p/203615

I searched for usb c hub mux tx1 tx2

[fringe jacinth]

I might have to learn first how to properly google 😬 There seems to be more available than I thought or found

[unique patio]

doing good websearches is an art

[rain remnant]

bing ai search might help

[fringe jacinth]

uh right, thats a thing now ^^ Or just use chatgpt directly haha!
Alright thanks danh! I'm sure with that I will find something that I can actually implement

[pale umbra]

Do you think it would still complain with this ?

I don't even think it's big enough for the surface area that would be taken by all the usb ports it could power 🤣

[pale umbra]

ChatGPT is great at convincing you a false answer is true

[unique patio]

I just found this interesting observation: https://social.coop/@DrewKadel/110154048390452046

Something that seems fundamental to me about ChatGPT, which gets lost over and over again: When you enter text into it, you're asking "What would a response to this sound like?" If you put in a scientific question, and it comes back with a response citing a non-existent paper with a plausible title, using a real journal name and an author name who's written things related to your question, it's not being tricky or telling lies or doing anything at all surprising! This is what a response to that question would sound like! It did the thing! But people keep wanting the "say something that sounds like an answer" machine to be doing something else, and believing it is doing something else. It's good at generating things that sound like responses to being told it was wrong, so people think that it's engaging in introspection or looking up more information or something, but it's not, it's only, ever, saying something that sounds like the next bit of the conversation.

[pale umbra]

Like if you asks it to write you a circuitpython program for a X chip board it will not only accept to write circuitpython for a board that doesn't actually support it but will also link you to python libraries when it's not even the same language

[unique patio]

yes, I had it hallucinate a Trinket M0 library analogous to the Circuit Playground library. But this is perhaps a topic better for #general-chat

[pale umbra]

And when you challenge it it will make more lies or you will get a yellow ban warning or suddendly you are out of tokens to use it on a free account (because challenging it is against ToS)

true danh but I feel like they should be warneds that asking a cat is a better option than chatgpt

[unique patio]

I posted a ChatGPT warning a while ago at the end of #welcome.

[pale umbra]

I joined long ago and would never think of looking at #welcome again - but yeah let's continue somewhere else if you are willing

[unique patio]

#general-chat is good; I don't really have more to say, though, just it gets frustrating that I have to tell people to stop believing ChatGPT

[hushed smelt]

It does feel like that when a project takes a lot more power than you were expecting.

In my defense the datasheet for the LED is mostly in Chinese.

[pale umbra]

which led and datasheet ?

[twin geyser]

Hi! My hobby project uses infrared similar to laser tag, and I was super excited to see the high power IR breakout (https://www.adafruit.com/product/5639) so I didn't have to solder my own little amps anymore. But now I've added a prop-maker (https://www.adafruit.com/product/3988), and my understanding is that each of the 3W RGB driver pins puts out ~350ma of current. Can I just add a resistor in series to drop the voltage down to 1.6V and use that as an IR LED amp? I actually want two IR LEDs at different viewing angles, so borrowing two of the RGB driver pins would be even better. I'm pretty new to electronics, and am concerned that I wouldn't find out that it was a bad idea until way farther into the project. Does anyone know offhand, or should I just try it out?

[unreal flax]

Yes, you should be able to do that. The Prop-maker has a 3-ohm resistor on the red driver and 1.75-ohm on the blue and green channels, but if you take those into account you should be able to add some additional resistance to dial in whatever current you want through the 1.6V IR LEDs.

[twin geyser]

Very exciting, thank you! I'll try it out tonight

[woven bluff]

I'm not a fan of spending my time contributing to someone else's supervised learning without being paid.

[woven bluff]

why interfere with natural selection?

ChatGPT is like the ML version of Tide pods

it smells nice and looks delicious

[dry pelican]

Well chatGPT is a really cool technology. It's just not factually accurate because it's not a real person. It's just a complex network of linear algebra that is EXTREMELY GOOD at predicting the next token in a series of tokens and is able to do so to form coherent answers.

AI is a black box that is good at doing imprecise tasks that have a lot of complexity (like image or text generation or recognition), but it can't be hard-programmed (at least not yet) to always do exactly what you want it to do

[latent jungle]

Context is a very abstract concept.

[woven bluff]

AI is not a black box, people wrote it from scratch after all

the hardware that ML runs on is also designed from zero

[unreal flax]

It's a black box in the sense that the AI's logic was trained rather than designed, so it's hard to predict or understand what features of the data it is paying attention to or how it's making its decisions. It's kind of like children: you can raise them and discipline them, but you can't really control what they do or know what's going on in their heads.

[teal void]

Is there a way to confirm that SCK is actually GP18 on the KB2040?
https://learn.adafruit.com/adafruit-kb2040/pinouts

[unreal flax]

Yes, the "downloads" section has the board schematic, and it indeed shows SCLK connected to GPIO18.

[teal void]

Is there a test to see if my board is malfunctioning or something? I've got some really weird issues and idk what else to do.

[unreal flax]

A good general troubleshooting technique is to simplify. Remove code and pin connections until you get back to something that definitely works, even if it's just a blinking LED. And then gradually add things back one piece at a time, keeping them working at each step. That'll let you pinpoint where things go off the rails and make it easier to figure out what's going wrong (or easier to ask for help with a very specific problem).

[teal void]

I think you're right I have to go back to the very beginning and take it a step at a time. I've tried posting my very detailed issues between channels and two other discords, but not getting much help yet.

Since this is for a keyboard i'm building, I was able to get everything to work EXCEPT keys 1, 2, 3, +, and / i'm struggling trying to figure out which pins they require

[unreal flax]

Generally I'd expect they could just use any unused pin, if the code specifies which is which. Is the difficulty in finding pins that are unused?

[teal void]

I'm going to take a step back...I'm trying to matchup the pins with the one on the right here

Reading through https://learn.adafruit.com/adafruit-kb2040/pinouts i'm trying to figure out if the SCK is D13, MISO is D12 and MOSI is D11? The article mentions D12 and D13, but it's referring to the STEMMA QT connector? and I see no mention of D11 on the KB2040

[supple pollen]

The RP2040 offers two SPI buses, and can connect them in various ways. Bus 0 SCK can be D2, D6, or D18, for instance.

[dry pelican]

Yes

Look at the datasheet for which pins can go with which buses

[teal void]

Is the KB2040 flash chip W25Q080 or something else like GD25Q64CS?

[dry pelican]

Just a QSPI flash IC

[teal void]

okay so theoretically MOSI can be D11, and MISO can be D12, etc?

[dry pelican]

Pretty much anything woukd work with it

[teal void]

Okay, I'm just trying to figure out if I need to #define RP2040_FLASH_GD25Q64CS or similar, or leave it default

[dry pelican]

Oh

Uhhh

Idk

[unreal flax]

The exact BOMs for Adafruit boards are sometimes hard to find, so it may be easiest to look at the chip markings on your particular board.

[teal void]

The creator for the PCB i'm pairing this with said, "checked our SPI flash testing, the W25Q080 on the KB2040 should be compatible with the RP2040_FLASH_W25X10CL we use"

[hushed smelt]

You can usually at least find the component values in the schematics which are usually found in the Downloads section of every learn guide. https://learn.adafruit.com/adafruit-kb2040/downloads

Granted some schematics are better labeled than others. While designing schematics sometimes documentation is the last thing on your mind.

[supple sky]

So I have been using this USB switch to switch between my work and personal PC for about 3 years now. It's worked well but it's getting kind of slow and sometimes fails to switch. It can take 5-10 seconds to fail a switch. I was curious if anyone uses a USB switch often (preferably one with a remote switch like this). Also, if you have made one, I'd love to hear more/see if this is something I'd like to find out more about hardware wise. Like, what hardware would a switcher need to have to be faster? I switch like 100 times a day, so my budget for something fast is pretty high.

https://www.amazon.com/gp/product/B07RC8F2L3/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&psc=1

[supple pollen]

Most of the KVM switches I've used have been very fast.

[supple sky]

Well, I don't really need a full KVM, just keyboard and mouse switch. I have a monitor that I use for my personal PC only that's below my ultrawide. I just switch inputs on my main monitor when I am fully done working for the day.

[rustic linden]

Wanting to design something that plugs into a breadboard, but how the power strips are offset is bugging me

[hushed smelt]

That's actually a good idea. The only thing you really need to control different PC's is a keyboard/mouse these days and they're USB now not PS2 so a traditional KVM is kind of an old way to do it. I suppose it still qualifies as a KVM though just a USB KVM.

Google Synergy is a networked software version of a KVM but it's a pain to setup correctly and if your PC is using wifi and wifi goes down so does your ability to control the other PC's. As long as you have 2 or more PC's sitting right next to each other a hardware KVM of some type still makes the most sense.

I could see something like a USB KVM being beneficial for switching between microcontrollers too,

[long wraith]

You mean the rails don’t follow the grid of the rest of the board? Unfortunately that’s also not consistent among all breadboards…

[hushed smelt]

They do but they're offset intentionally so someone doesn't accidentally use an IC to jump from a main rail. Not a problem when using breadboard wires but with IC's it's an intentional safety feature to offset the rails.

Alpenglow makes a neat little breadboard power switch called the Switchtrick. Maybe that's kind of something you're looking for?

https://www.alpenglowindustries.com/switchtrick.html#/

[long wraith]

I guess I have unsafe breadboards, lol. Some of mine don’t have an offset

[hushed smelt]

Yeah guess it depends on the breadboard. There are all kinds of different types out there.

[long wraith]

Someone should make a standard… XD

[hushed smelt]

I use the large one from Adafruit. https://www.adafruit.com/product/443

[long wraith]

I think my big ones are the same offset as those

I just have a bunch of random loose half and full breadboards

[hushed smelt]

I actually tend to stick with permaproto's and perfboard for small prototypes. It's rare I need to breakout the big one.

[long wraith]

I tend to favor 2 half breadboards together if one isn’t enough

Most of my projects that go beyond breadboard just get direct wired

[rustic linden]

nah, it was more for connecting power to an external circuit over providing power to a breadboard

[hushed smelt]

That's true too. Usually I'll just put stacking headers on a feather and use dupont wires to directly prototype. It becomes a spaghetti mess pretty quick.

[long wraith]

Spaghet

[hushed smelt]

The prototype that eventually got modified for the Dragonskull halloween mask.

Noods are... noodley.

[long wraith]

https://tenor.com/view/jaiden-animations-cute-gif-12028223

Noodles

[supple pollen]

Assume you (and your chute) are facing 0° (north), and that you're not crabbing at all, moving that way at 1m/s (we'll overlook vertical motion for now), relative to the parcel of air you're in. Now suppose the air at that altitude happens to be moving 135° (southeast) at 2m/s. Your actual motion relative to the surface of the earth will be the sum of these two vectors: 1.5m/s at 106°. This is the heading your GPS receiver will give you, so while you're facing north and not crabbing, your heading will be 106° (almost east).

[hushed smelt]

That makes sense. direction and heading are not mutually exclusive.

