help-with-hw-design

[latent jungle]

The 74LVC245 is a bunch of MOSFETs.

If you need a single channel, then the `245 is overkill. If you need many channels, the space needed for discrete FETs is relatively large.

Also, the `245 will not work for I2C. The chip can only work in one direction or the other. It isn't bidirectional in the way that I2C is.

[knotty tiger]

discrete MOSFETs are often designed for higher current, and are likely overkill for level shifting

[icy hawk]

well they're cheaper than a daughterboard which is what I'm currently using because I couldn't get a discrete MOSFET to work

[knotty tiger]

for something like an addressable LED string, you likely need two MOSFETs per line (one for high, one for low). I2C can get away with a single FET because it’s designed to have an idle line pulled up by a resistor

[icy hawk]

interesting. I also tried using these (two MOSFETs in one, idk if they're wired right though) and couldn't get them to work either: https://www.amazon.com/dp/B0893MKNB2

[latent jungle]

Using a MOSFET intended for Power (high current) to do level shifting is overkill and may not work as expected.

Even though we think of transistors as digital devices, they really are not. They have very non-linear analog behaviors. For example, to fully open up the channel between source and drain, that 15A FET needs at least 10V between Gate and Source.

So at 3.3 or 5V, the channel barely opens and may be dropping a relatively high voltage (causing an unwanted voltage divider.)

FETs like the BSS138 or 2n7000 are far more suited for the low current applications like level shifting. (although, arguably, the 2n7000 is overkill as well.)

And in general, as the channel gets larger for higher currents, so does the gate (to drain) capactiance--which affects overall switching speed.

[icy hawk]

ok so it sounds like the advantages of buying a daughterboard to do level shifting are that you don't have to worry about which MOSFET to use, how many, how to wire them, or whether you need resistors or other parts to make the level shifting work correctly

[elder peak]

Yes, the people who picked out the right part and made a board should be FETed for making your life easier.

[latent jungle]

(designing your own is a fun exercise, but I wouldn't recommend it without knowing how to use an oscilloscope first)

[worldly schooner]

Is there a schematic available for the Push-button Power Switch Breakout? I’m looking to design something similar and wanted to study the kill pin in this design.

[timber wyvern]

https://github.com/adafruit/Adafruit-Push-Button-Power-Switch-PCB has the EagleCAD files, they've never made a 'learn' or published the schematic or mask-diagrams for that one AFAIK.

[icy hawk]

trying not to get too stuck on this but I do think I'm learning 😅 so I need a fast logic level shifter to be able to control DotStars with a Raspberry Pi. there's the QT level booster which is for logic, but I couldn't find the specs for how fast it can switch data. then there's the pixel shifter which was intended for Neopixels but maybe can support DotStars as well?

the DotStar guide only recommends a 74AHCT125 chip which I really don't want to wire up myself. the QT level booster doesn't specify what it's using to handle the level shifting from what I can tell. the pixel shifter has a 74HCT2G34 which is a different chip than the DotStar guide but it can handle up to 10 MHz...

DotStars are SK9822 LEDs which means they can handle up to 30 MHz serial data input. would really be nice to be able to use all of that since I'm using a Raspberry Pi (can send up to 32 MHz)

ok according to this page the 74HCT2G34 can actually handle 32 MHz? so the pixel shifter should be fine, right? let me know if I'm totally off track here

[timber wyvern]

A lot of Pi models can handle MUCH faster than 32Mhz, the base frequency for the SPI pinouts on a Pi5 for example is 125Mhz.

[latent jungle]

You're over-thinking it. I believe DotStar switching frequency is something around 1 MHz. The bs1338 of the QT level booster can easily handle that. It has an on+off time of about 25 ns. which means there's plenty of margin for a 1us (1 MHz) signal.

oh, my bad. I was thinking of NeoPixels.

[knotty tiger]

beware that the breakout board is wired for I2C, so it can only actively pull low, and depends on 10k resistors to pull high. this might or might not be fast enough for a DotStar, depending on how fast you’re going

[icy hawk]

I want to go the full 30 MHz

I’m doing persistence of vision

[latent jungle]

Across how many LEDs?

I just want to make sure you're not mixing up 30 ms (30 Hz) with 30 MHz.

[icy hawk]

just 100 LEDs

well I want to make an animation, possibly around 60 fps. and every frame will need several updates in order to actually draw the image

60 frames * 100 LEDs * ~60 slots in a circle that I want to draw (maybe more idk) * 32 bits per LED? not sure. but that many bits per second

[timber wyvern]

24 bits per LED, and either it's 60fps or it's higher fps, it's not both.

So it's 2400 bits per update to refresh all the LEDs (or 3200 if it's RGBW?) 32Mhz would be 10-13k FPS, not 60.

Like are basically wanting to 'black frame insert' for varying periods like a crude PWM to test how small of an image pulse an individual needs to see the video it sounds like?

[icy hawk]

by fps I mean fps of the image, not fps of the LED strip:

the more frequently I can update the strip the higher "resolution" image I can display

with each LED strip update I'm only displaying a fraction of an image frame

I think 60 divisions will probably be enough but I'm not sure

60 image frames per second is probably overkill; I think 30 would be fine

but it would be cool to push it

[timber wyvern]

Ah, trying to do some kind of motion-reliant visual like the 'picture on a bicycle wheel' or the 'shake the stick back and forth and it shows an image' type visual effect? So yeah you'd need FPS * pixels * travelling-distance resolution updates per second then.

Sorry I misunderstood before. 🙂

[icy hawk]

yeah it's for a bike wheel and I'm using a magnet and a Hall effect sensor to detect the speed and adjust on the fly

[timber wyvern]

I thought it was a DotStar based fixed square panel or something, since I've seen 10x10 sized panels of addressible LEDs before.

[timber wyvern]

A simple rotary encoder attached to the pivot or rolling against the wheel like one of those tiny generators for bicycle headlights you can get would likely be an alternative to look into, just display the slice based on the current rotation angle instead perhaps.

[icy hawk]

how fast can those spin? I can easily reach 25 mph on my bike which translates to ~420 rpm on the wheel

[timber wyvern]

Depends, but just checked prices and they're fairly expensive new so likely not helpful unfortunately. Great if you can salvage one from a scrapped CNC or similar though.

[icy hawk]

eh I'm fine with the Hall effect sensor. that's what the OG spokePOV project used. was that the first ever Adafruit hardware ever sold?

[timber wyvern]

The SpokePOV was https://www.adafruit.com/product/5 and the MiniPOV was https://www.adafruit.com/product/1 so basically, yes.

[tough matrix]

there are plenty of quadrature encoders that can easily handle 420 rpm, with high enough resolution. E.g. this one can go up to 10 000 RPM at 8192 ticks per revolution:
https://www.revrobotics.com/rev-11-3174/

but indeed hall effect sensor would be easier and cheaper

[eternal bear]

Hello,
I have an adafruit matrix portal s3, and I want to connect a hall effect sensor to it. The sensor I had in mind was a NJK-5002C due to the form factor for its intended application. If I connect it directly via the JST connector at 5v, will it work ok on a0 with the internal pull up?

[unique patio]

the output voltageof the NJK-5002c is the same as the input, so in this case, 5v. So you would want to do level shifting. There is a zener protecting the D2 input but that's not a good way to do level shifting. Use a voltage divider with a couple of resistors. There is an example of using a divider in the NJK-5002C datasheet, but for different voltages (12v).

[deleting your other post -- please don't cross-post, since then you might get two answer threads]

it is possible that the part might actually work at 3.3V. You could try it and then you won't need a level shifter. I did some shopping for these and I see lots of voltage ranges possible. Also the prices are all over the place.

[eternal bear]

thanks for the reply, I'll do a bit more digging, barrel form factor at 3.3v has been hard to come across

have seen a few other comments saying some of these will function at 3.3v, if it's undervolted is it just a case of it either works or it doesn't?

[unique patio]

I'm not sure we could say without testing or asking the mfr whether it would just not work or be flaky. You could ask OpenTec (if that is the supplier)

I have seen 3.8 min, 2.7 min, etc.

[unique patio]

NJK-5002C says 6v min in one place, 5v in another.

So worst case is you just use a voltage divider

which is not a big deal

[eternal bear]

i thought it would be a case of a resistor or level shifter rather than a voltage divider for the NJK-5002C output. I am very inexperienced though, just looking through the datasheets and looking at terms

[unique patio]

a voltage divider can be a level shifter: https://embedded-systems-design.github.io/level-shifters/

[eternal bear]

thanks, useful doc

[halcyon bobcat]

hello, good people. I have two questions:

I'm interested in building a thermal camera. the supplies and tutorial are available from adafruit; however the tutorial is for the pybadge, which is out of stock.
(1) would the same code for the pybadge work on the CLUE (also programmable in Arduino) if only needing some tweaks? I am not skilled in programming, but I can read and follow code.

The CLUE is nice because it also has many built-in sensors and the whole board can be used and repurposed, but the display is tiny.
(2) Is there another board and TFT combo that would work (1.8" to 3" range)?

Thanks, I'm a physics lab manager at a university, and I'd love to use microcontrollers to build small devices and work with student on projects, but it's my greatest skill weakness.

[lyric meteor]

Hey @restive brook - I see that you are the moderator for CAD models - can you add the model for 1212 SMT Breakout? I have a bunch to do and I want to laser a stencil. I looked in the repo but it's not there https://github.com/adafruit/Adafruit_CAD_Parts

[flat vigil]

I think filing an issue on that repo is how to make a request

[lyric meteor]

Did that - thanks

[round stump]

Anyone out there trying to launch a hardware company these days? How rough are the tariffs and such? I launched something back in 2021 and had some pretty decent traction and I'd like to start again. However, I'm wondering if now is a terrible time to begin.

[tough matrix]

I have a very small scale tindie shop, but it is not really a company, just some way to make money to cover my own electronics buying. It runs ok, but i am not investing any effort in promoting it.

As for tariffs: JLCPCB now adds 45% tariffs to all orders. So you have to account for that in your prices.

[worldly schooner]

Are those tariffs applied to shipping costs? MOre often than not those shipping costs outweigh the boards I'm actually ordering...

[tough matrix]

no, fortunately not

so they charged 45% tax on (merchandise total - discount): 0.45* 440 =198

[worldly schooner]

Phew, that's a relief. For sizeable orders like that, it's still quite substantial though....

[icy hawk]

I'll need a couple types of plastic for a project. one will be a stiff board around 6x12" that will be mounted to the back of a HUB75 LED grid. another will be some tough but very flexible clear plastic (honestly, a shower curtain is pretty similar) for wrapping around buttons and displays. both need to be waterproof. is this something I can find in reasonably small quantities at some kind of industrial supply store? or should I buy a similar consumer product and repurpose that?

[lost portal]

You could look at an ABS plastic adaptable box/junction box for the hard plastic. They usually come in grey or black (occasion cream) and range from about 2” square to 18”x24”

[tough matrix]

mcmaster-carr has all types of plastic one might want, though their prices may be high.
Another vendor is online metals - despite the name. they sell plastics too:
https://www.onlinemetals.com/en/buy/plastic

For stiff plastic, you can also buy a large cutting board (they are typically made of HDPE) and cut to size using a saw.

[proven pine]

Anyone have insight into FCC compliance? I'm trying to make a rf enabled dev board at a very very small amount. (Without a module). I can offer it without FCC intentional radition checking?

[tough matrix]

Will you be selling them?

[proven pine]

Maybe, but the intent is to give them away

But I do have other project in similar shoes that I intent to sell

[tough matrix]

AFAIK, if you use a ready module - like one of esp32 modules - they are already FCC certified. Otherwise, you do need to get a certification, and it is a major pain, but I never did it myself (but know people who did).

Maybe @thorny cloud has some experience?

[proven pine]

You mean not using one?

It's something like this

Intented to be like a badge that I give away to people at tradeshow etc

But appearently that still counts as distribution under FCC law

[thorny cloud]

@null that looks more liek a prodict than a development board - which lands you in the "require FCC to sell it or even give it away".