[woven bluff]

thins is kicad official footprint

but I'm not seeing thermal via on the other side

'

[supple pollen]

Looks like they're there, but that's a weird rendering. I ended up crafting my own TI footprints, after I decided I didn't like the TI-supplied ones.

[distant raven]

I didn’t like the suggested pad size of the thermal/gnd pad for the ice40 qfn chips. I shrunk it a little bit to route a wire or two without going under it

[supple pollen]

In my case, I didn't like that the pads were built up out of a welter of strange little polygons so I replaced them with rectangles with filleted ends. This was also cleaner for DRC, which objected to the overlapping polygons.

I also made the pin 1 indication more prominent (realizing later that it makes a strange little face)

[vast flume]

Ah I see. Yeah not technically crabbing but for all intents and purposes is no different in this case lol

[flint zodiac]

can I mark a product I designed as compatible with adafruit's STEMMA QT, and if so, what are the requirements?

[rustic linden]

https://learn.adafruit.com/introducing-adafruit-stemma-qt/technical-specs

Those are the technical specs for Stemma and Stemma QT

[flint zodiac]

oh that's pretty clear then. I had only seen some of the other pages. Thank you!

[oblique prawn]

quick question, when it comes to spec'ing out a DC switch in terms of amperage, is it based off of the source amperage or the limit of amperage being pulled by the switch?

[supple pollen]

The current being pulled through the switch into the load.

[oblique prawn]

perfect, just confirming

Also, does anyone have any good sources for electrical component libraries for Fusion 360?

Attempting to use the Electronics Design component in it to document a circuit as well as layout a few PCB's to get sent off, but the library is fairly limited. I could move to Altium, but I'd like to stick to Fusion so I can migrate the PCB model out to my 3D design of the case and other hardware it's interfacing with as a reference part.

[supple pollen]

SparkFun, AdaFruit, Microbuilder, and many others offer component libraries.

[oblique prawn]

not just the step files, but the actual importable library of the component for the PCB design and simulation?

[supple pollen]

I don't know about simulation, just the circuit symbol and footprint.

[oblique prawn]

so just a schematics

[supple pollen]

And board layout, yeah.

[oblique prawn]

I'll have to keep digging, everything I see is a whitepaper PDF

[supple pollen]

For reference, here's the AdaFruit one (which also has a link to the Microbuilder one): http://www.ladyada.net/library/pcb/eaglelibrary.html

[oblique prawn]

Thanks

I've been checking the product pages to see what was available for some of the connectors and mechanical components

[dry pelican]

LTspice is a circuit sim with a lot of parts

[oblique prawn]

thanks

[woven bluff]

how does digital benchtop DC power supply controls voltage?

[supple pollen]

Generally by having a pass element that is varied to conduct enough to provide the voltage specified.

[spice turtle]

generally by varying the feedback loop using a DAC

[woven bluff]

that's true for analogue DC power supply, but digital version does not have pots

[woven bluff]

PIC SMPS?

[spice turtle]

By varying the duty cycle while reading back the voltage from a feedback system. You can do it in software or hardware

Microchip has tons of App data sheets on this subject

"AN1086 - Switching Power Supply Design with the PIC16F785" is one example

AN3725-Buck-ConvDesign-Feedback-Ctrl-Using-CIP is another

[woven bluff]

the programming looks very intimidating

[spice turtle]

If you do it in software (ie, read ADC, then do something like update duty cycle) it can be easy

the hardware way, using built in stuff, does seem difficult

[woven bluff]

you mean peripherals?

ofc they are used, like timer-counter for PWM and ADC interrupt stuff

[spice turtle]

yes. But there are peripherals like the Op-Amp that can be used

and basically be a hands off approach, ie no software over head

[woven bluff]

if CPU hangs, smokes...

[dry pelican]

I was thinking a hybrid approach where maybe the mosfet is controlled by a stable pwm controller that is controlled by the micro

And most micros have pwm peripherals that are not dependent on cpu iirc

[woven bluff]

me too, DAC injection to feedback node

[dry pelican]

Maybe a DAC functions as the voltage reference

That could work

[spice turtle]

hence the hands off approach. But even Microchip used software to do it

[dry pelican]

Now I wonder how my bench PSU does it

[woven bluff]

I think the proper way to do it is to use as much interrupt IRQ and peripherals as possible and minimum CPU overhead

[spice turtle]

See also: AN1047_Buck_Boost_Driver
TB3102-Boost Converter Using the PIC12F1501 NCO Peripheral

[spice turtle]

problem is you are limited by your ADC read frequency. Then you have to factor in computation and logic

[woven bluff]

for cheap type with just pot with knobs off-line flyback

[dry pelican]

Knobs with flyback

[woven bluff]

I have no idea how digital types work

[dry pelican]

Exactly. I took some pictures of the inside

Didn’t look at the digital section though

Maybe for a flyback type, some stuff goes on with the optoisolator feedback

TL431 is often used in SMPS, so maybe by tweaking that voltage

[spice turtle]

Possibly they just generate a constant voltage and use an Offline dedicated SMPS and then use a digital Buck like I mentioned above.

Thats how I would do it

[supple pollen]

I didn't say anything about pots, I mentioned a series pass element.

[woven bluff]

how does that connected to the digital front panel?

I want to roll a battery powered portable DC supply

0-30V, 80W max, 4x 18650

and there's those cheap chinese SEPIC modules with XL6019 regulator

and cc/cv coupled SEPIC module with TPS40210 (which they filed the part number off the chip)

I originally wanted 4 SW BUCK-BOOST but seems too complicated, now I'm leaning to SEPIC

[supple pollen]

There are a myriad of possible ways, but the usual lashup is some sort of feedback loop where the actual output voltage is compared to the desired output voltage, and the difference is used to control the pass element.

[woven bluff]

control the pass element how? DAC bias injection?

[supple pollen]

Again, there are a myriad of possible ways, both analog and digital. Using a DAC is a reasonable approach.

For a well written description of how power supply control loops work, you can find copies of the "Kepco Power Supply Handbook" easily for <$5

If you want an off-the-shelf module, an RD6006 would probably run fine from a quartet of 18650 cells.

[spice turtle]

evblog did a review on those units and he gave them a glowing review. He also went in depth on how to design your own as well

[spice turtle]

With your power requirements, why not go with a single FET? Your FET wont be dissipating the whole 80W. I think this is easily do-able with a External controller.

[supple pollen]

That's the advantage of switchers, the notion is that the FET spends most of its time either not conducting (and therefore not dissipating power) or in a very low resistance state (and therefore dissipating little power). The tricky part is switching between them quickly enough to keep dissipation low, which requires high current gate drivers to charge and discharge the gate capacitance quickly.

[spice turtle]

💯

[dry pelican]

UCC27614 go brrr

They claim 5ns rise time

Insanely fast

[supple pollen]

Into what load? Some of those gate driver chips have interesting dies, with huge tapered transistors driving amperes of current

[woven bluff]

JLC charges $40 for Castellated Holes...

[rustic linden]

It's frustrating

[dry pelican]

It claims 10A. So you could probably calculate what the max capacitance for 5ns woukd be. But that rise time figure may just be the time it takes to switch the internal mosfets that switch the output.

[supple pollen]

PCBGoGo doesn't.

[woven bluff]

do you recommend that?

[proud siren]

Hey all , I have a weird question that Google isn't helping with. I have an adafruit feather nrf52840 and I bought the battery that comes with it (3.7v, 500mAh). The device works great when connected via USB. But when I plug in just the battery, it doesn't work, it just seems dead. While I am plugging in the battery, the neopixel and the led directly above the power flash red, and when I click it in snugly, everything shuts off.

The other extra weird detail -- if I start with both the USB and battery plugged in, then unplug USB, the battery continues to power it. Anyone know what I'm doing wrong?

[copper bolt]

What is a go to for learning Electrical Load balance ect.

[dry pelican]

The first issue sounds like a contact issue. The second one sounds like a possible battery charging or battery protection circuit issue

Weird

[proud siren]

Do I just need to get a new board? Or new battery?

[finite hemlock]

@proud siren volt meter the battery by itself check for 3.7-4 v them make sure the battery connector wires are in the right polarity as some aftermarket ones have been notes to be backwards

[proud siren]

Thanks for responding! I'm getting 4.23v from the battery itself. I got this battery with the device, so I would imagine it's got the right polarity? Also, it keeps the power on if the device starts plugged in to USB and then I transition to battery only. I think it wouldn't do that if the polarity was switched?

I tried swapping it anyway, and yeah, no luck.

I don't know if there's any value in this little detail, but while the battery is connected and it's not working (ie, I didn't first also have it connected to USB), if I press the reset button, it does the same momentary flash of the 2 LEDs in red. Is that some sort of error code? I tried looking it up but couldn't find anything

[supple pollen]

I've used PCBGoGo before, I have no complaints about their service

[finite hemlock]

@proud siren I think you're at the point where unless you want to trace the power rails and start testing voltage regulators you'd be better off ordering another board to test against. Refreshing bootloader and firmware would be a hailmary

[woven bluff]

PCBWay also charge $15 forCastellated Holes... and PCB Go Go cannot send email to my gmail address, probably because of the GFW

[distant raven]

OshPark allows castellated pads

[rustic linden]

Makes me want to run an experiment

[woven bluff]

does 1oz and 2oz make much difference?

[distant raven]

For current carrying ability, yeah

[woven bluff]

what is the most common?

[unreal flax]

1 oz is the default at most fab houses.

[rigid plume]

Hey folks, will this battery charger circuit and 5v vbus and battery switchover mosfet circuit work? Thanks

[supple pollen]

I consider 2oz as kind of standard (for exterior layers, 1oz for interior), but yeah, the default at most of the less expensive board houses is 1oz, which is fine for most uses.

[woven bluff]

They do, they just add afterwards after checking

[supple pollen]

That's obnoxious. I just did a quick check with their quick quote tool, didn't realize they'd pull that.

[woven bluff]

now I'm just adding this site to my ublock blacklist

together with rationalwiki

and the YouTube video recommendation column

[supple pollen]

And amazon, pinterest, dx, touchofmodern, etc.?

[woven bluff]

I use amazon, for appliance, grocery and AWS

pinterest is already there

hopefully I can move to the US for my postdoc, OSHPARK might be cheaper

[hushed smelt]

All the chinese fabs do that. It's only that cheap if it's a small one sided board. The more options you choose the more expensive it gets. Papa Johns does the same thing with their Pizza "deals" where it's only for a thin crust cheese pizza. Start adding options and voila you've gone from an $8 pizza to a $30 pizza...

[distant raven]

stares at my $200 cart in JLCPCB

[crude sedge]

Hey guys can you help me with this code pls?

[unreal flax]

For code you'll probably want to ask in #help-with-arduino or #help-with-circuitpython, or another channel depending on the platform you're using.

[crude sedge]

I need help with the adafruit_pcd8544 library for python....which channel do I go to?

[unreal flax]

Sounds like #help-with-circuitpython would be the best, then.

[crude sedge]

ok thanks

[rustic linden]

a mood, unfortunately

[woven bluff]

why OSH does not offer conventional solder mask color?

[supple pollen]

Since their business model is panelizing a bunch of peoples' projects onto big boards, it's more economic for them to just support a small number of options. They went to a lot of effort to get a particular æsthetic they like, and stuck with it. On the other side of the fence, PJRC used to make nifty looking black boards, but then discovered the traditional green soldermask had some technical advantages for them, so shifted to that.

[woven bluff]

that's the business model of every fab house... if they just want to maintain a single line, then green solder mask is the obvious choice

is something like this acceptable?

I'm not sure about the optimal length and width of the tab I should use

[dry pelican]

Sparkfun tested a lot of mouse bite sizes

They have an article

[supple pollen]

https://cdn.sparkfun.com/assets/home_page_posts/4/4/0/0/Mousebites_Whitepaper_Final.pdf

Reading that, I found that there are depanelling tools available to neatly separate boards at mouse bites with minimum stress on the board and leaving a nice finish

I had generally just used flush cutters

[vast flume]

I'm trying to communicate with a display driver over 8 bit parallel, but I am super confused. The datasheet hasn't been much help with explaining how the protocol actually works and I can't find anything online

I am just trying to turn on the display and then enable all of the pixels

But how do I tell it when the data I am sending it starts & ends?

[dry pelican]

I guess you just turn those pins on and off in a very specific order based on the timing diagram

And they're all measured in nanoseconds

uhhh

I don't know if you need to time it right

maybe just turn the A0, /WR, /RD, and data in pins to the right values and it'll work

what is this display?

It says 8080 series MPU and that's a really old processor

I think

[vast flume]

The display uses the ST7565R controller

[unreal flax]

It means an 8080-style parallel bus.

[vast flume]

Ohhhh

https://electronics.stackexchange.com/questions/20259/is-there-any-8080-series-parallel-interface-standard

Thx @unreal flax

[unreal flax]

You might find a SPI interface easier if the display driver supports it.

[vast flume]

If only I didn't already have the pcb 😅

But yeah def a big regret lol

SPI and I2C are natively supported while I have to figure out parallel from scratch lol

[shell heron]

Hi all! Hope someone could advise on this question.

I am working on a simple Audio Amplifier for a Handheld Gaming Console, with an 8 ohms speaker and a headphones jack, using the LM4875.

I am looking for a working amplifier, rather than high quality audio.

My question:

What kind of Capacitor could I use for Cout (100uF)? Tantalum (Polymer) or Electrolytic?

For the other capacitors, I am using Ceramic Capacitors (and around +/- 10% tolerance, 50V, X5R). I am looking Through Hole components!!

Thank you in advance for any recommendation!

[distant raven]

Probably an electrolytic. One thing to note that X5R is typically a temperature derating assigned to multilayer ceramic capacitors, usually surface mounted ones. I haven’t personally seen it for through hole components though

https://www.digikey.com/short/c44ddwpt

They tend to come in larger packages like this 1206 100uF X5R MLCC

https://www.digikey.com/en/products/detail/FK22X5R0J107M/1008928

I’ll be darned, there’s a through hole X5R

Hopefully that helps

🙂

[shell heron]

Thank you! What temperature coefficient would be more typical for through hole ceramic?

[distant raven]

What’s the power input you are expecting?

If it’s more than 5V you’ll need something different than that through hole I shared

[shell heron]

Yeah, your feedback is really helpful!
I found that last ceramic capacitor the your shared (FK22X5R0J107M), but the product page has the comment "Not Recommended for New Design). Not sure what that means.

[distant raven]

https://www.digikey.com/en/products/detail/KTD500B107M90A0B00/6189627

[shell heron]

Yes, the input is 5V. It will work with a Raspberry Pi Zero

[distant raven]

NRND means that they aren’t planning on doing any revisions to the design and it is approaching end of life

Through hole components are becoming increasingly rare as surface mounted components get more advanced and tend to be cheaper

[shell heron]

I see... I really wanted through hole components, to make it easier for people to put the console together, since it will be for game developers. But I might switch to bigger surface mounted components, instead.

[rustic linden]

I find it to be a bit of a shame, tbh, but I can understand it

[shell heron]

Or I will try an electrolytic for Cout, as you mentioned 🙂

Thanks @distant raven !!

[supple pollen]

I think the CS signal starts and ends a transaction

[vast flume]

I read it was either WR rising or CS, whichever occurs first

So I tied cs1 to vin 😅

[dry pelican]

You can get cheap kits full of good enough caps from Amazon or another online store. LCSC or another company may also have some caps.

[shell heron]

Thanks for the info. Yeah, I was just wondering it an electrolytic capacitor would be ok for the Cout (100uF). I will try that.

[supple pollen]

Since it's a single supply amplifier, a polarized capacitor like that should do

[shell heron]

Awesome! Thank you!

[regal lodge]

how's this?

[supple pollen]

I like the square pads denoting the ground pins (one of which has an extra-wide trace to it)

[dry pelican]

What is that IC?

[distant raven]

What’s the use of this board? I only have a minor quibble and it’s the fact that the USB isn’t centered 😵

But there’s probably a good reason for not centering

[regal lodge]

They all do but on an inner layer

[regal lodge]

Couldn't move it over without making the traces longer

[dry pelican]

Why 4L. It seems like you could do it with only 2L

Might also be good to move the high power traces to top or bottom since they can radiate more heat that way

[regal lodge]

Power handling mostly. Inner planes are power and ground

[dry pelican]

It they’re thick, they won’t radiate much heat, but I guess it’s still better to put them on the outer layers

[regal lodge]

There's lots of vias to sink heat

[dry pelican]

Yeah probably would be more work for not much gain to move the power traces

[regal lodge]

The paste covers them on front but you can see them on the back

[distant raven]

Is this true USB 3.0 or just 3.0 compatible PD? If it’s just compliant PD, length matching traces for data isn’t really necessary.

[regal lodge]

2 layer would require more layer changes

[dry pelican]

I think usb3.0 also requires stricter impedance matching than 2.0

So traces need to be certain width and be a certain distance to ground plane and stuff

[distant raven]

It does

Width is dependent on distance to reference ground

So ideally signal ground power signal

[regal lodge]

It is USB 2.0 and PD 3.0

[distant raven]

You could get away with not length matching on 2.0

Heck, even impedance matching is negligible for 2.0

[regal lodge]

You can but with it being a breakout I did it since whatever I connect it to still likely not be matched

[distant raven]

Unless it’s HS but most 2.0 is FS

[regal lodge]

It is HS

[distant raven]

480Mbps?

[regal lodge]

Yeah

[distant raven]

What chip?

[regal lodge]

This board is only doing PD

Trying to work out the power first

[distant raven]

Ah

[regal lodge]

In theory it can do up to 20V 5A (100W)

Hence, power plane... Lol.

[distant raven]

Personally, I’d do a bigger board with being capable of 100W

That’s a lot of power and you want much more thermal mass to sink heat too

[regal lodge]

I won't be doing 100W on this board. Just testing if I did the circuit right

So will do 20V but not 5A

I want to use this as a supply for a few different projects

This can even do source, not just sink, but I need to work out how to do that circuit lol

[dry pelican]

Coolest PD thing would be a bench power supply that uses PPS or leverages PD

Maybe also add some qwiic connectors

Bc everything needs qwiic

[regal lodge]

Lol. Easy to do tbh. Since 20V 5A is more than enough for most hobby stuff

I will OS it when it's finished

[cloud cliff]

Part of a datasheet for a full bridge rectifier. Is the max admissible load capacitor the maximum size of the smoothing capacitor or something else? And why does it go down with a rectifier that has a higher max voltage?

[supple pollen]

Yes, that's the smoothing capacitor. The base problem is the turn-on surge when the capacitor starts from a fully discharged state, and maximum current flows through the rectifiers to rapidly charge it. The limiting factor is generally power dissipation, so higher voltage rectifiers (which generally have a higher forward voltage drop) will dissipate more power for a given current, so have lower surge current specifications (and therefore smoothing capacitor size).

[cloud cliff]

Ok, so if my ac power source will only be able to deliver a constant ~30w (~6v input, so 5A), this won't be a problem even with a capacitor with a higher capacitance (22000uF)?

[supple pollen]

It will still be a problem at turn-on. You can use other techniques such as current limiting or decoupling, if you want a larger capacitor for some reason.

Alternatively, you can use a more robust bridge rectifier that can withstand the greater surge current.

[cloud cliff]

so this will be fine? https://www.partco.fi/fi/elektroniikan-komponentit/aktiivit/tasasuuntaussillat/10168-kbpc5010.html

datasheet: https://www.partco.fi/fi/index.php?controller=attachment&id_attachment=578

[supple pollen]

Yes, that thing should be fine with a large smoothing capacitor (in fact, I have a supply like that and it has worked well for many years). Happily, high current bridge rectifiers are not particularly expensive.

[cloud cliff]

thx

[idle python]

i need to learn more about pcb design

these sim modules run at the same voltages

why does this one have a polarised capacitor and a diode

but this one doesnt have a diode

they both have polarised capacitors but both have a different max voltage and farad rating

[blissful frigate]

The zener is clamping the VBAT pins on the module to its zener-voltage

its a voltage reference/simple overvoltage protection

[idle python]

i understand that but why do the capacitors have diffferent farad ratings

nvm i looked at the datasheet

i think its what the manufacturer recommends

[distant raven]

To obtain a specific value

TI recommends doing this for their products as well. The closer to the calculated value, the better. Space be darned 🙂

[rustic linden]

Someone help me I keep wanting to re-route my board to make it forever smaller

[supple pollen]

I can't leave my boards alone either

[rustic linden]

glad to gear I'm not alone

I've already made 2 full rerouting revisions and I keep having to tell myself perfect is the enemy of the good

[distant raven]

I have an obsession with making smöl boards

[dry pelican]

https://www.sparkfun.com/categories/tags/smol

[rustic linden]

Same, I always want to make things smaller

[unreal flax]

I sometimes run into the situation of spending time making a board more compact, but I am aways viewing it at like 1000% magnification on a big monitor. So when I finally get the board fabricated: "Wow, this is so much smaller than I thought!"

[distant raven]

I’m not sure why I’m so surprised how small feather sized boards are when I get the PCBs in the mail

Every time lol

[waxen halo]

I'm designing a power supply frontend (ATX PSU provides actual power) and I wanted to have my own power good signal that trips if any I detect any overcurrent. Does it make sense to just run the ground of these outputs through a mosfet that's on as long as that signal is good or is there a different way I should approach that?

[supple pollen]

If there's just one voltage, that seems like it would work, but if you're using an ATX PSU, you can just shut down the supply with its enable signal

[waxen halo]

I don't want to shut off the whole thing, just the outputs

I also have a touchscreen / MCU integrated. I guess if the stuff I want to keep on is just on 5VSB that might work.

Though I did just discover that dedicated power switch ICs are a thing and that's looking like a better option.

(Yeah there's just one voltage, in this particular case I'm providing monitored USB ports since that's how most of my projects are supplied power anyway. I have a separate circuit for an adjustable voltage supply)

[supple pollen]

Ah, the power switch ICs can help with overcurrent detection too (I'm looking at some for an IC tester)

[waxen halo]

I've already got a current monitor IC for that (and so I can show the consumption on the display), the alert signals from it are what will bring down the power good signal.