[proven pine]

Yea

Man making small scale hardware is such a pain here in the states

[tough matrix]

SparkFun had written up their experience with FCC cert: https://news.sparkfun.com/3124

[thorny cloud]

The FCC has this concept of a "sub-assembly" that unofficially (I am not a lawyer) means that a device that is specifically designed to be attached to other devices, or have other devices attached to it, to function, can skip the FCC part - and the "final thing" that is “assembled" needs to be FCC certified.

But honestly, it's a super grey area. The fact you are giving them away, and I assume in small numbers, you should just do it. FCC costs on products with intentional radiators on them are VERY cost prohibitive.

But again, I am not a lawyer, or in any way qualified to give anyone advice, and you should take anything I say (and anyone else here) with caution and scepticism.

[proven pine]

Maybe I will just remove the RF if I make it offer for sale and do a SDoC or third party 15B check

[thorny cloud]

BTW, that boards looks great!

Can folks flash their own firmware on it?

[proven pine]

Yes

it will be full open sourced

hardware wise & software wise

[thorny cloud]

Ok, please dont remove the antena. If you have no antenna and folks try to enable the radio (wifi or BLE) they'll likely damage the S3 chip.

[proven pine]

Well I guess I can efuse them off

(Not sure if that's an option in the efuse)

Another alternative I can offer

is a board without S3 on it

And people need to put one on themself

but this is probably too much to ask

[thorny cloud]

I'm not sure - I don't recall seeing anything like that, but I've never really looked.

[proven pine]

Yea. That's something I really don't want to do anyways...

[thorny cloud]

so they way you "may" be able to get around it (again, discussion, not advice) is to put a uF connector on the board, sell it without an antenna (BYO for customer) and state shipping firmware includes no RADIO use.

Technically, it's the board and its use with shipping firmware that get's FCC'd. Not the HW itself, but it's intended use.

[proven pine]

It seems like there has to be no way for end user to add antanna. Even a footprint can't work. (From what GPT told me)

I guess there's a gray area of yea we try to stop user but we didn't try very hard kind of things

but hard to draw the line there

[thorny cloud]

If the intended use has no radio functionality and it's off - you may be able to claim it's non-functional.

But again, it's such a messy grey area - every company/HW engineer needs to decide for themselves what level they want to go to.

[proven pine]

😭 😭

Thank you for your words!

[thorny cloud]

Nps - good luck with you decision making.

[tough matrix]

now I am uneasy whether I am legally ok selling my own product that uses adafruit's esp32-s3 qt py, as I am using the wifi functions of that dev board

[proven pine]

I don't think so

because that board doesn't have FCC ID

[tough matrix]

but I am not going to get it certified, that's just impossible

[proven pine]

So if you sell full product. You are required to do intentional radition on it

Yea...

https://www.seeedstudio.com/XIAO-ESP32S3-p-5627.html

You can try to switch to this instead. It's FCC certified

Also taking about the s3 qt. From what I read you can not sell boards at retail if they are determined as dev kit. (I asked GPT to do research for me and it says that dev kit are suppose to be internal use only and can't be distributed in anyways). But I do know they are also available at Microcenter...

[tough matrix]

unfortunately, an external antenna is not going to work in my situation, so I can't use Xiao.

[proven pine]

Maybe just embedded a ESP32-S3 mini module?

[timber wyvern]

Cite actual rules or laws somewhere, not the hallucination machine please.

Like if it can't give you an actual reference to that claim it's not a useful claim.

[proven pine]

Well the issue is that... I did read teh actual rule

and it's a mess with gray areas since it was written in the 90s

[timber wyvern]

Since it may be conflating things like all the lawsuits about console dev-kits being illegally sold and such, and mixing up the terms being used in different ways across different decades.

[proven pine]

The usual gray area for devkit is this. They fall under CPU boards

Subassemblies to digital devices are not subject to the technical standards in this part unless they are marketed as part of a system in which case the resulting system must comply with the applicable regulations. Subassemblies include: 
Devices that are enclosed solely within the enclosure housing the digital device, except for: power supplies used in personal computers; devices included under the definition of a peripheral device in §15.3(r); and personal computer CPU boards, as defined in §15.3(bb);
CPU boards, as defined in §15.3(bb), other than those used in personal computers, that are marketed without an enclosure or power supply; and 
Switching power supplies that are separately marketed and are solely for use internal to a device other than a personal computer.

But this is only for Part 15B. So technically if your device have RF then they are subject to Part 15C testing

The the thing unclear is that can we treat Part 15C as an extension of Part 15B? I don't think so, and there fore this rule doesn't mean that custom RF devkits can be sold without Part 15C

[timber wyvern]

Thank you, sincerely, that's much better and more verifiable information. You're correct.

[proven pine]

So in the end, if we take this strict at face value. A lot of things on market simply doesn't compliant. Even the ESP32 S3 qt py

This is probably a very very wide spread issues. Either people have to accept the risk and just play in this gray area

or settle for something else...

Just really unfortunately if you are not a big company making millions

[pastel night]

last i checked there is exemptions for smol quanity (10?) aka prototypes. intentional emitters are another levels of testing over susceptible and unintentional emitters. There are some pre-compliance testing setups you can get (~5k$ software hardware bits and bobs depending on size and standards that apply). lab testing is going to another xxxx$ thousands min. Reasons to buying modules with FCC ID and only have to do minimal testing.

[formal scroll]

What's a good RF IC for a homebrew 900mhz long range FPV drone tx/rx ? There are wayyyyyy too many to sort through in a reasonable timeframe so I figured I would ask where experts hang out.

I was planning on fixed wing with long loiter, long range, for observational/mapping/zoning out looking at nature.

[tidal cliff]

anyone know what the type of connector on the right side of this board is?

[lost portal]

It’s a type of push fit terminal connector. Somewhere like cpc/farnel or mouser should stock something like that

They’ve become increasingly popular over the last 5-10 years in commercial electrical components instead of screw terminals. When you’re installing a few hundred lights in a building, the seconds saved per fitting add up to some serious time savings

[tidal cliff]

yea I'm having trouble finding pcb / smt mount though

[lost portal]

I think they’re usually through hole, rather than surface mount

[tidal cliff]

yea, or not pcb mount at all

[lost portal]

Screwless fixed terminal block https://www.mouser.co.uk/c/connectors/terminal-blocks/fixed-terminal-blocks/?q=pcb terminal

[unique patio]

This is a WAGO surface mount terminal block. I searched for the image and found https://magwled.com/pages/faq which mentions it is a WAGO. https://www.wago.com/us/discover-pcb-connection/surface_mount_pcb_terminal_blocks

better view: https://magwled.com/products/dgrz-led-cpd-36-1-mini?srsltid=AfmBOops4FixcrN-oSH1BI2eTIvTYS4ilzdzg-h_WhoEMNU4Dy6_w6pb

[lost portal]

I didn’t know wago did surface mount connectors.

[unique patio]

me neither

we (AF) use screw terminal blocks. We have a soldering machine that can do through-hole parts nicely

[lost portal]

I’ve seen (and used) their loose cam lock connectors and their din rail mount equivalent

Screw terminal is far more common outside industrial applications

[tidal cliff]

this might be it: https://www.wagoproducts.com/rail-mounted-terminal-blocks/series-2060/wago-2060-453-998-404/

looks like digikey's got it

not able to find an equivalent on LCSC though

[golden wasp]

Hi,

I'm doing my first ever project featuring a motor. I have this 5V DC small stepper motor: https://www.adafruit.com/product/858

The whole contraption is supposed to be a moon clock and look like the moon phase dial of a watch.
The motor would rotate a plexiglass disc, on it the moon, in front of it a cutout that hides the moon on the disc according to the moon phase. A magnet in the disc + a HAL sensor helps with the positioning. LEDs would light the disc from behind to have the moon shine.

I made a hole on the plexiglass in the shape of the shaft and made a small proof of concept with an ESP32 to rotate it around and stop when the HAL sensor activates. It works in a horizontal position but it will be upright when finished.

So my question is: how do I prevent the plexiglass from falling off (apart from having the static wood cutout physically in front of it)? I see a small indentation at the tip of the shaft but it doesn't seem to be a screw hole. What is it for? Shall I drill into the shaft and put a washer + screw at the tip to hold the plexiglass disc stable? There should be just the tiniest gap between the front cutout and the rotating disc.

Thanks for any advice and suggestion in advance.

Regards, Peter (the noob)

[stark arch]

How should I size the switching frequency for a flyback transformer? Have heard 25-150kHz is typical, but struggling to find how to choose an exact frequency within that range

[humble portal]

Hi Folks. Have a question about any damaging/life-shortening impacts to I2C-connected sensors when they are turned-on/off frequently (about once every minute). I need to reduce the power consumption on a weather station that includes a number of I2C sensors. It was recommended that I could use an Adafruit S-35710 (https://www.adafruit.com/product/5959) as a power/wake timer. Since even with deep sleep (which has no affect on the sensors) my sensors consume about 125 mA more-or-less continuously, so being able to turn them off between read/send cycles would be a big help. The biggest power hog is the PMSA002i air quality sensor (https://www.adafruit.com/product/4632), at about 100 mA continuous. Am I shortening the MTBF of my I2C-connected sensors by turning them on/off on a one-minute or so time scale? Thanks!

[timber wyvern]

Scratch that, missed you said 1/minute not 1/second, ignore me. XD

Mis-read.

[unique patio]

That should not be a problem. People do this all the time.

Sometimes a sensor has its own low-power mode you can command it into, but not always.

[fervent canyon]

Not sure if right channel, sorry if not! Looking to step down 24v input to 5 or 3.3 v. I don't see any buck converters for this on Adafruit website (max input is 21v I can find), but many LED strips are 24v and it'd be sweet to easily have a buck to step down to microcontroller input voltages. Anybody doing this / have any recommended step down converters?

[unique patio]

https://www.pololu.com/category/131/step-down-buck-voltage-regulators lists some quality buck converters. The max input voltage ranges from 36-60V.

[fervent canyon]

TY Dan!

[fervent canyon]

https://www.pololu.com/product/5593 voila

[brave vigil]

Did Adafruit choose screw terminal blocks over https://www.wago.com/global/electrical-interconnections/discover-pcb-connection/smd-pcb-terminal-blocks for cost reasons, due to not being aware of their existence, or another reason although?

[brave vigil]

I do recall that the Desk of Ladyada videos around https://www.adafruit.com/product/5768 mentioned some difficulty fitting the footprint of a terminal block onto a space constrained Feather board. However, I'd assume that a greenfield PCB form-factor, such as the Sparkle Motion family, would've been able to be designed around those limitations. 🤷

[tulip swift]

Screw terminals tend to be more compact where as those Wago type connectors since they need a lever arm which takes up some space, they just tend to be larger. Great for a larger board, but can look kind of silly on a small break out board. Really it comes does to how big you want to allow your breakout board to be.

[tough matrix]

usually one attaches a hub to the motor shaft, and anything else - like the plexiglass - is attached to the hub.
e. g. you can use these hubs
https://a.co/d/25WOPyC
or
https://www.pololu.com/product/1203

[brave vigil]

WAGO also makes a product that's a middle ground between the two. Instead of it requiring an unusually small phillips head screw driver, the WAGO 2059 Series requires their own "operating tool". From my understanding, they're somewhat similar to most SIM card slot ejector tools.

[lucid dagger]

what kind of screw fits a 2.39mm diameter hole? im guessing its some normal imperial measurement, but i dont know freedom units ;-;

its too big for M2 and too small for M2.5

3/32?

2-56 UNC?

[brave vigil]

Maybe the intention is to use the combination of a nut, bolt, and pair of washers. 🤷

[lucid dagger]

hmm

[timber wyvern]

Yeah that's meant to use a #2 UNC bolt, it has a big clearance to make sure it fits. It's not expected to be under big axial loads, it's a pushbutton switch not a slider so they left enormous tolerances for easier field assembly and replacement it looks like.

A #3-48 UNC screw should also fit if you need/want a tight fit, literally 0.001 too small for the threads but big enough for the shaft so you'd cut into it as you install it depending on machining tolerances on the screw.