I was struggling to work out the cooling requirements for the mosfet as a power switch, the power switch ICs seem a fair bit more straightforward

[supple pollen]

Normally a MOSFET used like that will either be fully conducting (in which case it's very low resistance and won't dissipate much power) or off (in which case there's no current flow so no power dissipation).

[waxen halo]

In other words for that usecase especially I should be looking at the charts for rds on instead of the typical?

[supple pollen]

I'm not sure which typical you're referring to, but there is normally a Rds on graph version gate drive, so the the value for the part you're looking at, at the gate drive voltage you'll be using is what you're after (many MOSFETs don't conduct fully until there are 10-12V of gate drive, don't be misled by the "threshold voltage" where it just begins to conduct)

[waxen halo]

I guess more of a max than a typical. The datasheet I was looking at listed an RDS on max at voltages right at the top. I figured using the worst case scenario was a safe way to design it, but it always seemed to require external cooling.

[supple pollen]

Yes, I tend to use the max value for figuring heat dissipation. However, modern MOSFETs routinely achieve <100 milliohms, so unless you're looking at some serious current, dissipation shouldn't be too much.

[glass wolf]

Hello everyone! I am totaly new on pcb design. What i am trying to do basicaly an arduino pcb with bms i've found tons of schematics but since i am newbie on this i didnt understand much.
For example here 2 pics and i didnt understand what is DNP /CE meaning and i couldnt find any component number for U6 which is on first picture.
I really appreciated all helps
Thank you!

[woven bluff]

is this considered flyback transformer?

[supple pollen]

DNP is "do not populate", normally that provides pads on the board, but no component is placed on them, they're just available for use later if needed. "/CE" is composed of two parts: the CE part usually means "chip enable"* and the "/" means "not", "inverted", or "active low" (there are several ways to express this, like a bar over the signal name or sometimes a ! instead of a /).

• I'm unsure why that design has a "CE" signal connected to the "TS" pin. Perhaps they're (mis)treating the temperature sensor input as a way to shut down the charging chip? I don't kow.

[supple pollen]

That does not look like a flyback transformer to me. Normally transformers designed for flyback use have a (topologically) circular core with a small gap in it, magnetic energy is stored in the gap, then when the current is shut off, that magnetic energy is quickly dissipated with the effect of boosting the voltage (this is the "flyback" referred to, for historical reasons involving raster scanning CRTs). Since that transformer appears to have a rod core, I would not consider it a flyback design, as there isn't a closed magnetic circuit with a gap.

[woven bluff]

I'm considering this transformer for my arc lighter project, but I'm not sure how to properly drive it (not the crude blocking oscillator in the picture ), while keeping footprint minimum.

since it's not flyback, SMPS chips are out of question

555 with a good logic level FET seems reasonable, but 555 require too many passives

also the whole thing is powered by lipo, I'm not sure if logic level FET will operate properly at 3.4V

[supple pollen]

Some SMPS chips work fine with non-flyback circuits. You're right to be wary of "logic level" FETs, as some are billed as such because they have a low threshold, but the threshold is not the important figure, what you really want is a low on-resistance with a low gate drive voltage.

I have a personal fondness for the LT107x and LT117x series of multiple topology, current-mode switching control chips. They require few external parts and the current-sensing approach can be used to avoid core saturation without complicated feedback circuitry. They don't work quite like traditional switchers, but Jim Williams wrote a raft of very good application notes on how to use them.

[woven bluff]

this is a 20V/30A MOSFET

[supple pollen]

I find the on-resistance graphs more useful than the "drain current" ones

[woven bluff]

5mOhm at 3.5V, approx from the graph

[woven bluff]

they don't have that graph

[supple pollen]

That one looks like it conducts some with 3V on the gate, but would presumably be dissipating a fair amount of power, while it would be solidly on at 3.5V. Depending on how the protection circuitry is set, you could have issues when the LiPo starts to become discharged.

[woven bluff]

I'll have UVLO at 3.4V

[supple pollen]

That MOSFET should be fine then.

I've seen some with UVLO as low as 2.9V.

[woven bluff]

may be add a first stage boost with commercial module?

[supple pollen]

Possibly: it's easier to design circuitry around a known voltage. However, the power draw of most circuits like this is a spiky mess, and some commercial modules (especially the cheap ones) will have trouble supporting a load like that.

[woven bluff]

LT117x are regulator with build in switch, but does not have primary-side sensing

[supple pollen]

Some designers take a split solution, and run the heavy load directly from the cell but have a boost circuit to provide just the gate drive voltage.

[woven bluff]

considering the secondary side is plasma arc... I don't know how to control feedback

[supple pollen]

They absolutely do have primary side sensing, but it's current sensing, not voltage sensing. This is a different kind of sensing, and gives different abilities. For an arc lighter, you generally don't need primary side voltage sensing for anything.

[woven bluff]

so what should I feed into FB node, which controls duty ratio?

[supple pollen]

Since you don't need to regulate anything, you can just feed a fixed voltage to the feedback pin. Then the duty cycle will be controlled by core saturation, which will give you the maximum power through the transformer for a given frequency.

This app note (cheekily entitled "Switching Regulators for Poets", implying it's accessible to people who aren't EEs) https://www.analog.com/media/en/technical-documentation/application-notes/an25fa.pdf includes this circuit which doesn't use the feedback pin. They're versatile chips, and you don't have to use every function they provide (such as regulation) if you don't need to.

Thinking on it further, I realize that it ends up basically being a more-integrated version of the blocking oscillator (which is also ultimately controlled by the transformer's core characteristics). Which approach you choose basically boils down to personal preference (which presumably includes factors like size, cost, and component count).

[dry pelican]

I'm designing a "locking PLL" that can lock its frequency when there is no input signal. This is for driving Tesla coils that need a signal close to secondary resonant frequency to start and perform best at resonance. Since the coil will be interrupted (turning on and off with an external signal), the PLL will go back to its original frequency whenever there is no feedback signal because the coil is off. I guess this will lower performance because the PLL has to adjust its frequency back to resonance every time the coil is re-enabled. This circuit should solve that by keeping the VCO at the same frequency whenever there is no feedback signal. That way, the signal fed into the coil will be at or near resonance whenever it restarts. Would this idea work?

[woven bluff]

I never thought about this improtant question, in constant frequency current mode SMPS what does feedback pin actually do?

[dry pelican]

I've thought about this design, and I think it'll just be better to use the interrupter to disable the VCO adjustment instead of the comparator and FBR thing that may not be able to shut off in time.

[vast flume]

Hey guys, 2 quick questions:
I want to use this display https://www.orientdisplay.com/wp-content/uploads/2021/08/AES200200A00-1.54ENRS.pdf, but it is not made clear if it has an onboard SSD1681 driver or if I need to have my own. Also, I can't seem to find where the refresh rate would be?

[proud siren]

Does anyone know where I can buy a push button that's fairly big (ie, like the size of the surface of someone's thumb) but also quite flat? For instance, these (https://www.amazon.com/gp/product/B07SVTQ7B9 ) are a nice button size, but it's so tall. It's almost as tall as the feather board is wide, which makes for an unnecessarily cumbersome package. I've been searching on Amazon but can't find anything that seems right. Do they not make big, flat buttons?

[supple pollen]

Basically it adjusts the pulse width/duty cycle. https://www.analog.com/media/en/technical-documentation/application-notes/an19fc.pdf page 7 describes how the chip works

[supple pollen]

It sounds like you'd want a gated sample-and-hold sort of circuit in the PLL tracking loop.

[supple pollen]

One approach is to use an ordinary small tactile switch internally, and then have a large button actuate it.

[proud siren]

Thanks for the response. I guess I might not understand what you mean. My problem is that every "big" button I find is unnecessarily tall. So if I were to put the PCB in a box with the button, the box has to be really big to contain that button. If I had a small button on the board, and a big button that actuates it, I still need to have room for the big button in the whole package

[supple pollen]

Basically, the button itself is a flat round plastic piece that rests on the actual small switch that does the electrical switching. Kind of like this https://hackaday.com/2019/07/06/3d-printed-buttons-printed-as-a-single-unit/

[proud siren]

Ohhh, I see what you mean. Thank you, that makes sense

[vast flume]

As some of you may remember/know, for the past year or so (through multiple big design changes) I've been working on creating an altimeter for skydiving. Started off as an esp12f powered board that just had some lights and sensors to now being powered by an stm32, having lights sensors and a display. I made a mistake on the latest iteration and used 8 bit parallel without knowing how terrible of an idea that was. There were also some issues I made with the power circuit for the display, as the datasheet was terrible. So pretty much, my question is, what display/driver would you guys recommend for me to use? I don't need a fast refresh rate at all, 1hz would be fine. The display I currently have is an ftsn lcd but I also started looking into epd's.

[dry pelican]

Ssd1306 OLED displays are versatile and low power. They're also cheap

[supple pollen]

I'm thinking eInk or maybe one of those Sharp memory displays might make sense. Or possibly one of the Ynvisible ones if they're available with the capabilities you want

[dry pelican]

Also some sort or SPI LCD from <online retailer> would work

[dry pelican]

That's exactly what I was going for with the bilateral switch that connects the low pass filter and the phase comparator output. I think that I'll just use the interrupter output that switches the coil off to also switch the bilateral switch because the VCO frequency will lock immediately instead of locking after a capacitor discharges beyond a certain threshold (which could cause the frequency to drift out of resonance).

[hearty ocean]

Hi all, I'm working on a project that needs a PN2222A transistor to activate an LED on GPIO (using circuit below), however these KSP2222A transistors (advertised as PN2222: https://www.adafruit.com/product/756) aren't working the same as my PN2222A -- the LED is about half as bright when triggered. Is there a major difference between these two transistors that I'm missing?

Edit: Looking at the datasheets, the only major difference seems to be collector current (1A for PN2222A vs 600 mA for KSP2222A). Would lowering the 330 Ohm resistor solve this?

[distant raven]

Make your 10k like a 5k

You likely need more current to saturate the base.

[supple pollen]

Note there are some variants (like P2N2222A) with different pinouts, and transistors will have less gain hooked up the other way around.

[dry pelican]

Maybe also decrease the 330 ohm resistor since NPNs have a voltage drop of about 0.6v across the collector and emitter

But madbodger is probably right with the flipped pinout

[woven bluff]

It did not say what FB does when not pulled to ground

I don't know how error amp would behave when one input is floating

looks like current amp does not directly control the driver

so, when FB is not needed, Vc should be tied to something static so current should get exclusive control?

not sure what to do with the third error amp that compare VIN and primary side ringing

[woven bluff]

$200 went to OSH Park, this really is an expensive hobby...🙃

[cloud cliff]

I have a moped with a dynamo that supplies "6v" AC to the lights, that i want to install 12v led lights on. I initially thought i could just get away with a full wave rectifier and a large filtering capacitor, but that now seems quite... expensive

I measured the real voltage at the lights, and it came out as 4-50V rms depending on the engine rpm. If i understand correctly that means that with the slight losses of the full wave rectifier i'll get nearly 70V dc at the output at full rpm. I also calculated that with at idle the power coming from the dynamo with the rectifier will be ~50hz, so with the 4v(ish) 1A im drawing, i'd need a ~30000uF 100V cap, and that's expen$ive (3*50€) (and also still a full volt of ripple)

Also the voltage regulator that i planed on using only goes up to 50V

is there any more reasoanable option for converting ~50-130hz 4-70V choppy DC (from the (large, 50A 1000V) full wave rectifier) to a smooth 12v DC (drawing 6W, though i would also like to add some supercaps so the lights will glow after i turn the moped off, so i might momenteraly draw ~20w)

[cloud cliff]

If i could somehow make the input always be at ~25V AC i could just us a ~4000uF 50/63/100v cap and be done with it

[spice turtle]

that seems to be quite a variance. 4VAC? You wont need that large of a cap actually. I think I had heard that a good rough estimate is 10,000uF per Amp. You can get away with a lot less. You also dont need one single cap, you can get away with multiple ones in parallel.

You could use a pre-regulator to drop it down even more, just a simple zener diode with a transistor for power before your main regulator.

https://www.bristolwatch.com/ele/zener_power_supply.htm

[supple pollen]

Maybe a simple circuit to clamp the voltage to 50V maximum or disconnect the input when the voltage rises too high. However, if you're considering a bridge rectifier and capacitor "expensive", then stuff like preregulators and clamping circuits may be prohibitive

[spice turtle]

🤔 you know, they could technically use a 12VAC transformer too! Might be bulky but they can get them on ebay or digikey

[vast flume]

I am trying to read from a bmp581 over i2c with an stm32, however when I read from the fifo buffer the read data remains in the buffer, even though new data is appended. Once the buffer fills up new data stops coming in. Also, the registers for reading the current temperature & pressure data values are not getting new readings either. Has anyone else experienced this? I can't for the life of me figure out what I'm doing wrong

[unreal flax]

How are you determining that the data remains in the FIFO? Do you get the exact same bytes on subsequent reads? Is there any chance that the second read is actually failing and you're just looking at the same buffer in memory again?

[vast flume]

I'm using the debugger to inspect the local memory on the stm32 after reading from the sensor. I'm getting the same bytes on subsequent reads, with new dataframes appended after until the fifo buffer is full (at 15 dataframes which lines up with the datasheet)

What do you mean by your last point though? That could possibly be the issue honestly

Also another thing that has been strange is I seem to end up also reading the register that stores the count of stored dataframes in the fifo if I try to read 24 bytes (4 dataframes?) at once. Very confusing since the register for the fifo count is well before the fifo data buffer. I was thinking maybe I was having an int overflow somewhere but I think I've been able to rule that out

[unreal flax]

Hmmm, well if you're seeing new data showing up, that indicates something is happening. How are you choosing how many bytes to read each time?

[vast flume]

I configured the fifo to store temperature & pressure which would be 6 bytes according to the ds

So I'm just reading it like this:

uint8_t temp_arr[6];
BMP_Reg_Read(BMP_FIFO_DATA, 6, temp_arr);

void BMP_Reg_Read(uint16_t reg_addr, uint16_t reg_size, uint8_t* data_output)
{
    HAL_I2C_Mem_Read(bmpConfig.i2c_config, (uint16_t)(bmpConfig.address<<1), reg_addr, reg_size, data_output, reg_size, 100);
}

[unreal flax]

Something is a little weird in the Read() call. The two reg_size parameters have different uses. The first should be the size of the the address, I think, which would be 1 byte.

[vast flume]

oh

hmm

So theoretically an address could be multiple bytes long?

[unreal flax]

Yeah, if there's more than 256, you could see a 16-bit address sometimes. I can't think of an example off the top of my head, though.

[vast flume]

Ah interesting, didn't know that

Trying now with it set to 1

omg

I can't believe that was it 😭

24deg c 😎

Thank you so much!

[unreal flax]

Great! So what was probably happening is that it was sending a bunch of 0x00's after the address, so you were really reading registers starting at 0 instead of the FIFO, I'd guess.

[vast flume]

Yeah that makes total sense now. Was wrapping all the way around such that it somehow was 24 bytes before the fifo count register lol

Crazy coincidence

[unreal flax]

Yep, that matches. It's address 0x17, so it would be the 24th byte read if it started from 0x00.

[vast flume]

Omg

Well thank you very much!

[cloud cliff]

10000uF is a good estimate if i'm getting 4V 60hz (120 trough the rectifier) electricity, but i'm getting 25hz(×2=50hz) at the input, so even a perfect world the 10000uF cap will give me 3V ripple (that's probably fine for the lamps, but idk if the voltage booster/regulator i'm using will handle that)

@supple pollen
With all the supercaps and voltage regulators/current limiters the price would be creeping north of 200€ on a moped that cosy me 500€ lol. A simple voltage clamp would probably be fine atleast price wise (i can't disconnect the input because then i wont have lights while going 40kph down some dark backroad at 1am), but i'll still have the problem of the voltage being too low at idle. Is there a sinple circuit that progressivly changes the current going to one component smoothly to another when the voltage rises (and disconnect the 1st component when the voltage gets wayy high). I could then use some cheap 30000uF 35V cap(s), and a smaller higher voltage cap when the engine rpm and voltage rise (5000uF 100V)

[woven bluff]

is Q Package 5-Lead Plastic DD Pak the same as to-263?

[spice turtle]

Where are you getting 3V ripple from? I can tell you it will be no where close to that. Maybe 30mV, but not 3V. Usually voltage regulators can handle ripple, maybe not 3V but they can probably handle 100mV or so

[supple pollen]

What current draw are you using to get these figures?

[cloud cliff]

I/(2•ACfrequency•V)

So 1.5A/(2•25hz•1V)=0.03F=30000uF

[cloud cliff]

4V at idle, so 1,5A (6w)

[supple pollen]

Let's back up a step. Where are you getting 6W?

[cloud cliff]

Basically from nothing, the leds that are available for the ba15s bulb socket are pretty inefficient, and fairly bright. The tail light doesen't draw that much, maeby a watt. I know how 4w of led light looks like, and those bulbs i have in my home are for sure more efficient and less bright than the ones i have fitted on my moped.

[spice turtle]

I really think you'll be OK with such a small load. LEDs can handle ripple, in fact some converters basically forgo a rectifying diode all together and just feed them pulses.

[cloud cliff]

Oh yeah they'll be fine. I just don't want them to flicker. Leds are technically illegal to fit on a moped that isn't aproved for them (though nobody really cares), so i'd like them to look like incandescents as much as possible

I could just string a series of batteries together and measure the power consumption though

[supple pollen]

I think your ampere calculation is off, and you can get by with much less capacitance

[cloud cliff]

This is off or you think they don't draw 6w?

[supple pollen]

I think they only draw 6W at full voltage and less at lower voltages.

Additionally, where are you getting 60Hz?

[cloud cliff]

more like 50, the moped idles at ~1500rpm (rotations per minute), and 1500/60sec is 25. So the engine rotates 25 times per second, and with each rotation of the engine the ac cycle goes down and up one time. So trough the full bridge rectifier that's 50 peaks a second (yeah not technically ac 50hz, but 50 peaks a second)

[cloud cliff]

i'd like to get the full power from the leds, so it would be ideal if i could use them at full voltage, so i can like, see infront of me more than three meters

[supple pollen]

So you want full power at low voltage? The easiest way forward is probably a small battery.

The 12V 7Ah batteries used for emergency lights are inexpensive and common

You could probably also use lithium technology, but for such exposed to outdoor condition use, I tend to stick with old fashioned reliable SLA.

[cloud cliff]

no i want full power at full voltage on the leds, when the moped idles. One solution that i have already come up with is to use bigger lower voltage caps when the moped idles and somehow to switch to smaller higher voltage ones when the voltage increases. I just don't know how to do that.

[cloud cliff]

yes, but the i'd have to remember to keep it charged. It's also heavy and i don't really have any space to fit it on the moped

[supple pollen]

You wouldn't just charge it from the existing alternator? Basically, it would get topped off when you're running above idle, and drawn from when you're at idle.

[cloud cliff]

with ac?

the moped doesen't have a battery, i'd ofcourse be taking the power from that if there was one

[supple pollen]

No, you'd still rectify it

Yes, I'm suggesting adding a small battery to the moped.

[cloud cliff]

and i'd still need to regulate it, lead acid batterys don't probably really want to charge at 40v

[supple pollen]

Yes, battery charge regulators are common parts for bikes, mopeds, tractors, and the like. It wouldn't really charge at 40V anyway, the battery would drag down the voltage as it absorbed charge. How it would balance would depend on what type alternator you have.

[cloud cliff]

the alternator coil that goes to the lights goes ONLY to the lights and nowhere else

[supple pollen]

I assumed you were going to reroute it through a rectifier anyway?

[cloud cliff]

yes

but i'd still need to fit the battery somewhere

[supple pollen]

You'd have to do that for the capacitor anyway...

[cloud cliff]

yes, i have small creavices where i can 3d print multiple boxes and put diffrent parts of the rectifier/regulator there

A battery is just a big hunk, i have nowhere to fit it

maeby on the rear luggage carrier with cargo straps or something, but i suspect the police won't really like that

[supple pollen]

An NP-7 is about the same size as a big capacitor, and there are plenty of smaller ones available

One of those, and a regulator like https://mechanicsurplus.com/products/3597489125 and you have a nice DC supply of 12-14.4V for €30 or so.

[cloud cliff]

and is thet regulator happy with the 50V the dynamo will push when not being drawn power from?

[supple pollen]

Yes, it's designed for exactly that sort of use. It also includes the rectifier, so it's a one-part solution.

[cloud cliff]

well i'll have to vandalise the cables a bit to make the work but otherwise it's fine.

So if the lamps draw power to the lamps even when it's being charged trough that battery, they'll still get a flat voltage?

[supple pollen]

You'd have to do that anyway, I'm guessing. The battery acts like an enormous capacitor, but with less voltage variation. So when you're revving the engine, you'll get 14V or so out of the battery. At idle, or with the engine off, you'll get 12V or so.

[cloud cliff]

Ok, so then i'll just buy one of those. Just have to remember to net leave the lights on because the battery will go flat in less than 3h.

[supple pollen]

Exactly.

There are ways to automate that if desired (basically when the engine kill switch is actuated, it shuts off the lights)

[cloud cliff]

the moped doesen't have an engine kill switch

[supple pollen]

How do you shut off the engine? Just turn off the fuel supply?

I found this handy wiring diagram at Revival Cycles

[cloud cliff]

no, it's got a decompression valve

[supple pollen]

Ah, that's interesting. You could probably add a switch to that if you wanted, or just remember to turn the lights off.

[cloud cliff]

"OEM fuse", lol, no fuses

[supple pollen]

It's not like in a car, where you can't start the engine if the battery is dead anyway.

I like to have fuses to protect my wiring (the battery can push a lot more current than the alternator if there's a short circuit). But that's a personal choice.