[lucid dagger]

thanks! good to know i was on the right track 🙂 ill order some of both and do a test, but i suspect i will still prefer the #2

[unique patio]

In addition to @Timeline's point about footprint, the number of reuses of the connector is probably greater for screw terminals.

[sick ravine]

Screw terminals also are zero-force connections

I just went through this whole song and dance with a new design before just going with the ol' screw terminals

You don't need a sturdy mount or anything for them whereas the push force for wago and other similar designs is non+negligible

[lost portal]

Push fit don’t tend to like stranded cables much either. For best results with stranded cables, you need to ferrule or tin the end to stop it fraying.
Screw terminals tend to be more forgiving in that respect

[timber wyvern]

Not even 'tend to be' they straight-up are by basic physics, yeah. The push fit stuff is intended for (semi-)permanent installs not constant re-use as well. Screw terminals are much more re-usable.

[lost portal]

Push fit can be as reusable. They’re intended to speed up installation on large scale commercial installations, nothing more.
Also means they can employ fewer electricians as a labourer can be trusted with push fit as they don’t require any tools.

[clear matrix]

Anyone here have experience with using board mounted spade connectors instead of screw terminals?
I am currently using a Phoenix connector type screw connector in my project, but it's 14-ish mm tall and makes my enclosure bigger than I'd like. Perhaps spades are a better alternative.

[elder peak]

So, lesee, I started out with dupont-clone connectors but kept having problems crimping them. Then I moved to regular screw terminals everywhere, then spring terminal blocks, then 0.25 faston-clone crimp connectors and phoenix-style terminal blocks. I even made some wacky hybrid footprints able to handle a few different connector types.

Lately I've been a fan of Molex Mini-Fit Jr and Micro-Fit 3.0 connectors, JST XH connectors, although screw terminals and phoenix-style screw terminals are handy.

[unkempt latch]

Hi! I put together a set of two PCB designs for my vertical mouse, but this is my first time designing anything with either Li-ion battery charging or an antenna, so I was wondering if anyone could take a quick look and give it a sanity check.

The PCBs are designed in KiCad, and are available in the pcb directory of this project: https://codeberg.org/whitelynx/lynx-mouse

[golden raptor]

Looks good at first glance, I'm gonna take a look tommorow but I won't help with rf(I'm an idiot in rf)

[pastel night]

any design the touches humans... ESD protection anything that can be capacitively coupled and has a direct path to micro inputs or power paths... e.g. switches, leds, exposed contacts. common method is a TVS / Zener + a small resistance.

[pastel night]

rf... folllow the chip's layout example as closely as possible.

[pastel night]

22 ohm termination resistor on D+ and D- lines between the USB connector and u2

[unkempt latch]

OK. The reference schematics didn't show those, but I can add them 👍

Thanks for taking a look!

[unkempt latch]

I updated the design with ESD protection on the USB data lines, switches, and debug port lines, as well as correcting the crystal values and the impedance matching network for the antenna to match the reference design. Hopefully that looks better 😺

[prime raptor]

What design tool are you using?

[latent jungle]

That's KiCad.

[prime raptor]

I'm also using KiCAD, but I'm guessing there's some kind of theme function? I much prefer the darker backgrounds...

[unkempt latch]

Yeah, I'm using the Gruvbox theme 👍

[prime raptor]

Thanks, I'll give that a try, looks very nice.

[brave vigil]

IIUC, for the variants from WAGO that contain either a mechanism that requires their special operating tool, or just a regular push-button, if you do the procedure that would otherwise release the conductor, but simultaneously insert a conductor into it, that entire process shouldn't really require too much force.

[sick ravine]

Well, it's zero force in two ways right? There's the wire force and then also the force on the little PCB

Dunno about wago's connectors but the springs on Phoenix contact terminal blocks aren't insubstantial to overcome

The connectors that use a lever instead of a push button are probably better in that regard

[brave vigil]

This webpage lists the variants that I'm referring to: https://www.wago.com/global/electrical-interconnections/discover-pcb-connection/smd-pcb-terminal-blocks

[sick ravine]

Also a lot of these connectors put the pushbutton at an angle which is good in some contexts and bad in others

Oh I know what you're talking about

PC has these 45deg versions which I used at the top of my printer ebox here

Also note the PC pushbutton terminal blocks at the bottom

Really really really like pc terminal blocks

[brave vigil]

The one in the bottom-right corner of the PCB at the top-center is another option, that's listed at a completely different webpage of WAGO's website, than what I've hyperlinked within #help-with-hw-design message.

[sick ravine]

It's actually a Phoenix contact part lel

I was considering wago connectors for my little motor driver board but the screw terminals are way cheaper and imo better for non-automated assembly/fast assembly purposes

I was a little surprised to find pushbutton terminals that are almost exactly the same size as these small 3.5mm pitch screw terminals tho

[lost portal]

They were developed for things like light fittings, where they often don’t have the space for bigger terminals to replace the screw terminals already being used

[sick ravine]

Eh these ones in particular were yeah

But there's a whole bucket load of push-in spring-button w2b connectors out there now

[brave vigil]

Is https://www.wago.com/global/search?f=%3AresultType%3Aproduct&text=250-* the copycat or the original?

[sick ravine]

Dunno lol

[icy hawk]

bad news, this does not actually have bias power. I turned the volume all the way up and while it can record from something that has line level power it cannot record anything from my Shure MVL. unless there's a setting I'm missing somewhere...

[timber wyvern]

Oof.

[icy hawk]

gonna go with a Audio Technica ATR2x-USB in case this helps anyone else. specs specifically mention that it provides plug-in power

[silk cypress]

How do I check for the presence/absence of the termination resistor in one end of a CAN bus network?

[unique patio]

I didn't know about this, but I see that there is supposed to be a 120-ohm termination resitsor between the two wires of the bus at each end. Are you asking about checking for this resistor programmatically or with a multimeter?

[silk cypress]

Checking programmatically is the goal.

[tulip swift]

You can't. Any specialized tools for this are meant to check the network wiring and termination but not to be used while the bus is up and running. Basically either your Canbus works or it gives you issue and you need to power it off and check it with a multimeter to see if you measure roughly 60 ohms. Closer to 120 probably means only one resistor is on the line and values well under 60 ohms probably mean there are 3 or more terminals hanging out there.

[silk cypress]

But there are voltage levels on the CAN H and CAN L when the bus is idle or not. With or without termination, these voltages change.

[unique patio]

what is your use case for checking for the resistors? Do you want to report bad wiring to the user?

Are the devices on the bus open-drain when not in use? We have code for I2C that checks for pull-up resistors. We pull the bus line down and see how fast it recovers. Mostly we are looking for totally missing pull-ups.

if you know that the devices on the bus are electrically disconnected (open-drain) you could maybe use the ADC to measure some characteristics. You could pull a pin one way, and drive the other pin strongly the other way and do a voltage divider measurement. But it would need calibration. The internal pull resistors are typically not well spec'd (like, say 60k-85k on an RP2040), so you would need to calibrate on a known good setup.

[tulip swift]

The bus voltages ideally should be logic high or low but you can only see and measure when those logic levels change state vs actually seeing voltages. As Dan mentioned, maybe you can try to measure the period between a low to high transaction of the signals, but that assumes things like at the time no other device are trying to signal on the bus too. What if two devices pull the lines low and yours let's go, starts to time things and the other device continues to hold the bus low? Now your code thinks there is a terminal resistor issue?

[timber wyvern]

The test seems like it would be fairly simple, you're checking if the lines have two parallel 120 ohm resistors on it (terminates at both ends of the bus), so you'd need to setup an ohm-meter circuit, if it reads 60 ohms both terminators are in place, 120 ohms means one terminator is missing, and infinite resistance means there's no terminators in place. Put a little padding around that so say 55-65 and 110-130 and you can detect 'unknown state' if it's not in those ranges but also not nigh-infinite.

[tulip swift]

But @silk cypress wants to do this detection programmatically which I assume means being able to detect the terminators and then continue with his code now communicating on the bus. Once you start adding ohm-meter circuits to the bus, you might be changing it so the communication is hindered. If you only want to check the bus for resistors, your trusty Fluke multimeter is all you need.

[timber wyvern]

There's plenty of 'use a RPi as an ohmmeter' projects, so it's easy to automate and a relay or something to enable/disable the extra circuit bits.

[unique patio]

I was suggesting using the internal pull up / down resistors as part of the ohmmeter circuit. I think you would not need anything else in that case. you would be measuring the voltage at a voltage divider that is partly the external resistors and partly the internal resistor.

[twilit crane]

Hi folks - anyone know if the single ended nOOds filament can be trimmed from the non-contact end?
https://www.adafruit.com/product/6133

[timber wyvern]

Nope, can't trim noods at all, big fat first 'con' on the main learn page for them.

[twilit crane]

thank you!

[native lodge]

how would I connect a lipoly battery to my qt py rp2040
i was thinking of using "https://thepihut.com/products/lipo-amigo-lipo-liion-battery-charger" but i didnt know how and was wondering if there was a way to do it easily on the pcb i am putting the qt-py on, along with some sensors and a screen.
i wanted to use "https://thepihut.com/products/500mah-3-7v-lipo-battery?variant=42143258280131" as the battery.
how hard would it be to make it charge with usb c OR "https://thepihut.com/products/wireless-charging-module-5v-1a?variant=27740714769" ?

i was also wondering if there was a Fritzing file for "https://thepihut.com/products/6-dof-sensor-mpu6050?variant=27741016337" and how i'd connect it.

[tulip swift]

One option for a charger is the Adafruit BFF charger https://thepihut.com/products/adafruit-liion-or-lipoly-charger-bff-add-on-for-qt-py which is designed to stack with a QtPy/XIAO board. The thing you have to watch for though it that if you use the provided headers and blindly solder them on, you effectively lock out all the pins. So to interface other wiring connections to the Qt Py be sure to plan ahead before adding the headers to the BFF. You might want to leave some positions empty for wires. The BFF minimally only needs the Vbus and Gnd pins but also uses an optional signal to a GPIO for monitoring the battery voltage. But if your sensor and a display all need pins, the BFF might just get in the way. You can still use it without stacking it of course.

No experience with the wireless changer but looks like it can put out as much as 5V so it looks like it is really more of a Voltage transmitter rather than a charger, meaning you would still need to put in a proper lipo battery to limit the voltage to the battery to 4.2V and control the charge rate.

[native lodge]

my concern with the bff is that it may be difficult to use it when the qt-py is on a pcb

[tulip swift]

Yes, the BFF is really intended to be stackd on a Qt Py and the pair kind of the stand alone thing rather than being mounted on a PCB. Are you custom making a PCB? If so, could you simply add the lipo charging circuit to your PCB design? If you look at the ones used by Adafruit and others, the lipo charging is very simple where you can basically just follow the recommended circuit in the datasheet.

[native lodge]

i am trying to make a custom pcb using Fritzing, but its my first attempt and making a physical pcb, do you know a good circuit i could use?
i also want to give it a gyroscope which would be cool to integrate. i can probably work out how to add a basic temp/humidity sensor myself.
thanks for the help

my current plan is to use:
https://www.adafruit.com/product/5743 attached to pcb for control
https://www.adafruit.com/product/5297 as the display, again attached (and an led or 2)
some kind of light sensor
Ultrasonic Distance Sensor like the "HC-SR04"
https://thepihut.com/products/6-dof-sensor-mpu6050?variant=27741016337 or similar

[tulip swift]

Looks like everything except the DOF sensor supports the Stemma connector which means you can hook all the baords up to each other with simple Stemma cables. If you get the Adafruit version of the DOF https://thepihut.com/products/adafruit-mpu-6050-6-dof-accel-and-gyro-sensor-stemma-qt-qwiic then you get a stemma connector on that as well. Instead of daisy chaining them all in a long line, you might want to pick up an Adafruit Stemma Hub https://thepihut.com/products/adafruit-qwiic-stemma-qt-5-port-hub

[native lodge]

i was hoping to have a go making a pcb to learn and make it neat and compact... when everything has its own pcb i feel like it could get rather bulky.

idk tho

[native lodge]

i dont know how i'd work out the voltages components etc needed for the imu and batt charger, is there a way to put a prebuilt one directly onto the pcb...

[brave vigil]

i was hoping to … make it neat …
At least for the goal of neatness, https://learn.adafruit.com/swirly-grid is an option to consider.

[humble portal]

Hi Folks. Trying to use an S-35710 Low-Power Wake Up Timer (https://www.adafruit.com/product/5959) to turn the power on/off on the stemma i2c-connected devices on a weather station. Have not figured out how to use the OUT pin to turn power on/off on the i2c devices. Any suggestions?

[torpid trout]

Just wondering: Is the issue that OUT is a logic-level rather than power-level output?

(What are the actual I2C boards you're trying to power down?)

[latent jungle]

Since the OUT signal defaults to LOW, the best option would be to use a P-Channel MOSFET as a high-side switch for the I2C devices. They'll default to ON when the S-35710 powers-up and then you can use the timer to turn them off.

That said, why wouldn't you just use the microcontroller to turn the I2C devices on and off?

(As the description says, the board is probably better suited as an external watchdog timer and not a power controller.)

[humble portal]

It is a logic-level output

[humble portal]

I would love to be able to do that, but apparently CircuitPython does not have that capability (yet). BTW, the S-35710 was recommended as a solution to this problem, but I probably did not provide a clear enough description.

[latent jungle]

What do you mean? You just need to use a GPIO pin to toggle the P-FET (or load switch)'s Enable.

It's how I do it in CircuitPython

[humble portal]

Losing me a bit. Do you mean toggle the power pin on the I2C bus? Can you point me to an example?

[latent jungle]

No. The I2C devices have to be connected powered by a transistor (or load switch) that you toggle on and off.

[humble portal]

Got it. Thanks.

[latent jungle]

I am not sure where the suggestion for the S-35710 came from, since it would still reply on "waking up" the separate power source for the I2C devices.

I suppose you could use it to power down the regulator on a feather board via its Enable pin. However, the product description explains why that is tough. And it would mean having a separate power source for just that timer board

[humble portal]

Right. The new products video that @mint moth provided with the S-35710 notes that you cannot control the power to the power with this wake timer.

[latent jungle]

You can't, if the 35710 is powered by the same feather ...

[humble portal]

I suspect this means that I cannot use stemma to chain the i2c sensors that I want to control the power to.

[latent jungle]

Depends on the board. Some boards power stemma connectors through a dedicated switch.

[humble portal]

It is an RP2040 feather with the RFM95 transceiver. I should also point-out this PR (https://github.com/adafruit/circuitpython/pull/10023) which relates to low-power options to sleep modes and adding the ability to turn-off connected devices.

[latent jungle]

that’s an irrelevant PR. it’s talking about the MCU itself.

[humble portal]

It actually also addresses controlling device power (it is a rather long discussion...)

[latent jungle]

I disagree.

It’s specifically talking about the power domains in the RP2350

not controlling devices on a circuit board, like a feather

regardless, based on the schematic for the RP2040 Feather with RFM95, the stemma port uses the same 3V3 rail as everything else.

there is no way for that board, by itself, to power down external devices (I2C).

[wintry wind]

I know you are looking at a RP2350 and so may not want to consider ESP32-S3 based systems but this board I used for the very reason that you can control the (second) Stemma QT interface -- the first one like so many other boards also powers the main MCU like @latent jungle said.

https://www.adafruit.com/product/6399

Edit: Just for completeness, I had forgotten that the Adafruit standard Feather ESP32-S2/3 boards also let you control (the single in this case) Stemma QT via a pin.

[tulip swift]

One would have to do a little bit of wiring and circuit building, but you could alway unplug the 3.3V wire at the stemma than connects to the feather and then have power applied to that wire by a little switch circuit controller by an GPIO pin. Might need a bit of perf board to build it on.

[humble portal]

Thank you. I wondered about just this option (cut the 3V wire after the stemma connector on the board and insert a switch controller). Interesting option.

[tulip swift]

If you want to play with it, you can use an exacto knife or something tiny to left up the little ear on the connector of the STEMMA cable and slide the wire out. It helps to tape down the other three wires to the desk so you can better use both hands to lift the ear and slide out the wire. If you wanna avoid cutting it that is.

[humble portal]

Yup...some electronics surgery involved.

Would it be crazy to suggest that a useful breakout board with a chip that can do the power switching, through stemma i2c, would be a useful little breakout board to have? Basically something that you could insert in your stemma i2c chain that you can control to turn on/off the i2c devices further down the chain.

[tulip swift]

It does seem like a good idea especially for all the battery operated projects.

[tulip swift]

Maybe they can design another Stemma Hub board but have it be actively controlled. Still uses stemma connector but then also a GPIO input hole to shut down the 3.3V at the board and everything after it. If you don't use the GPIO shut off pin then it operates as just a hub.

[humble portal]

I do a lot of low-power (battery powered, often remote) projects that would benefit from such a breakout. I kind of wonder if such a thing already exists?

[humble portal]

Exactly.

[humble portal]

Thank you for the interesting ideas @tulip swift . I appreciate the discussion.

[knotty tiger]

switching off (disconnecting power from) an I2C device while still having voltages present on the I2C bus it’s connected to is not a great idea, unless it’s specifically designed to handle that

[tulip swift]

The voltages on the pull-up would be shut off as well. This is how most of the Feather boards do it. If another device was powering the bus, that could be a problem, but if all the devices are off that same Stemma connector, then they would all be powered down including the bus.

Where VSENSOR is Vcc for both the pull-us and powering any device on the Stemma bus.

[knotty tiger]

i think you can’t reasonably control the power switch from the same I2C bus that you’re trying to switch off

unless it’s also disconnecting the I2C signals

[wintry wind]

Now that I think about it I had this very same problem with the MagTag. During deep sleep the I2C bus stays powered (and doesn't have its own regulator like a number of the ESP32-S* feathers) so I couldn't power off completely my sensor. The green light (I know I could have cut the trace) stayed on showing me it was still powered during deep sleep. Just dropping in one of these newly imagined i2c bus extenders would have solved my problem.

And I don't see why one couldn't design an i2c bus that powered down everthing beyond it but still was powered up itself.

[humble portal]

I believe that would be ok. The use case I have in mind is a system that sleeps most of the time but wakes periodically to take some measurements. When it goes to sleep, shutting everything down, including the i2c signals, would be just fine.

[knotty tiger]

such a device might already exist! i haven’t investigated thoroughly

[humble portal]

If it does not exist, is it worth suggesting to @mint moth that such a breakout would be a useful tool for the low/battery powered projects that people like myself do?

[wintry wind]

Actually it does exist if you want to control via a digital pin. I don't think it supports controlling over i2c though. https://www.adafruit.com/product/4756

I should say you should test and verify for yourself. I'm just going by the enable pad on the board.

[humble portal]

Oh yes. I use one of these i2c extenders in my current weather station.

[humble portal]

Thank you @wintry wind and @tulip swift for the discussion. I unfortunately need to sign-off, but will dig into this further. If you have any further flashes of insight, please post!

[wintry wind]

I use it on a my sensor system for my network equipment rack. I measure intake and exhaust temps on my heavy duty PoE switch that has the biggest thermal load and then also the rack exhaust fan temp. I had long enough wire runs that I got that (and a spare) to use on the bus.

Take care, Jeff. Glad to have the discussion.

[tulip swift]

That is why I suggested a seperate input pin on an active hub. It would be used to shut down the voltage right at the hub from going any further. So your hook up would be the stemma cable and one extra wire.

[tulip swift]

No that is something different. That is to enhance what the passive pull up resistors do when driving long distances. It doesn't support shutting off the bus.

[wintry wind]

Thanks for the clarification.

I have a hw question. One of the cool things I liked about the Unexpected Maker Feather S3[d] was that it has a pin that indicates if you have USB connected. I've heard one could fashion something like that with a diode and resistors for something like the Feather ESP32-S2/3 boards. Is that documented anywhere? I guess you might also be able to use level shifting chip.

[orchid galleon]

The I2C isolator (4903) or perhaps the bus buffer? (5159) could be used as a base for powering down a bus segment, but I'm sure there is something more appropriate to the purpose. I'm using (4903) with a slide switch for manual control.

[knotty maple]

threw together a buck converter tonight based on TI TP62840DLCR. let me know if you spot something wrong!

[tulip swift]

Well I hate the way you drew most of that, especially the input caps, the component & connection choices look like they agree with the datasheet.

[knotty maple]

Thanks for taking a look and thank you especially for checking the components and connection choices. I'm still refining my schematic style so I'm definitely open to any tips or examples you think look more organized

[tulip swift]

For easier reading:

Put both input caps side by side below the line.

Move or hide the "GND" labels on pins 1 and 3 because it looks like you labelling pins 2 & 4 as "GND".

The "3V3" label on pin 8 is kind of hidden in the intersection. I would draw a line out the right and then move the label out there. Or at least slide the 3V3 label left, but I think if it sticks out to the right it is clear it is your 3.3V output to other parts of the schematic. Adafruit does this a lot making it hard to find the EN line on the LDO of many of their schematics so when you are first trying to find where the EN pin goes to it takes a while to spot it kind of tucked in with the other components.

[knotty maple]

Ah I see, thank you! Those would clean it up a bit for sure

[humble portal]

Thank you for pointing-out those two i2c breakouts. Indeed 4903 is really close to what I need, I think just missing an enable pin to toggle the connection (which would I think include power) between the two sides. The bus buffer (5159) looks interesting also.

[rancid lagoon]

is there a generally accepted way (as in most things will just charge at the allowed speed) to signal either 5v 0.5a, 5v 1a, or 5v 2a source power available in a USB-A port for charging stuff

[tulip swift]

I don't think so. PD chargers start at the basic 5V and to the best of my knowledge can supply whatever the max current is for 5V. I am not aware of any signalling to restrict the output current to something lower. So if your USB A port can do 2A then if your load wants 2A it is going to get 2A.

[latent jungle]

Quick Charge is a method for higher current. But it was proprietary, so I am not sure how difficult it would be to implement.

[unique patio]

not on USB-A

[lone basalt]

Your device will only draw what it needs current wise though, so you can kinda soft limit by optimising your part selection for your peak wanted current draw

[rancid lagoon]

Then what about replicating the apple 5W and 12W power things

[rancid lagoon]

Wrong end, the question is about the power source

[lone basalt]

I'd have guessed it's either a custom protocol or an actual hardware limitation that they can only build by owning both sides of the stack

resistors limit current

[tulip swift]

Replicating them how? Are you just talking about 5W and 12W USB chargers?

[rancid lagoon]

The 1a and 2.4a limits they have

If there’s no generally accepted way might as well use a way that works well enough for me

[tulip swift]

The limits are due to the electronics inside them being capable of only those currents. It isn't like they are secretly 5A devices that are pretending to be lower current devices and are artifically restricting the current. What is it you are trying to do?

[rancid lagoon]

Tell the connected device to only use 1a or 2.4a max from a 5 volt power source

[tulip swift]

OK but why? A device typically draws whatever current it needs. With PD devices the device can indicate it is able to accept bigger voltages, but voltage supplies don't normally "command" a device to use less current.

[rancid lagoon]

If everything connected took as much as it wanted that would be more than the supply could do

They have to do something, don’t they? Otherwise anything on a charger too small would just overload the charger and have it cut out

[tulip swift]

Yes, generally that is the way it works. If your load (device) requires 3A and you plug it into a source only able to supply 2A then it isn't going to work. Depending on the source it may just end up reducing the voltage severly trying to supply the current or a fuse will blow or maybe it will overheat. You can not generally just have a device use less current on demand to use a lower wattage source.

[rancid lagoon]

The phone or tablet or whatever doesn’t require the 3A tho, it’ll just charge slower if there isn’t enough juice available

[tulip swift]

Yes but that is because of the charging circuit in the phone/tablet/etc. where there is usually some current control involved on the device side of things. And if it draws too much current and the supply voltage output begins to sag the charger will likely be running at some lower duty cycle to keep the charge voltage to the battery correct (e.g. when the charging voltage falls, the charging circuit will typically stop charging until it comes back up again).