[cloud cliff]

Well yea i'll probably add one

i don't mean that someone took it out, it didn't come with one from the factory

[supple pollen]

Yeah, some of these things are pretty bare-bones.

[cloud cliff]

oh this is way tighter than i remember it being, i'm gonna need a smaller battery (toothbrush for scale)

like a 3x4x15cm would baeby go in there

oh not even that, more like 3x2x15cm battery

and that's pushed right up to the engine

i would go lithium, but if my mounting is bad and it falls ontop of the engine i'm gonna have a bad time

[supple pollen]

Yeah, you were never going to fit a 35000µF capacitor in there.

[cloud cliff]

yeah but wait what am i doing

i have a device that turns whatever volts ac into solid dc (battery charger)

i can just use supercaps

(and an extra charge/voltage regulator)

also 3•10000uF, and that defineatly does fit

oh yeah no supercaps are really expensive

[vast flume]

Well if anyone wants a library for using the BMP581 with an stm32, I made one which can be found here: https://github.com/kkingsbe/STM-32-Libraries/tree/main/BMP581

[rapid garden]

is it just me, or does this filter seem to be more resonant at negative voltage than positive (bottom graph)? why would that be?

for context it's a switched cap vcf using noise + comparator instead of pwm. resonance path has diode clipping to prevent self-oscillation

[supple pollen]

It may just be hitting the rail

[rapid garden]

i found the cause, turns out even though JFETs are "symmetrical" they ever so slightly bias

fixed it by putting n-channel on the first stage and p-channel on the second so their biases cancel out

[glass wolf]

Hello everyone, i am trying to learn pcb design by myself and using easyeda for design.
Any idea why this cap and also other caps which is connected to VDD their footprints both on GND
Any suggestions?
Thanks!!

[spice turtle]

that seems like an error 🤔 check the signal name for that cap on both ends

[hushed smelt]

@glass wolf Where did you get the capacitor footprint from? Are you using one drag and dropped from the Commonly Library or is it perhaps a custom user footprint?

[hushed smelt]

Is it giving you any net flag errors in schematic view when attempting to "update PCB"?

something like this?

If you don't have a pin header or other kind of pin attached to the 3V3 net it will throw an error about incomplete net. You can get around that by giving it a pin header like this.

no errors

Check your net label for the individual GND pad that is on the incorrect net.

Relabel the Net with the correct net you want. In this case 3V3 and it should automatically connect back to 3V3 net.

If it changes back to GND after a schematic update then the problem is in your schematic not PCB view.

In which case you need to figure out why 3V3 net is changing to GND net. I did try it myself and found some odd behavior. Possible bug.

Do not ignore netflag errors. They're as important as DRC errors but for schematic view.

The problem, I think, GND is a valid net and so is 3V3 so if it's switching to a different valid net that's a sneaky kind of bug and might be something to report to EasyEDA in their forum.

[narrow jetty]

better version

[dry pelican]

Looks good. It could be compacted more though

Maybe put the ICM20948 under the feather and mount it on the other side

Make sure to make the pinout work if you're doing reverse mount of course

[narrow jetty]

thank you i will work on making it smaller in mk2/3 i will try about putting the icm under it

[proud siren]

I have a series of naive question as I have no industry experience in this area. I have a working prototype of a product using the Huzzah esp32 board from adafruit.

1. If I wanted to hypothetically mass produce the prototype I made (ie, 100 units), would it be cheaper to design a custom PCB that only has the components I need (microcontroller, some input pins, BLE chip)? Or is it roughly the same cost as just buying a bunch of these feather devices?
2. The places that make your custom PCBs -- will they also write my firmware to each device or is that on me?
3. In general, when someone is making a new electronics device for the purposes of selling many units of it, I imagine they don't buy a bunch of boards and then solder on the buttons manually, screwing each board into a case, etc. Who provides the surface of assembling an electronics device once you have a designed PCB and firmware?

Apologies if these are obnoxious, I wasn't sure how to google this.

[fervent lance]

1. it depends, sometimes you can get cheaper designs, sometimes you don't, considering only hw cost and not devolopment
2. you should ask them but most do
3. manufacturers such as jlc do often provide assembly and other services (just dm them if something is not included, they might accept your porpose), some other people although prefer manually soldering their stuff due to cost, or not dealing with knock offs or whatever

[proud siren]

Wonderful, thanks so much for these answers!

[bright cape]

friends... I'm beating my head against a stick on this one....
reed switch is used to swithc the system on, then the PA6 of the MCU should pull the gate of the PMOS low and keep the VBAT flowing to VDD until the MCU goes to sleep and put PA6 HIZ Z (thus, cutting power to the entire MCU). Theory is good, I think.
but the MCU comes on regardless. I thought maybe it was current going through R1 INTO PAD6, and through ESD cell or something?
I increased R1 to 100K, no change.
I even removed R1. no change

[supple pollen]

I'm guessing the protection diode on PA6 is conducting, because when Vdd starts to drop, PA6 becomes higher than Vdd.

[bright cape]

i figured it was something horrible like that. I might either switch out Q1 for a proper load switch, or add an NMOS between PA6 and the gate of Q1 to pull it down properly. that'd likely isolate VBAT from PA6.

i'll have to invert the output of PA6... no biggie

thank you @supple pollen - you gave me some good stuff to think about

ok - i cut the trace between PA6 and the gate -- and the board doesn't come on!

(tha'ts a good thing)

and, by grounding the gate, it switches on beautfully.

[rigid fiber]

EasyEDA PCB is putting a nunch of through holes underneath my ESp32-S3 chip, in the 3x3 grid of ground contacts for the chip. But this means that the centre contact can't be routed to anything because the usual 0.254mm trace conflicts with the design rules by getting too close to the through holes. Do those through holes really need to be be there underneat the esp32-s3, or can i remove them?

[rigid fiber]

Hmmm... ot looks like they were put there by the footprint of the part. Can't remove them.

[dry pelican]

You can edit the footprint and move or remove the vias if you want to

But usually you want those vias for a solid connection to ground and heatsinking iirc

[supple pollen]

Yeah, they don't have to be connected, but normally they're connected together anyway.

[lone basalt]

I'm trying to wire up a RPI 2040 in easy eda, I'm having issues with track width on the pads, if I make DRC happy I can't connect every pad, but if I shrink the track so that I can, DRC gets mad, what should I do? I've found where to edit the design rules, and what JLCPCB's minimum is

[ocean coyote]

has anyone messed with pio usb? on the pico

im looking to connect a st7735 lcd with sd card slot, and 5 buttons alongside a female usb a port setup as a host to relay packets through

i need the micro usb set up as a usb device setup to appear the same as the device im plugging in

the issue im having is im currently using the pins I believe are associated with pio usb

https://cdn.discordapp.com/attachments/1036181426581815377/1099115857445138472/image.png

also being a pico w, I think that the wireless chip uses spi on pins 23 24 25 29

does that interfere with the spi busses ive chosen for the lcd or sd card?

[latent jungle]

The native USB signals are not broken out to the edge pins.

And yes, one of the SPI ports (SPI1) is dedicated to the WiFi module. (You can use the SPI port if you are not using the WiFi module.)

There are two available on the RP2040 / Pico

Page 13 of the RP2040 datasheet has a table that shows the muxing of the pins and various functions: https://datasheets.raspberrypi.com/rp2040/rp2040-datasheet.pdf. (Or the pinout diagram https://www.raspberrypi.com/documentation/microcontrollers/raspberry-pi-pico.html.) Use that to determine which pins you should use to access SPI0.

[ocean coyote]

sorry I should have linked this https://github.com/sekigon-gonnoc/Pico-PIO-USB

I believe its also integrated into tinyusb now, so the plan is to use tinyusb and this pio usb to have a pass through as such

[latent jungle]

where on your diagram is that indicated?

[dry pelican]

Does Adafruit have the schematics for those pico-based brains boards that are used for testing newly manufactured boards?

Those use the pico as both USB host and device

[ocean coyote]

its not yet, but its going to use gpio 0 and 1

[ocean coyote]

good point

[dry pelican]

And also the software

Hopefully it’s also open source

[ocean coyote]

do you remember what there called?

[dry pelican]

But it is understandable if they don’t want to open source their internal testing things

They’re just called brains boards iirc

Maybe someone in this server who works at Adafruit has info

[latent jungle]

No worries. My bad for wasting my time on a question not actually documented.

[ocean coyote]

found it https://www.adafruit.com/product/5723?utm_source=youtube&utm_medium=videodescrip&utm_campaign=newproducts

[dry pelican]

I forgot that that one just released lol

[ocean coyote]

was there another?

[dry pelican]

It was featured in their videos

But it was internal

That feather will work

[ocean coyote]

do you think id encounter issues with current without a battery attached?

the usb devices i will be plugging in draw around 300ma maybe more

[dry pelican]

No. Shouldn’t be an issue if trace thickness is at least 0.5mm

And usb is spec’d for at least an amp iirc

[rigid fiber]

This spec sheet from JLCPCB is all in Chinese. Is that a 5 pin relay just with doubled pins to control multiple items?

Oh fudge, I just realised that I was concentrating so hard on getting to the PCB stage that I forgot relays in my fan project. As it stands now there no way to turn them off, only slow them down. D-:

[hushed smelt]

How to control 12V RGB LED's for a shop lamp from a 3.3V feather.? I'd prefer to control the RGB's from a feather even though they'll be 12V RGB's

The lamp is currently 2 rows of 5V white & yellow LED's powered via 5V USB. About 60 white & 60 yellow tiny LED's.

I'd like to make a 12V custom ring PCB with 12V RGB's for a much brighter shop lamp.

How to control the dimming, on/off, using a 3.3V feather as the controller? Is that possible?

[unreal flax]

Are you planning to use 12V smart LEDs, or regular LEDs that you are controlling via PWM?

[supple pollen]

No, it's an 8 pin relay. 2 pins for the coil, and 3 pins apiece for two sets of form C contacts.

[hushed smelt]

probably a matrix of 12V RGBW's with built-in driver chips in each LED. I don't like dealing with external driver chips.

Will try to use much bigger LED's maybe the 4mm variety or whatever the big ones are.

[unreal flax]

You might need a level-shifter to go from 3.3V to 5V for the data line, then, since I think the 12V varieties tends to have higher logic levels, but that's worth checking in the datasheet of the LED you pick.

[hushed smelt]

good point I haven't even started looking at data sheets. if i could control 12V LED's with 5V that would work out so much easier.

[unreal flax]

You should be able to. It'd be rare for them to need 12V logic.

[hushed smelt]

Can't find any 12V RGBW 5050's with built-in I2C 😦 Seems like they all require separate driver chips.

[supple pollen]

Most of the drivers I've seen either use a timing based one-wire protocol, or a synchronous bipolar protocol (like SPI, but unidirectional). I haven't seen a synchronous unipolar one like I2C used.

Individually addressable 12V LEDs would be very inefficient anyway. Most of the 12V addressable strips use separate LEDs in groups of 3.