[unique patio]

USB devices report how much current they draw when they enumerate. This is not necessarily accurate but can be an upper bound. But this is on the device side, sent to the host.

bMaxPower, in units of 2mA

[tulip swift]

Yes I believe on the source side, since it is pretty dumb in that the source doesn't change how much current it supplies (and probably can't anyway), it simply gives the device a yay or nay on whatever current the device wants to set itself up to draw. Or put another way, device indicates "I can draw 3A" and the source responds "Nope" and then device will likely default to some "safe" lower current. Or the source says "Yep, I have that" and the device can turn on its high current draw. In all cases, regardless of what the device draws, the source doens't do any sort of current control.

[latent jungle]

There is no open way to do it. There are multiple proprietary methods, like Quick Charge.

And whatever Apple is doing.

Keep in mind, most cables aren’t designed to deliver more than 900 mA (if that.)

I suspect some of these proprietary methods do a voltage drop measurement, like what USB-PD supports, to know if a cable is too resistive.

[brave vigil]

The closest thing would be USB Battery Charging

[lucid dagger]

question about the Grand Central M4 Express board: is this pin an extra power pin for adding a new power switch?

i opened the project in kicad so idk how much i can trust this board setup, but it looks like a normal header hole for mounting a new power button for the likes of a case-mounted switch or something?

it doesnt seem to be listed in the documentation as anything of note though https://learn.adafruit.com/adafruit-grand-central/pinouts#power-connections-3015119

ah i suppose it doesnt matter for me, since ill be plugging this in via usb anyway and that power button is only for the DC jack..

[lucid dagger]

unless there are somehow data-only powerless usb cables 🤔

i wouldnt put it past the usb consortium

[latent jungle]

I don't think there is an official standard for such a cable. There are adapters like this one that you could put inline with a cable: https://www.adafruit.com/product/5972

[lucid dagger]

oooh!

[plain field]

Hi, I first posted this on an unrelated Discord server whose owner builds Feather-compatible boards, but it would be relevant here..... a general comment about Feather form factor and enclosures:
When designing for battery power, it can be difficult to fit into tight enclosures - because the battery JST connector exits the PCB sideways and the wires take up space and shouldn't be pinched. In my project, I found it's possible to clip off the JST connector and resolder it vertically, maintaining polarity.... but mechanically it sucks.
However, I had an idea: if Feather PCBs snuck unpopulated VBAT and GND thru-holes underneath the battery JST connector, wired appropriately, perhaps it would be possible to resolder the JST connector pins in these holes, placing the battery connector in a vertical orientation? Would there be drawbacks to such a hidden feature?

[unique patio]

Boards often have other traces routed in the area. Note that the BAT and GND pins at the edge of the board are electrically the same as the two pins on the JST connector, so you could wire up a battery directly or with a connector using those pins.

The choice of which way the JST is oriented depends on the use case. Feathers were designed to be stacked with FeatherWIngs, so a horizontal exit made the most sense.

[plain field]

Makes sense, thanks for the explanation. I like having the detachable battery, so perhaps putting the female JST connector on a short wire and soldering that into the board would be a good solution. Have you considered selling a product like that? Re: wings, so many projects are possible with I2C/STEMMA now, I don't use them much anymore, but I understand the accommodation.

[unique patio]

https://www.adafruit.com/search?q=jst-ph+2-pin

In particular: https://www.adafruit.com/product/3814

Also see the breakouts, including one with a switch, in the Search Results above

[plain field]

Excellent, thanks!

[unique patio]

Double-check the polarity. Our stuff should all be the same (red/black on the same pins), but some third-party batteries and cables have them reversed. There is not a uniform standard.

[plain field]

Noted, I always buy the adafruit batteries for that reason.

[icy hawk]

update on this, it might work but it's a TRS cable and my mic has a TRRS cable lol. so I might just have to buy an actual audio interface. or actually get a different mic that just has TRS

anyone recommend a good mic I can wear under a bike helmet? that can block wind?

[indigo inlet]

An adaptor like this might work... perhaps try one but be ready to return it if it doesn't work? https://www.amazon.co.uk/dp/B073ZDDTH2

[indigo inlet]

If I read correctly, you'd like to connect a Shure MVL (which has a TRRS output and needs bias power) to a Raspberry Pi 3 A+? From reading around, seems surprisingly tricky. Perhaps MVL into Rode VXLR+ https://www.amazon.co.uk/VXLR-Minijack-Adaptor-Power-Converter/dp/B071LNDKBJ/ (which is mentioned at https://sound.stackexchange.com/questions/50971). Would probably need a TRRS to TRS adaptor between the two. At the other end of the Rode, we then need XLR (with phantom power) to USB for the Pi. Perhaps Shure MVX2U https://www.amazon.co.uk/dp/B0C8W8D4PD and then a USB-C to USB-A cable. Or Pyle PDUSBPP10 https://www.amazon.co.uk/dp/B07MTGT27G and then USB mini B (?) to USB-A cable. Or some simpler XLR to USB adaptor without phantom power, plus a box that injects phantom power, eg Vonyx VDX10 https://www.amazon.co.uk/dp/B0CFYRDVHP All a bit disappointingly expensive and complicated.

Or build the circuit that Shure give for the MVL https://www.shure.com/en-US/docs/guide/MVL ... on the right hand side here: a TRRS socket, two resistors and a capacitor... and then you have an audio output that you could connect to a more standard "mic to USB" interface. You would need to provide 2.7V DC for the bias power. The Shure page says this would take 245 µA. I lack experience here... I'd guess that could be generated from the Pi 5V output via a potential divider... or perhaps a buck convertor would be needed (but that doesn't seem like a lot of current, so perhaps not?).

[icy hawk]

yeah I'm already on my 3rd failed adapter. it's wild that building the circuit myself is within the realm of the "easier" solutions at this point

which is why I was considering just switching out the mic, possibly a headset style dynamic mic that hopefully doesn't require TRRS

except according to the diagram you posted the MVL doesn't even use everything in TRRS? it could even be TS if it wanted??

I could cut open a 3.5 mm plug/jack extension cable and fudge the wiring to make a custom adapter

[indigo inlet]

Oh yeah! Lol! (That diagram is from the Shure MVL page by the way.)

I think you'd still need a TRRS socket... think a TS socket would only connect to the tip and perhaps the top sleeve of the MVL jack

Are you comfortable with soldering?

[icy hawk]

yeah

[indigo inlet]

Cool... so one option would be to get a TRRS socket (from RS / Farnell etc) and a short length of audio cable (one core plus shield), connect the TRRS socket to a breadboard and try and build that circuit that Shure describe

[zinc schooner]

Anyone know why my display has these banding issues? I’m using the default test code, alternate image shown because they show up better

This is circuit python qualia on 4 inch non touch screen

please ping me if anyone has an answer

[fresh fiber]

Is there a guide somewhere on creating a simple board? I have some feather wings I would love to create, but have zero idea where to start.

[flat vigil]

https://docs.kicad.org/9.0/en/getting_started_in_kicad/getting_started_in_kicad.html ?

I believe it has symbols and footprints for feather already

[fresh fiber]

Oh awesome!

And anything on ordering assembled PCBs?

[flat vigil]

there is a kicad plugin to export stuff for jlcpcba

[fresh fiber]

Wonderful

[wintry wind]

I'd been wondering the same thing, Thanks @flat vigil

I've been wanting to put together a USB power present circuit for some of my Feather ESP32-S2/S3 boards kind of like the UM Feather S3[D] has on board. I have a circuit design for resistors, diode and maybe a capacitor to safely reduce the 5V to <3.3V via a resistor network and use an analog pin to measure.

However I've been looking at TXB0104 Bi-Directional Level Shifter https://www.adafruit.com/product/1875 and I have a few 74AHCT125 Quad level shifter ICs https://www.adafruit.com/product/1787 already. I believe that these are more 'digital' than the 'analog' resistor network but would they work as a digital input (5V vs not-5V present)?

Sometimes I prefer to buy a small ready made board like the first link rather than build my own circuit for such a simple purpose.

[fresh fiber]

Hmmm, missing something obvious. I get errors in DRC if I place anything on top of the feather footprint. Is there a way to make that the base?

[flat vigil]

what is the error? I haven't actually used it so I'm not sure.

[fresh fiber]

It's a keep out error. I think the footprints are for if you want to add a feather to your board, but not use it as the PCB base. Searching some people have made projects you can copy (I haven't found one for the newest version). I might just need to start from scratch

[flat vigil]

ah, ya maybe

that's make sense

You can right click and edit a footprint for the local board

[fresh fiber]

I know Adafruit uses Eagle, but it would be cool if they had a base project for feathers (wings or boards), so all one needed to do was clone it and add components

[latent jungle]

are you making a wing in KiCad?

[fresh fiber]

That's my goal

[latent jungle]

What's the DRC error you're getting?

In the past, I've used this template: https://github.com/Bucknalla/kicad-feather-wing-template. (I ended removing, changing, or moving all the silks to the back side).

The schematic is a couple of connectors and the PCB file contains an edge cut layer, holes, and headers in the correct positions.

If you're using a footprint, make sure it doesn't contain a polygon on the courtyard layer.

[fresh fiber]

I'll try that template and see what happens

[fresh fiber]

That's what I was looking for. Is there an easy way to remove the headers and just have the pin holes? Or do I need to make my own template with pads for that?

[latent jungle]

delete them in the schematic. and press F8 to update the PCB

use the selection filter in PCB (bottom right) to only select text to highlight and delete silk

[icy hawk]

ended up going with this and my mic finally works!

splitting TRRS and providing plug-in power were the two things I missed earlier

[indigo inlet]

Yay! 🎉

[fresh fiber]

2 questions on Kicad if anyone has a moment:

If something has a ground connection and that ground is over the ground plane, do I need to actually connect it?

I've seen on some feathers there is a vias that connects the ground pin from the bottom of the board to the top, but doesn't go anywhere. I assume this is fine because it connects to the ground plane? (And thus can ignore the warning)

[flat vigil]

1, nope. It should be connected after you fill the plane (ctrl-b iirc). 2. I usually add a via so drc is ok. even though the pin hole itself should act as one too.

[fresh fiber]

Okay, I still get warnings for things not connected (and it shows the light blue unconnected lines between all of them)

[knotty tiger]

how are those ground connections shown on your schematic?

[fresh fiber]

🤦‍♂️ I just realized that the ground plane was on the back, so wasn't connected to anything...

[fresh fiber]

Also curious, why are the 2 mounting holes on the USB side of a feather created like pads, but not connected to anything?

[unique patio]

I don't know -- that is an AaE kind of q

[fresh fiber]

Oh, not familiar with that Acronym. What does it stand for?

[tulip swift]

Annie an Elephant kind of queen. Duh! 😁

[unique patio]

Ask an Engineer - our Wednesday video show, which has become rather irregular as of late.

[fresh fiber]

Ahhh, I will ask the next time I'm there, and add them in the meantime

[fresh fiber]

Does anyone know what SD card reader part is on the feather boards? Trying to find one I can use

[unique patio]

you mean the card socket? Do you want the alternate-action push-push one or the simple friction fit one?

https://www.adafruit.com/product/1660

[fresh fiber]

That's the one! Do you know if there's a matching part on JLCPCB? I am not good at SMD soldering and am hoping to have them do my component assembly...

Looks like they can order it, but that's quite the price increase... https://jlcpcb.com/partdetail/AdafruitIndustries-1660/C17490034

[unique patio]

JLCPCB sell a whole bunch: https://jlcpcb.com/parts/componentSearch?searchTxt=microsd

you could just pick one to use and use their footprint design, etc.

there's no strong reason to match exactly what we picked. You want to use one of those and import the part into kicad or whatever

[fresh fiber]

Are they pretty much all the same? So just find one that fits?

[unique patio]

Limor discussed picking one on at least of Desk of LadyAda

[fresh fiber]

This is very new territory for me. I've bought hundreds of dev boards from Adafruit. Time to start building what I created....

[fresh fiber]

Do you know what episode?

[unique patio]

no... on the phone, I'll get back to you somewhat later

[fresh fiber]

Absolutely no rush. Thank you

[unique patio]

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

also of interest: https://www.youtube.com/watch?v=kuCjS1MZvbU

[fresh fiber]

Thanks!

[pale grove]

Bit of a specific question.

For example on the RP2040 propmaker feather there are the amp gain pads.

Assuming I don't have these pads - is there a way for me to verify what the gain of the amplifier is at the moment?

With a multimeter for example

[unique patio]

What the gain pin is connected to determines the gain. Are you saying you're using the same amp (MAX98357) elsewhere? You can measure the voltage on the gain pin, but one of the settings is "unconnected", which is somewhat hard to measure

[pale grove]

Yep using the same amp on a custom board

But somehow my speaker is not as loud. I wonder if I put a trace wrong somehow

[unique patio]

if you connect a GPIO to the gain pin, you could control it yourself: low, high, open-collector (as "unconnected")

[clear matrix]

Hi guys, I searched the Adafruit web page, can it really be that there's no official piezo electrical vibration sensor board?

[latent jungle]

Almost any piezo element can be used as a sensor. They generate voltage when they vibrate. And they vibrate when voltage is applied.

The cheapest of buzzer can be used. However, you might need to remove it from its casing.

[unique patio]

can you give one or more product examples of such sensors?

[clear matrix]

Like this, but cool. And with stemma.

[unique patio]

You could send a product suggestion to [e-mail]. We do have that ribbon piezo sensor: https://www.adafruit.com/product/4931

I am not sure why we don't have a sensor like this, but maybe because an accelerometer often serves the purpose. How would that not meet your needs?

[clear matrix]

I want to make interactive airsoft targets, sometimes the energy of the hits will be tiny and, I assume, very short. Not sure an accelerometer could keep up.

[unique patio]

maybe worth trying. Most have tap detection.

but i can see how the above may be closer to your needs.

[clear matrix]

Maybe the made in China one is fine, but if there was an Adafruit version, I'd probably go with that first.

[dense blade]

Hello! I'm trying to figure out a good DC-DC USB-C power source that sits between my 2s LiFePO4 batteries (so about 26V) and laptop, ideally providing 100W through the USB-C ports. Adafruit has a power sink board (https://www.adafruit.com/product/5807), but I can't find anything similar for a power source.

I've found a blog post with someone planning to make the kind of board that I need, but I don't really understand how the ICs on this board work and it sounds like he never actually went through with the project: https://community.element14.com/technologies/power-management/b/blog/posts/100w-usb-type-c-power-delivery-source-getting-started

Is what I'm looking for complicated/niche or is there a solution that I'm missing here?

[icy hawk]

https://cdn.discordapp.com/attachments/873752706983030804/1468426420769521766/IMG_5395.mov?ex=6983fa2c&is=6982a8ac&hm=bdcd0d0165e4d101dc9b928d4db149a531016df378fe91bbe41c4cc71713020e&
I have this LED matrix panel and I'm using the Adafruit LED matrix bonnet with a Raspberry Pi 3 A+ and the Adafruit recommended library and I am successfully able to power the panel and display images. however the top half is showing only blue LEDs, as you can see in the video. I ran several tests (this video is just one of them) and no matter what I cannot get the red and green LEDs in the top half of the panel to turn on, just the blue. I suspect this may be a defect in the panel since full color works on the bottom half just fine. but it could just be a configuration thing. currently hacking at pixel-mapper.cc to see if I can make a custom color channel mapping. any suggestions?

https://cdn.discordapp.com/attachments/873752706983030804/1468426420769521766/IMG_5395.mov?ex=6983fa2c&is=6982a8ac&hm=bdcd0d0165e4d101dc9b928d4db149a531016df378fe91bbe41c4cc71713020e&

[copper marsh]

I have the Neotrellis and a bunch of Neotrellis blocks. I’m excited to put them together, but I also don’t want to solder them all together into one monolithic unit. I was wondering if anyone had success putting magnetic connectors onto them? If so, were you able to get the exact spacing right, or did you have to use wires to connect the magnetic connectors?

Alternately, what is the estimated difficulty in just using a 3D print to hold pogo pins in place and solder pogo pins directly to the unit, and then using a 3D printed case to glue magnets into? Are there any particular pin types I could look up that would best suit this type of installation?

[unique patio]

The connection pads at the edges are all just wired in parallel, and the JST 4-pin connector in the center is also wired to those pads. So you could get multiple JST jumpers and connect them in parallel. You don't have to use the board-edge fingers.

You can secure the NeoTrellis boards in position using the mounting holes and some screws. You could 3-D print a backboard with screw posts, or screw them to some kind of rails.

[icy hawk]

update on this, I tested with a different ribbon cable and completely different make/model of panel; same result. looks like it's a hardware issue with the bonnet

[coarse lark]

Any recommendations on a 5V LDO?
Preferrably smaller and more efficient than the AMS1117.
Its for powering servos on a robot controller board so it needs to be able to deliver at least ~2A.
Or should i look into a proper power management chip/buck converter?

[latent jungle]

2 amps is a lot for a LDO. And still a lot for a traditional Linear.

[coarse lark]

yeah

https://giphy.com/gifs/meme-pray-guy-praying-OmqRHI3lSX7jR97WjZ

i knew this would happen one day
where i have to drop the LDO and design an actual power circuit

[latent jungle]

there are off-the-shelf switcher modules

[coarse lark]

i guess i could integrate one into my design
why reinvent the wheel they say

[unique patio]

https://www.pololu.com/category/136/voltage-regulators has a lot of nice small efficient regulators if you don't want to do your own

[icy hawk]

found the actual problem, the bonnet was shorting to the USB A enclosure on the Pi. added some electrical tape to the enclosure and now it works. for those in the future searching here are some keywords: RGB LED matrix panel bonnet raspberry pi half one color

[lyric nymph]

Does anyone know what the maximum refresh rate for the eInk displays typically is? For example, https://learn.adafruit.com/adafruit-2-13-250x122-quad-color-eink ?

[flat vigil]

Two color can be in single seconds range. Multi-color takes 10s of seconds usually

[solar sun]

I have one of these. I just powered it on to check. Took around ~15s

[lyric nymph]

!! Holy moly that's slow. Ok, thank you for the help!!

[tulip swift]

My 2.13" Tricolor display refresh (~12 seconds) https://imgur.com/a/refresh-time-of-typical-2-13-tri-color-epaper-display-vDOGEpm

[coarse lark]

maybe i should have read this earlier, i spent yesterday's afternoon making this. i'll keep these in mind for future projects.
i scaled down to only 2 servos (my goal is to make a simple robot car with a claw/grabber so 2 servos is enough), i chose the AP63205 because it needs few external components.
It's a 2A step down which should be enough for this project (SG90 servos have a ~650mA stall current so 1300mA for 2 servos + say 300mA for an ESP32S3 with wifi enabled and some sensors, total ~1600mA).
it was hard to choose a power chip because most either couldnt supply enough current or needed more than 10 components to work (i am working with a limited budget rn, a lot are in QFN packages which i dont feel ready to work with).

now im looking into a mosfet for reverse polarity protection on VIN (something with a drain current > 10A, my current candidate is the AO4407)

[jovial yew]

So I'm working on a small project with some very tight tolerances, and I initially had the button attached with wires, but I plan to make quite a few of these, and the wires are not the easiest thing to do. So I'm currently considering an edge-to-edge connection, like in the 3rd pic, but I'm not too sure how to execute it. Does anyone have any suggestions on how to make sure it will fit, or any other suggestions of how to eliminate the wires and make it easier to assemble? Through hole pins won't work since there's a surface mount device on the other side.

[jade wedge]

Quick questions about USB C charging. I'm designing a board that will charge from USB C. Nothing fancy at all, just needs 5v 500mA tops.
Am I reading correct that I want two 5.1k resistors to ground, one on each CC pin on my schematic?
Also, is tying shield to ground appropriate (or at least acceptable)?

[elder peak]

Two 5.1k resistors yes. Don't try to use less than two 5.1k resistors; that's a repeated design failure.

Ground to shield is ... complicated. Someone else can correct me but the least bad approach is probably to just leave them floating?

[limber rover]

yo is this ready for production?

25 kalih choc v1 key switches

2004 20x4 lcd screen with i2c adapter

adafruit s32-s2 feather
lipo batt 3.7V

[cinder anchor]

Yes, leave them floating.

[cinder anchor]

Hard to tell without seeing the diagram.

[limber rover]

you mean the schematic?

[cinder anchor]

Yes.

[limber rover]

hold up ill bring it up

i kind of didnt do it that much

ill finish up

[limber rover]

something liket his

by production i mean its personal use

[jade wedge]

Ok. I just remember trying to figure it out for USB micro, and found pretty much every answer possible. Yes, no, "isn't it obvious", "it doesn't matter", "only if you want everything to explode"

I think eventually I looked at what Adafruit does with their Feather boards, and I think the shield is grounded there.

[cinder anchor]

I don't see anything immediately wrong with it. Did you add ESD protection?

[limber rover]

hmm difficult question

do you see some?

i didnt specifically go into that too much

first time i do something like this lol

[jovial yew]

I hope this is ok to ask here. I just got the first version of this board, and unfortunately the edge-to-board fingers didn't fit in the slots, as the bevel in in the slots was too far out, also, the pad for solder was hitting the top and bottom edges. I updated the design with some measurements I took by eye, but before wasting another 12 dollars on boards, I'd like a second opinion on if it will fit. I use OSHPark if that matters.

[manic magnet]

Could anyone judge my digital (on off) automotive input protection schematic? I don't really know what I'm doing 😅

[tulip swift]

Looks like your zenor is in backwards? Right now you are clamping the node between R28 and R27 at 0.9V.
Your voltage divider on the input will give you, at that same node, a high of 2.77V if the zenor was turned around.
With the two schottky diodes on the left, is the zenor even needed?

[manic magnet]

It's true, I did some simulations in LTspice. Seems I can raise the 3k and I should reverse the zener

[manic magnet]

I changed the design and added a pullup/down option with solderable links, any more thoughts?

[tulip swift]

What is Vcc? Is it different than 12V and from 3.3V? If it is 5V or 3.3V then if you jumpered to Vcc your 12V input then high will cause a "backwards" current flow of a milliamp or two "up" through R33. What is the intended purpose of either of those jumpers?

[manic magnet]

VCC is 12V yes. The two jumpers are to select between a "floating" and GND input or a floating and +12V input as both are common in automotive environments. However I do see now that R25 will likely act as a pulldown when R33 is not jumpered

So I don't think I need the Pull-down jumper

[tulip swift]

Oh I see. You are anticipating your INPUT_FROM_12V to be coming from an open-collector/drain output of something. I can't imagine you would need to pull down. But having a pull up if needed is fine but not the way you implemented it. As drawn above if you enabled the pull-up for a "high" on the 12V side meaning the output of your sensor/device is open and you are depending on your pull-up, 3V3_GPIO is only going to be about 2V (voltage divider of 14.7k and 3k) which may not be high enough. For example ESP32-S3 I believe only guarantees a high detect if your input is at least 0.75x3.3V = 2.48V.

[manic magnet]

I see, so the pullup should be after the voltage divider and on the 3v3 rail?

[tulip swift]

I think you should design the circuit for whatever signal source is on the 12V side. If the 12V "thing" outputs 0 or 12V (push pull configuration) make the circuit a voltage divider. If your 12V "thing" is open collector then you are controlling the voltage to whatup you have it pulled up to and then you only need a pull up resistor on the 3.3V input.

The other thing to be careful of is how your "thing" outputs 12V. If it is a push-pull mosfet kind of thing where one mosfet pulls it to the 12V rail and the other to the 0V rail, then the voltage divider is what you want. On the other hand if it instead uses a pull up resistor inside to 12V and then the output acts as a switch pulling the output low, then you have to factor in the device's own internal pull up with your pull ups. If the thing has a 10k pull-up for example, you again get into the same situation you have above with the 4.7k pull-up on your circuit. The input won't get high enough.