If you were going to run R, G, and B LEDs at the rated 20mA from a 12V source, you'd be dropping 9V apiece for the blue and green ones, and 10V for the red one: over half a watt of wasted power to dissipate! A 5050 package wouldn't last long under those conditions.

[dry pelican]

Yeah. Might be better to put in a buck converter to 5v so that neopixels can be used

[wild glade]

Does this look compliant?

Should I keep the P1x.x in the name? I've added those as a alternate pin names

Same for ALE/PROG, PROG is added as alternate definition with input set as type

[hushed smelt]

After looking at 12v strip designs i realized its all just 3v or 5v led’s run in series. I had no idea it worked that way. Why would anyone do that instead of running them at 5v? Save money on resistors? Makes no sense when i2c leds are so much easier to deal with.

[wild glade]

(It's KiCad, by the way)

[hushed smelt]

Like, it doesn’t make them any brighter in series at 12v. I don’t get it. Maybe at one time the price difference made it a more cost effective way to do it?

Think ill just go the neopixel route and make everything easier. Thank you for the advice. 🤗

[wild glade]

Also, how would I deal with the alternate pin numbers of different footprints?

[supple pollen]

I normally put the signal name on the symbol and only tie it to a specific pin when combining a symbol and footprint into a device.

[wild glade]

Fair, designing it for DIP so I'll just keep it as is

[unique patio]

I think 12V is a semi-standard voltage for low-voltage applications: lots of power supplies available, etc. Also you can use thinner wires because the current is lower

incandescent 12v bulbs were very common, so maybe originally these were drop-in replacements.

[hushed smelt]

That makes sense. Can get away with longer & thinner wires with lower current. Thank you.

[supple pollen]

However, you can only get the lower current if you run multiple LEDs in series, which means they aren't individually addressable, but addressable in groups.

[wild glade]

What does C A and B mean here?

[unreal flax]

Those are "GD&T" symbols, giving some subtle mechanical-engineering information about what the measurements are in reference to.

[wild glade]

Ahh, now to figure out what datum means in this context, fun not having english as my main language, but, TIL 🙂

Manually naming 486 pins is fun tho

Why can't I just say "this is 0,0, the rows are this, the columns are that"

Oh well, guess it's occupational therapy

[supple pollen]

I think "Datum" in this context means "the point from which everything else is measured"

[wild glade]

Turns out you can

I hope I did it right

Not worried about missing or misplaced pins though

Just that index mark

[hushed smelt]

That makes sense too since the current lamp has only yellow or white LED's and not individually addressable.

[wild glade]

So many pins

I'm sharing the files when I'm done, nobody should have to do this much work lmao

[formal scroll]

Can someone here recommend me a tutorial for using an arduino to make a PWM controller for a 12v motor that needs to run for a set time, be reversed in direction, then stop? I am building an automatic litter box using a 12V DC motor with gear reduction. I have been looking on amazon for a good controller but I don't know which one is reliable and the one I saw with a timer doesn't sound like it will reverse motor spin, and I was hoping someone had experience with this so I can build this cheaper than the exorbitant cost of the automatic litter boxes available. Thank you for any information you can provide me!

[rigid fiber]

OK, I think this is the final version to be sent to JLCPCB when the outsources parts arrive. https://oshwlab.com/coryca/esp32-s3-pwm-fans Hopefully in 2 weeks I will have the PCBs.

[supple pollen]

An easy motor controller is the AdaFruit Motor Shield, which supports variable speed and reversing, and is reliable. Another possible choice is the Crickit. Normally to run something for a specified time, you'd get a timestamp, add the desired period to it, then start the action, and loop until the current timestamp is greater than or equal to the computed timestamp.

[formal scroll]

I figured it out , thank you though. I am using a 12V 150W DC mnotor with a 43amp driver module I found on amazon. I have to make sure its more than powerful enough to roll a barrel with possibly 50lbs of cat p 00 and sand. LMAO not allowed to type a relatively benign word

[dry pelican]

For 43A, it's probably best to use a relay instead of something solid state. Because any mosfet or igbt powering an inductive load will have to be huge and there will have to be a huge flyback diode. A relay is a much simpler device (if you don't need speed control) and it is much more robust. You do still have to take into account the inductive spiking since arcing can damage a relay, and n inductor with 43A going through it is going to want to keep 43A going through it (motors are not purely inductive, but it gives a good maximum to design for).

[hushed smelt]

19" Workshop Lamp PCB

The feeling when you hit DRC: 0

5V 6A full power. 😬

[distant raven]

Who are you trying to fab through?

[hushed smelt]

Probably have them assembled by JLCPCB, uses LCSC parts that it says are in stock.

They assembled the Bleeding Rainbow strip and I'm pretty happy with those results.

[distant raven]

I was curious about the size tbh

That’s definitely pushing the limits of size

[hushed smelt]

One of the nice things about using EasyEDA is the direct integration with JLCPCB & LCSC. I know it geared to drive sales to them and I'm fine with that. The EDA works well, easy to use, and it works for me.

[supple pollen]

I'd be tempted to make a bunch of concentric rings if the cost were about the same

[hushed smelt]

Oh, yeah umm I haven't checked yet. 😅

I thought about that but I really don't have the use for them and having a bunch of rings would be very expensive. Pretty sure I'll only get charged for the 1 ring and not the center volume.

since minimum order is 5 it's going to be expensive enough as it is.

i'll have 4 left over hopefully for some other projects. maybe a drone with ufo lights? who knows.

[supple pollen]

I'm used to being charged for the smallest rectangle that can enclose my board, but it's possible they do clever-enough panelization to put something else in the open area inside the ring, or they compute prices using a different metric. I did a run of boards that cost me $75 for the boards alone a while back.

[hushed smelt]

$514 👀

$200 additional charge for the extra large size. $100 engineering fee. oof.

Think I'll just zip tie a flashlight to my lamp now. 😦

[supple pollen]

You might consider making it in 90° arcs like the larger AdaFruit LED ring, it might be a lot more cost effective.

[hushed smelt]

aha and minimum order is 5, genius

[supple pollen]

Good point! I did take advantage of that sort of scaling in a few builds. Like my 24-channel light dimmer, I just had a trio of 8-channel boards made at OSHPark.

[hushed smelt]

https://tenor.com/view/gold-star-good-job-thumbs-up-you-get-a-gold-star-you-are-a-star-gif-13205174

thank you @supple pollen

[dry pelican]

72 degree angle PCBs. Now you can use all 5 for a complete circle 😛

Another option might be getting a premade neopixel strip an using some sort of partial conical reflector. But partial conical reflector will be hard to find unless you have necessary metalworking equipment.

[hexed mural]

WOW

[hushed smelt]

Even better, that cuts it down to 20 LED's per section.

[hushed smelt]

@dry pelican 72 degrees 👍

[dry pelican]

Nice. How will you connect the boards?

Also, I just suggested the 72 degree boards as a joke, but I guess it also makes practical sense

Are you getting them assembled by JLC?

[dry pelican]

That will be extremely bright tho, and because the eyes work logarithmically, it isn't that much brighter than a lower control level to the neopixels

but 30W is actually pretty reasonable for LEDs

with power losses might be like 25W

[hushed smelt]

Brighter the better, I can control the brightness with circuit python 😉

I just hope they'll be bright enough.

[wild glade]

modularize every individual LED or small groups, and 3D print a ring instead, that allows you to keep PCBs square and/or panelized boards

[dry pelican]

At full brightness, that thing might be able to light up a whole room depending on efficiency

[wild glade]

should be cheaper

[hushed smelt]

https://www.printables.com/model/452082-solder-webcam-lamp

[dry pelican]

also, the white tends to have a very bad color (it's more blue and it makes the colors of certain things look wrong). Maybe gamma correction or something would help

but the best option is to use RGBW

[wild glade]

Interesting, then deffo make little square boards and hide them behind a diffuser

[wild glade]

Sticking to white Cree LEDs won't give you the good warm white? 😛

[hushed smelt]

I can gamma correct np

[dry pelican]

but the separate rgb channels still make a lumpy spectrum. Still better than most lighting strips

[hushed smelt]

yeah, we'll see how it goes

i really don't want to redesign it for a 3rd time using RGBW instead with a different footprint. spent all day on this one.

Looks like it'll be about $100 total

If it works as a bright lamp it'll be worth it. My current one is on the fritz and dying.

gah plus shipping is $133, well at least it's not $500.

this is going to be the most expensive lamp i've ever owned

[hexed mural]

Doesn’t Dyson make a crazy expensive lamp? Time to order the full ring in one piece and take the crown!

[hushed smelt]

lol no thank you. they can keep that crown.

if it works the best part is i'll be able to tell people my lamp sucks

[hexed mural]

But what if you make it so you can remove the ring and also wear it as a crown

[hushed smelt]

that is... a possibility

why wear a crown when you can wear a dragon mask though

[hexed mural]

https://www.dyson.com/lighting/desk-lamps/lightcycle-cd04/black-black you’re close! It’s $600

[hushed smelt]

there's no way that's for real. it's a tiny bulb in a stick

and it doesn't even suck

[hexed mural]

Oh it’s real. The floor lamp version is like $850

[hushed smelt]

now i'm gonna have to 3D print that as a parody and power it with circuit python

and i'll market it as "designed to reduce wallet strain"

[hexed mural]

There’s some patent around the sliding mechanism, which is actually pretty cool. Not $500 cool but still. There was a hackaday article about someone trying to reverse engineer it

[hushed smelt]

it's two sticks... lash them together with twine and call it Casablanca Shiek style.

pic from far away almost makes it look like 3D printer v-rollers which would actually make a pretty cool automated lamp

a single stepper motor in a base with a long lead screw, height adjustable lamp...

[hexed mural]

You mean like this https://hackaday.com/2023/01/02/diyson-lamp-hides-cables-between-the-seams/

[hushed smelt]

You can hide cables with roller bearing channels that transfer power. Cars use them for dimmer switches and they only cost like $15.

[dry pelican]

Shouldn't be that expensive. Usually shipping goes up like that if the package gets really big. But a lamp is not $133 big

[hushed smelt]

Already added everyone who helped into next weeks circuit python meeting hug report. Thank you all for the advice, that was a great discussion.

[hushed smelt]

Oh sorry missed your questions up there. No, your advice was spot on. I'll daisy chain them with some hookup wire. Each panel has 3 mounting holes. I can heat set insert screws into the 3D printed light housing where ever I need. Even though it might have been an off the cuff joke from your perspective it was absolutely the right answer.

turn their 5 PCB minimum against them. 😛

[woven bluff]

I have solder mask being easily destroyed at those place, any idea why?

[supple pollen]

Because it's pretty narrow and only anchored at one end?

[formal scroll]

@dry pelican thank you for the additional information! I plan on running the motor at half power and that means the driver will be running at almost 1/8th load, is this design still cause for a rethinking of the parts I will be using? I have way more skill in programming than I do in hardware/electronics so I can't really answer these questions for myself. I bought two of these items for controlling the motor.
https://www.amazon.com/BTS7960-Driver-Module-Arduino-Current/dp/B07TFB22H5/ref=sr_1_8?crid=ZHGZ27SO7XM4&keywords=h+bridge+motor+driver&qid=1682422373&sprefix=h+bridge+motor+driver%2Caps%2C86&sr=8-8

And I am using this motor
https://www.amazon.com/gp/product/B07K9KPDNV/ref=sw_img_1?smid=A18IHNL4DD28Y&th=1

I don't see a relay on them.