[manic magnet]

I see, that's an interesting thought. I didn't think about that. I want to make my board to control as many "things" as I can, so I can't say for sure what it will be. I think I will be best off removing the pullup, not making this so complicated for no reason, and adapting the external wiring to be compatible with this type of input when it comes to it. Thanks for the advice!

[tulip swift]

You might be able to do it with a simple mosfet circuit. Please excuse my high-tech advanced CAD tool (BIC = basic ink CAD).

Then it doesn't matter if your 12V thing is outputting 12V directly or if it also has an internal pulls up to 12V. Both should work here. Just note that your 3.3V input will be normally high.

[tulip swift]

☝️ Forgot to use the reply option so see above. Also note that circuit doesn't anticipate any of the normal problems you might see in an automative environment so you will probably want to add basic protections stuff (Transient Voltage Suppressor (TVS), MOVs, etc.) around for voltage spikes, load dumps, and such.

[maiden spear]

sorry for reposting, just thought this might be more relevant in this channel.

Hi, I was wondering If anyone had any advice for my project. I am trying to make a custom 4 layer PCB that incorporates GNSS and BLE antennas. I'm driving the Bluetooth Low Energy from an STM32 (STM32WB50CG) and am planning to use a chip antenna. For the GNSS I want to use a Ublox MAX-M10S via a Passive Patch Antenna. I've attached a few schematics and am starting to think about layout, I ideal it need to be quite compact eg. 80 x 35mm ideally (I know ill sacrifice GNSS performance as I dont have a large ground plane but I'm happy to have reduced connection quality as the device will only be used with clear sky view).

Ultimately, I am unsure of the separation that I need between the 2 antennas. Should I have them on opposite sides of my rectangular board? can I place them next to each other (30mm separation)?

edit. If I place them on opposite sides of the board I may end up having the Chip antenna arround 20mm from the USB port. I doubt I would have a problem with 5V line as I can route it far away but would having a large metal SMD device nearby effect the signal

Any help would be appreciated. Many thanks

[unique patio]

The BLE transmission may desense the GNSS receiver. The antenna datasheet you linked to says:

The antenna should be placed at the center of the ground plane with a length and width of 70mm.
Maintaining a square symmetric ground plane shape and symmetric environment around the antenna is
critical to maintaining the excellent axial ratio and phase center performance shown in this datasheet.

I would try to keep them as far apart as possible, and of course you want no ground plane under the BLE antenna, assuming it's a trace anntea.

but I have no particular experience with this

[maiden spear]

Ok! Thank you for your help

[manic magnet]

I thought about this, it would nicely "isolate" from any voltage spikes and such. I'll look into this a little more

[manic magnet]

I simulated a 60V load dump scenario, eventhough it's an electric car and the 12v should be a lot cleaner than internal combustion. The simulations showed that a 60V input over half a second would result in less than a milliamp flowing through the BAT54 diodes, and the voltage remaining at 3.6V. Maybe adding a TVS is a good idea regardless

[bright fiber]

Any chance someone here knows how to edit a part in fritzing? ended up with the wrong version of a part and jsut need to modify hole sizes for the connectors but havent been able to figure out how it works.

[teal zenith]

does anyone know any good sources for small rotary switches? i've been digging all over digikey and it seems like every rotary switch is like 1" diameter. I'd really rather have a small one, and i know they exist because i've seen them in parts bins before

actually i may have just found what i was looking for on tayda ... i thought i'd searched them before but it looks like they have some

[teal zenith]

they have some.. but not sure if they're exactly what i want. i've designed myself into a bit of a corner so my options are limited

[unique patio]

They are available but they're expensive. Start looking at $15 and up (e.g. Grayhill, NKK).

what specs are you looking for? Could you use a rotary encoder?

[teal zenith]

an encoder would work too, but basically it's for a cosplay prop and i need a 3-4 way selector switch, old military radio style with more of a toggle lever rather than a knob, but i don't have much space to put it

[teal zenith]

it may not really be possible, that's right at the edge of the object, the only real way to make it work would be if the switch was like 8mm in diameter which isn't really realistic

[unique patio]

the smallest ones I can find are 0.5", and they are very expensive. You could maybe hide a gear behind there that controlled a rotary switch in a more spacious location. Or use a slim potentiometer or a rotary encoder

that looks like just two positions, is that right?

[teal zenith]

yeah, i'm doing a pretty lose adaptation so ii'd prefer 3 or 4 stops

it looks like NKK and grayhill both have some that might work, they're in the $16-$45 range. the list of parts has so many similar options so filtering through it is a bit rough

[unique patio]

there are rotary dip switches with shafts that are quite tinybut they have a short life expectancy

if you have more room to the right and above you could make something that moves a larger "blade" behind the panel. Not sure how to make that activate a switch.

does this have to pass current or do you just need to read the position?

you could use a rotary encoder and attach a bar to the shaft that hits against stops so you can't move it more than say 1/4 turn

[teal zenith]

no current, jsut reading digitally

[unique patio]

"limited rotation rotary encoder" is the term

but I don't see any that aren't fancy

hencd the idea of attaching something like a pointer to the shaft behind the panel to limit rotation. Or you could do the same on the front: use a pointer knob like above, and add limit stops with studs at the end of the rotation limit

[teal zenith]

hmm.

[unique patio]

also there are potentiometers with multiple detents, like: https://www.amazon.com/Multiple-Detents-Linear-Control-Potentiometer/dp/B0CLBGP2YP?th=1

so could read that with an ADC

[teal zenith]

that is an interesting idea

[unique patio]

i can't find them on digikey

[teal zenith]

would make wiring easier

17mm body diameter, hmm

[unique patio]

does this have to look like something from a movie or something?

[teal zenith]

it's concept art from fallout 4

so it's a pip boy but kind of a weird alternate reality version. i already made a great working one based on the fallout 76, so i want to do something different now, and i've never seen someone do the one from concept art, so it looks like something from the radio compartment of a b-17

[unique patio]

you could also do the detent mechanically with the pointer arm

[teal zenith]

i don't know if my 3d design skills are good enough for that, i can handle big and chunky pretty well but that kind of fine mechanical dseign is a bit outside my realm

that mechanically detented potentiometer idea is interesting, and nkk has two different options that are close to what i need, it's just actually fitting it into the space i have is difficult

[unique patio]

maybe you could use the detent mechanism from a cheap rotary encoder and connect a pot to the shaft

[teal zenith]

hmm, something like that could work. i think need to think mroe about how loosely i want to adapt things, because to make everything fit in an actual wearable item involves some pretty big compromises. i'm not a stickler for accuracy, i'm happier with broad-strokes recognizability, lol, so that makes some of it easier

[unique patio]

good luck! hope this was helpful

[unique patio]

https://www.alibaba.com/product-detail/New-PTV09A-3-Pin-Rolled-Flat_1601289908781.html?spm=a2700.7724857.0.0.2f443b8eQ3gmam 10mm

[teal zenith]

it was extremely helpful, i didn't know about those brands before, so now i've found a number of options that i didn't have before, and the detented analog pot was something i hadn't ever considered before either

<https://www.alibaba.com/product-detail/

[merry folio]

Would something like this suffice? https://tech.alpsalpine.com/e/products/detail/EC10E1220501/
It's a quadrature encoder, so the positioning is relative rather than absolute, and that might make selecting a specific setting tricky. On the other hand, it's very compact.

[jade wedge]

This is reasonable for a USB power bypass, right? This project will charge from USB (VIN here), but also be able to run straight off USB. So the important part is putting a diode between the battery positive terminal, and VIN. Correct?

[unique patio]

but you also don't want to backpower VIN (assuming that's the USB 5VB line), so two diodes to protect both power sources

[jade wedge]

Ah, good point.

[unique patio]

you could use a three position switch and switch VIN or battery, with middle position both off. Then you don't need diodes, but the user has to make the power decision.

[jade wedge]

I think I'll stick with the diodes. I vaguely hope to actually make a product out of this, so I want it to be simple. It's bad enough (to me) that I have a jumper to pick between 100 and 500mA charging speeds.

Also, I was looking at the Feather boards for a bit of guidance on designing this. Is the only reason for a schottky diode the lower forward voltage?

[unique patio]

yes, and also note the Feathers specifically don't support external non-USB 5v power or external 3.3V power. https://learn.adafruit.com/adafruit-feather-rp2040-pico/power-management especially last section

[jade wedge]

Ok. It makes sense. A basic silicon diode has a drop of about 0.7V, which is a lot when you're talking about potentially only 3.7 to start with.

Ohh... Also I see the feather is fancier and uses a transistor or mosfet to entirely disconnect the battery while it's plugged in.
This is why I really appreciate the availability of schematics.
I think I'll just go with a dual diode package though. Simple for me.

Yessss... There we go.

[marble gazelle]

Schottky diode has a lower voltage drop

oh, you have it now

[jade wedge]

Yeah, since the schematic is really only for my purposes, I wasn't worried too much about getting the exact correct symbol.

But more importantly, as danh pointed out, two diodes, one for each source.

Now to shove it in... somewhere.
I can make room. It's not a problem.

[autumn snow]

Howdy y'all, I'm creating a custom board design with a charlieplexed LED array driven by the same IS31FL3731 driver IC as the CharlieWing (and other products) and want to make sure that my array is compatible with the companion Adafruit GFX arduino library.

The library implementation suggests that the desired width and height of the led display can be specified. There seems to be some hardware-specific stuff in the library's drawPixel logic, at least for the 16x9 driver breakout.

I intend on going from the 15x7 and 16x9 matrix sizes in Adafruit designs to the 96 LED matrix size (12x8) seen on the Arduino R4 and recent UNO variants.

My thought process so far is as follows.

The CharlieWing appears to use CA1 through CA9 and CB1 through CB8 for 105 LEDs.

Generally in charlieplexing, the use of n=11 tristate IO means n^2 - n = 110 LEDs, with the driver IC at n=9 per block means 72 LEDs per block as confirmed in the driver IC datasheet.

On paper then 72 of the 105 LEDs are charlie-plexed on block A and then block B is used at n=8 for 56, which makes me believe there are 105/128 addressable with 23 unmounted LEDs in the mapping.

I imagine I can start by physically 'unmounting' 9 already addressable within the 15x7 CharlieWing instantiation and running guess-and-check.

But ideally I can make a 12x8 matrix directly. Could this size let me cut down on the use of block B on the IC, with 96/114 (18 unmounted LEDs) or am I just going from 23 unmounted to 32 if I want to continue using Adafruit's library?

[jade wedge]

Before I send this off for manufacturing, does anyone see any egregious problems with this design? There is stuff further off to the left, but it's part of an existing design that I've already built, so I'm not concerned about that. I'm mostly checking on the charging (upper half) and boost (lower half) sections.

[spiral pagoda]

What's with the ground plane? Don't you want the mounting pins grounded?

[jade wedge]

For the USB socket I have not been able to figure out if I should, shouldn't, or it doesn't matter.
For the switch, I just didn't think to.

[spiral pagoda]

I'm assuming you're ok with no mounting holes and minimal indication LEDs

Maybe you're right that it doesn't matter if the USB or switch are grounded. It just struck me as odd

[jade wedge]

Yeah. The board is for driving LEDs, so a power indicator seems redundant. Also I don't really have an enclosure design as it's intended to be a very general purpose design for costumes and props.

[spiral pagoda]

Are you planning on soldering to those pads on the back side of the battery JST connector?

They're a bit big for test points, but the ground might be tricky to solder to with its solid connection

[jade wedge]

Pads are for if someone buys one, and decides to get a battery that doesn't have a plug.

[spiral pagoda]

Ok. Consider thermal relief on those pads

[jade wedge]

Will do.

[spiral pagoda]

Good job keeping your inductor loop tight

How much current are you expecting?

[jade wedge]

No more than 500mA. That's the rating of the diodes. Tracks are big enough. The 5V one in particular is way oversized.

[spiral pagoda]

Also, consider a DNP capacitor in parallel with R8 in case you need to improve the responsiveness of U3

0603 is fine

One more thing - J1 should tie to the ground of C3 and C4 on the top layer rather than forcing everything through that little via

[jade wedge]

Would bringing it up to the ground of J1 be sufficient?