I almost forgot to mention, I will be buying a more expensive, name brand, quality powersupply. This will be going upstairs in my home out of sight for essentially the entire day so I wanted the assurance that comes with "you get what you pay for" so far as firesafety goes. Would anyone have a recommendation for a 12v power supply that fits that perspective?

[supple pollen]

Mean Well makes some solid supplies at a reasonable price point. I don't really recommend buying stuff from amazon if you're interested in safety (or at all, really).

[dry pelican]

That motor does not look powerful enough. It says 3.97kg*cm for rated torque, and you can calculate the maximum load. Using pwm to control speed also limits power. A better option would be to use a belt or something for mechanical advantage.

[formal scroll]

I am printing a gear reduction for it so it has more torque

[woven bluff]

I had a failed meanwell boost, then I switched to XP powers and Recom

amazon and ebay SMPS are definitely Chinese junk made with cheapest parts possible

I had one coupled SEPIC module that arcs inside the core

[dry pelican]

Do you know how much power you need?

Maybe you don't need the whole 516W of motor power

And that motor cannot supply that amount of power

516W is almost 1 horsepower

[formal scroll]

there are like 20 motors listed on that page, I got the 150

or wait hold on

yeah I got the 150rpm at 150 watts

I am using a driver capable of supplying 49 amps of current

[hushed smelt]

That's a lot of amps.

[blissful frigate]

attached to an infinitely large heatsink?

[formal scroll]

lol no, I doubt it can go that high, the chineesium modules never work as good as they are advertised.

That's why I got that one. Can't rely on the specs too much but I bet I can run it at a much lower load and have it work well

[dry pelican]

Rds(on) is your best indicator of heat for mosfets. A mosfet might say it can do 100A, but if the rds(on) is too high (which it can be at lower voltages), it can't without blowing up.

[lusty heath]

I haven’t made many PCB. Made a mistake. I have Tx to Tx. Rx to Rx. Feather to RoboClaw motor driver. How does one go about “hacking” the board to swap them?

[unreal flax]

Typically you'd use a razor blade or X-acto knife to cut the wrong traces and then solder on short wires to make the correct connections.

[dry pelican]

Cut traces, scrape solder mask off, and solder some thin wires to the exposed copper

Now you have bodge wired a board

[unreal flax]

The copper is a pretty thin layer, so it's vulnerable to cutting and scraping.

[lusty heath]

Ok. Thanks both. I have an x-acto knife so I can cut them.

[spice turtle]

Does the feather not have re-mappable pins?

[supple pollen]

It depends on the Feather. The RP2040 allows a fair amount of pin reassignment, while the 32U4 is pretty fixed. There may be other options with even more flexibility.

[supple pollen]

Yes, join us bodgers! We have cookies! 🍪

[dry pelican]

Bodgebodgebodgebodge

[hexed mural]

That coilbodge rules

[dry pelican]

When your mosfet driver can't take the GDT's load, add in a chunky double complementary mosfet output stage to drive the thing

It's kinda unnecessary, but it prevents the gate driver from overheating and breaking

The circuit still needs some improvements

GDT waveform is messy, etc

But it's hard to argue with this (31v bus)

[hexed mural]

Whoa nice

[lusty heath]

My first successful bodge! It works now.

[lusty heath]

I looked that up. I believe that you can’t swap the Rx and Tx on a samd21.

[wild glade]

Will the photoresist stuff stay on the glass plate if I expose a photoresist PCB with my MSLA printer?

Is it wipeable? does it do anything to the glass? 🤔

[supple pollen]

Should stay on. Probably removeable with the right solvent. It shouldn't do anything to the glass by itself, but I've seen people use resist to etch patterns on glass using glass etchants.

[wild glade]

Hmm, so, cheap screen protectors then and expose as much as possible in one go.

[hushed smelt]

Bodge fixes at least means you're trying which is more than some even attempt. It happens to everyone and should be considered part of the process.

Final_Design filename often gets renamed Prototype_1 😁

[covert blaze]

Good morning everyone, I'm wondering the INA 219 on Adafruits website calls it a high side current sensor, but on TI's website it is listed as a bidirectional current sense

https://www.adafruit.com/product/904

[covert blaze]

In summary my current design uses this layout to shift any negative voltage back to positive voltage with a Vref offset, which I feed into a 24 channel MUX to read using a ADS1115. And I am wondering if I can reduce component count and cost by using some other design.

[supple pollen]

"High side" means it can sense current even when one end of the sense resistor is not at 0V potential. "Bidirectional" means it can measure current flowing in either direction. It is possible for a sensor to be capable of both, and presumably this one is.

I'm trying to figure out what you're trying to do: your description sounds like you're trying to measure (and shift) a voltage, but that sensor is primarily designed to measure current.

[covert blaze]

I should have been more descriptive, but I intend to read 64 different currents

For both negative and positive current

Shifting the voltage in the schematic is just to prevent negative voltage outputs

I already have a design that measures 24 current values, but I want to make a new 64 current sensing layout.

I am mostly focused on reducing component count now, with my current idea being to reduce the number of INA181s by using two 16:1 multiplexers with that output then being fed to an ADS1115 ADC

[supple pollen]

You're probably already at the minimum component count. You might be able to omit R5, R6, and both R10 resistors, and you might be able to derive Vref from an existing point in the circuitry instead of a separate regulator. Note that the diagram shows INA181 and the AdaFruit part you linked to uses a different but related part, the INA219.

Note that multiplexors probably won't handle the current you're measuring

[covert blaze]

Yeah I was playing with the idea of feeding the INA219 input using a multiplexer and then directly reading the current through I2C

Each signal will only be a few mA

Is the current plan

[supple pollen]

You aren't going to get much resolution measuring a few mA with an 80mΩ shunt resistor.

[covert blaze]

You are correct, it'll probably be 5 times higher or so in the new design

[supple pollen]

You could get away with more, as you're already adding 125Ω or so in the multiplexor

[covert blaze]

Shouldn't there be no current flow through the mux as it goes to and ADC which has high impedance?

[covert blaze]

Also forgot the INA181 has a 200x gain

[supple pollen]

Oh, you plan to use 64x INA181?

[covert blaze]

I will be using them, but I am trying to find a way to not use 64 of them

[supple pollen]

5mA times 80mΩ is 4µV: at that point, your measurements are likely to be swamped with noise, even if you add gain (or perhaps especially if you add gain)

[covert blaze]

Yeah definitely still have to increase the shunt resistor, probably to 250 mΩ
5 mA * 0.25 Ω = 1.25 mV
1.25 mV * 200 V/v = 0.25 V

[wild glade]

Non-standard symbols confuse me :v

L3 is just a inductor, C27 and C28 are caps, but... specific type? polarized or not? who knows.

I think it means ceramic cap?

[spice turtle]

those small values are probably non-polarized. Ive never seen those before

[supple pollen]

I remember some film capacitors mark the "outside foil" on one lead for use as self-shielding (normally that end is connected to ground or the lowest impedance point), but they're pointing the other way here. In tubes, a dot like that indicates a gas tube as opposed to a vacuum tube, but most capacitors contain neither.

[elder peak]

I can see how the schematic would make you dotty

[dry pelican]

it prolly doesn't matter what caps you use. As long as they're not high ESR or ESL

depends on application of course

[supple pollen]

4.7µF is a value that's often implemented with an electrolytic (polarized) technology, that one you would want to put the positive side to the positive voltage. However the 100nF one is normally going to be a ceramic or film capacitor, and those aren't polarized.

[dry pelican]

4.7uF could also be an mlcc

Depends if it's a tht or smd design. For tht, electrolytic would be used

[woven bluff]

4.7uF electrolytic and ceramic has similar footprint size when V>200

is there dedicated controller chips for electronic load?

I'd like to roll one with maybe SiCFET+gate driver+MCU+I2C current sensor, the sensor would update PWM registers of MCU

[spice turtle]

other than using an opamp? no.

[arctic gate]

Hello! I'm looking to do a project with an analogue joystick (like https://www.adafruit.com/product/5628) and the KB2040 Kee Boar Driver (https://www.adafruit.com/product/5302). Would anyone happen to have footprints available for these parts, or be able to point me in the right direction to find them?

[hushed smelt]

Might have to make your own footprint for that joystick https://cdn-shop.adafruit.com/product-files/5628/P5628_C17894-001-A+datasheet.png

I don't see a breakout board for it in the shop.

which is usually where you could grab a footprint from.

as for the KB2040 https://learn.adafruit.com/adafruit-kb2040/downloads

[arctic gate]

Yeah I couldn't even find anything for it on mouser or on the manufacturer's website

[hushed smelt]

footprints and schematic are usually in the learn guide for a product. if there's no learn guide then the schematic is usually linked to from the product page and it's up to you to design a footprint for it.

http://www.k-silver.com/html_products/JP19（正插）-829.html

Search for the model# in the schematic

[arctic gate]

oh i was just on the wrong site 😐

[hushed smelt]

Mouser, Digikey, etc.. will use that part # as the manufacturer part#

If it's still for sale, not obsolete or no longer sold, should be able to find it with the manufacturer model#

[arctic gate]

https://github.com/eggsworks/tamago60/blob/main/tamago60/tamago60.pretty/JP19.kicad_mod
Just found what I need, I think

[hushed smelt]

Seems about right, double check the measurements for everything against the datasheet to ensure it's correct. Don't rely on someone else's footprint without some basic measurement checks.

[arctic gate]

Yeah I will for sure

[arctic gate]

I've checked all the download links and I can't find any files I recognize as footprints (at least not any pretty/kicad_mod). Am I better off just manually editing a pro micro one?

[hushed smelt]

The .brd and .sch file are EagleCAD format. You'll have to find a way to convert it to your preferred format. https://github.com/adafruit/Adafruit-KB2040-PCB

[wild glade]

Can't you just make the footprint yourself?

[hushed smelt]

Footprints take a while to design, it is nice to have them available to drag and drop into a project. Some people know how to work with existing footprints but don't know how to make their own yet. I was there once, creating footprints is kind of headache and an extra step unless you can find an already existing one. Using an existing footprint is a more efficient use of time while designing. It's why footprint files exist so you're not recreating the wheel every single time.

[wild glade]

I'll make it if need be

[vast flume]

Hey guys, quick question. Is it possible to communicate with my stm32 board from my computer via the jtag connector?

I'd like to be able to update some parameters on it without having to recompile the whole program

[unreal flax]

For some value of "communicate", yes. You can use the debugger functionality to change values in memory, for example.

[vast flume]

Ah but no reasonable way for me to write a program on my computer to interact with the target board?

[unreal flax]

You don't need JTAG for that, you'd probably want to just use a serial port.

[vast flume]

The problem is that I already have the pcb 😅 . And I didn't put a usb on it