[spiral pagoda]

I'm not sure what you mean

[jade wedge]

Oh, wait, that's what you said.

[spiral pagoda]

Did you put your name/date/version on the back silkscreen?

[jade wedge]

I've got my name on it, but adding more would be easy.

[spiral pagoda]

Nice work!

Now a design question rather than layout: Have you considered what you want to happen when the lipo is drained?

[jade wedge]

Thank you.
And thank you for taking the time to give it a look. I will definitely see about adding thermal reliefs for the battery pad, and the extra cap to go with R8. I've already added a new track to go from C3 ground to J1 ground

[spiral pagoda]

It looks like TPS61023 can operate down to 0.5V, but drawing your lipo below 2.5V (or even 3V) can really damage it

[jade wedge]

I'll be using batteries with protection built into them, and suggesting that when someone is picking a battery that they should do the same.

[spiral pagoda]

Ok. Cool that this could even operate off a rechargable AA

Not with the charging circuit, of course

but low-voltage boosts are kinda rare

[jade wedge]

I was going to make another design that was intended to work off AA batteries, but I hadn't thought of just not populating the charging circuit.

I suppose that does make it way easier.

[spiral pagoda]

Yeah. You could even make a bypass for D2

[jade wedge]

Slap a 0ohm resistor across it and call it a day.

[spiral pagoda]

On the diagonal, yeah

[jade wedge]

Awesome. Well, again, thank you for your time and input. This makes me much more confident moving forward. Will I actually be able to do all this SMD work without breaking anything? Time will tell.

[spiral pagoda]

Honestly, a hot plate works wonders for SMT soldering, especially since you have everything already on one side

Ideally a small one with good temperature control, but it's not that critical

[jade wedge]

Right now it'll be a hot air gun, and regular iron for touching things up. The previous revision used a USB micro port, and I managed to do that. U3 is about the same size, but more exposed.

[latent jungle]

I call that the “surface tension tap”

[fervent arrow]

This is a really specfic part I'm having issues with, but if someone could give me some advice or ideas I would love them forever.
I have a protoboard with a HT8950 voice modulator 18-pin IC chip. The LAMP pin constantly lights my led, which is only supposed to happen when it gets input from a mic, but it does it even when theres nothing actually connected to the audio in pin?

[spiral pagoda]

If the pin is floating it could be picking up radio signals or radiation from AC wiring. Try a pull-down to keep it steady

The example schematics all show a 33k resistor from AI to AO which is probably to provide bias but might be enough to reduce the erroneous input

[velvet steeple]

Does anyone have experience with water pumps, hose, and mister heads for terrariums/vivariums? I'm trying to figure out a hardware design to be controlled with a microcontroller and a humidity/temperature sensor. I already have a Qt Py ESP32-S3 with the sensor logging data via MQTT. I just need to know where to go from here to actually have it mist the environment when humidity dips below a specific threshold. Can I do it with a Qt Py or do I need to upgrade to a feather or rasppi?

[spiral pagoda]

The QT Py should do just fine

Your mister requires a certain pressure and flow to operate. Make sure you choose a pump that provides that, and think about where the pump will be getting power from

The absolute simplest method to control a DC pump is a low-side MOSFET, but you may actually prefer the flexibility of a SSR.

Looks like giant African millipedes prefer 70-80% humidity and 72-78F. Sounds like a humidor. Except cigars tend to mold above 75%

[velvet steeple]

I started with a fogger. It kept the air humid but not the substrate, which needs moisture for things to rot as they are detrivores. Too much moisture in the substrate = bad, not enough = bad.

[spiral pagoda]

A mister makes sense then. Will you have this connected to your house water or are you going to pump from a reservoir?

[velvet steeple]

A bucket or container of some sort holding distilled water

[spiral pagoda]

Many garden misters expect 30psi

[velvet steeple]

it will be lower than the tank too...

[spiral pagoda]

Is dribbling ok or do you actually need the mist?

I’m guessing mist

[velvet steeple]

We would need to minimize dribble as much as possible. None is ideal

[spiral pagoda]

Got it. Then high pressure is needed. Aquarium pump (impeller) will not suffice

Parastaltic?

Or maybe a vane pump?

A piston pump will cause bursts of water which might be noisy and frightening to the millipede

Could dampen with an air accumulator of course, though that will take some tinkering to get right since we don’t know the dynamics of the system

[velvet steeple]

I'm honestly googling most of the things you are saying 😄

[spiral pagoda]

But if those are too expensive, you could DIY it with a soap dispenser and a RC servo

How much water do you want to dispense at a time?

[velvet steeple]

Just enough to nicely dampen the top layer of substrate. I'll probably need to have 2 heads, one for each side. 29 gallon tank. 30.19" W x 12.4" D x 20.86" H. I have about 6-8 inches of substrate

[spiral pagoda]

I don’t know the calculation for that, sorry

Are you talking about a cup? A tablespoon?

Right now you’re spritzing by hand with a spray bottle, yeah?

How many spritzes does it take? Do the same amount into a measuring cup and see how much it is

[velvet steeple]

I'm still dialing that in with my daughters spray bottle for her hair 😄

I think I found a path forward with a MOSFET and a 12v diaphragm pump that does 80 PSI

[spiral pagoda]

Nice! A diaphragm pump will still produce some pulsing, but hopefully less than a piston pump

Millipedes are sensitive to vibrations, aren't they?

And don't forget the flyback diode

[velvet steeple]

Yeah but we are talking twice a day.

[spiral pagoda]

cool

[velvet steeple]

The pump is 12v 3 amp. I got a IRLZ44N MOSFET (47A 55v), IN5404 diode (3A 400V), a 12V 5 amp power supply, and a 12v to 5v 3 amp buck converter to power the qt py. Does this sound solid?

[spiral pagoda]

Great choice of MOSFET but you'll want a schottky diode

1N5822?

Honestly you could use a 1N5817 and it would be fine

[velvet steeple]

what is the concern here?

[spiral pagoda]

The concern is switching speed. The MOSFET has a maxium voltage above which it will arc internally and fail. When you shut off the pump, it temporarily acts as an inductor and wants to keep the current flowing

The diode provides a path for this, which is why you ned a diode that switches very quickly

Otherwise the high current with nowhere to go will translate into high voltage and hurt your mosfet

[velvet steeple]

and you're saying the 5822 switches faster than the 5817?

[spiral pagoda]

I'm saying they both switch faster than the IN5404

[velvet steeple]

I appreciate your help thanks. We'll see how this goes

[spiral pagoda]

Sounds good

Did you pick a mister nozzle?

[velvet steeple]

https://www.amazon.com/gp/product/B0DZTB68F5/ref=ox_sc_act_title_6?smid=A3A27ZYQMHSCY9&th=1

I'll probably just need to chain 2 of these together and I have holes that they can slide into

I'm already thinking about ways to adapt this to our rain barrels and using it to water our garden

[spiral pagoda]

Nice choice

[worn iris]

any UPS recommendations with li-ion bat connector for a PI? perferrable one without the big 18650 battery holders

designing my cm4 carrier board rn and don't trust myself quite enough to do managment stuff with li-ion, would have just nabbed a geekworm one since those ar rated 5v 5A but they don't seem to offer schematics or anything aside from a dxf image where nothing is labelled except the silkscreen text for the website and the product name/rev

[pastel night]

i made my own but did you see this: https://www.adafruit.com/product/5705

p.s. you don't need 5Amps... unless you doing a pi5?

there are others similar.

[worn iris]

unfortunately there is none in stock, like at all

so not in stock on those 2 sites but across the pond theres a few sites that do have them in stock i just gotta wait a bit longer

[pastel night]

any of the pass thru power banks are also an option... https://www.adafruit.com/product/4288 but... 1) keep the output cable short, 2) if rpi 4 or 5 use one with a 3 amp output rating.

it's a "dumb" solution but is "off the shelf".

[smoky terrace]

question for folks who've worked with the Black Acrylic diffusion panels, how well do they laser cut? (https://www.adafruit.com/product/4749 for reference)

I've got some isolinear optical chips I made out of neon acrylic and I want to use the black difusion acrylic to make something that looks Star Trekish to slot them into

[jovial yew]

I'm looking into making an EM radiation sensor with an ESP32 for non-ionizing radiation like Wifi, Radio, power lines, and to a certain extent people. I want it to be as sensitive as possible, so pretty much always detecting, I don't really care about accuracy or precision so fluctuation is fine. Could I use a TL072? I've seen lots of projects online, but a lot of them use higher power like 5V or 9V, or don't use a microcontroller.

[spiral pagoda]

I would recommend a log amp, though they run about $20 each

[spiral pagoda]

https://www.digikey.com/en/articles/log-amps-meet-rf-power-measurement-signal-scaling-needs

[jovial yew]

Interesting, pretty pricey, but I'll look into it, thanks for the suggestion.

[vestal iris]

I had the black acrylic laser cut from a sign shop here and it worked great.

[smoky terrace]

Awesome, I'm thinking of using it for an LCARS display for isolinear chips

[vapid lark]

is there anyone in here who can help me check over my entire sheet? i THINK i am done! but i've been working on it for hours and days and my eyes are tired! PM me and i'll send a screenshot of it (i am using easyEDA software btw)

[latent jungle]

don’t ask to ask, and most people prefer a public review to avoid wasting time suggesting things

[spiral pagoda]

We can make a thread if you’re worried about cluttering the channel

[muted skiff]

Would anyone be able to help me diagnose why my USB Power Override section (right-side) does not work as intended? I've simulated the section using Falstad and the logic seems to check out. I essentially want to cut off VBUS whenever I have a barrel connector plugged in (12V to 5V buck on top side).

I've confirmed both power inputs work independently from one another on my PCB but when I add the DMP2045U, VBUS does not make it to VDC no matter what the power state of the buck is.

[spiral pagoda]

At minimum, Vbus should conduct through the body diode of Q1. Check the forward voltage with a multimeter to make sure it’s installed correctly.
Also check continuity from the USB port to Vbus. Did you make sure to solder closed JP13 and JP18?

[muted skiff]

Alright I've tested the continuity and it seems like the footprint was the issue. The standard SOT23 on KiCad has a different pin numbering than the DMP2045U datasheet expects. Woops. Thanks!

[latent jungle]

Not sure where you got the symbol, but for discrete parts like FETs, I prefer to use the KiCad's default parts that list the pin order in the symbol name:

It makes matching the symbol pinout to the footprint much easier.

[muted skiff]

Neat, thanks! I just used the PMOS symbol from the Simulation_Spice library. This seems a lot clearer.

[latent jungle]

I wouldn't use the Spice symbols for anything but simulation.

In this case, you wanted Q_PMOS_GSD

[muted skiff]

Gotcha yeah that's the one haha

[spiral pagoda]

You might consider accounting for the voltage drop of D11 in your buck regulator output: put the feedback resistors after D11

[quasi flare]

anyone know how many amps the "5V" line on a feather can supply? (up to the max for the USB connection, obviously)

[spiral pagoda]

500mA, limited by the MBR0540 diode

[quasi flare]

dang. need 1A. guess i'm gonna have to split the usb cable before it gets there.

[spiral pagoda]

Oh wait. The 5V line doesn’t go through the diode, does it

So I guess you’re ok

[quasi flare]

i don't think so, haven't looked at the schematic recently.

was more like can the traces handle 1A. 🙂

[spiral pagoda]

Yeah, I think they’ll be fine

[quasi flare]

building a thing to turn a wyze cam on and off depending on data from google calendar. so need 5V1A to go through a SSR to power the cam.

[spiral pagoda]

Expect a significant voltage drop though from the USB cable. You’ll probably only get 4.6V at 1A

[quasi flare]

considering that the cam will run with a 20' cable, i think it's OK.

🙂

[spiral pagoda]

Is this 1A continuous or intermittent?

[quasi flare]

not sure. specs just say 5V1A.

might be worth splitting the power pre-feather so it doesn't overheat.

[spiral pagoda]

Apparently a 10 mil 1oz trace will only heat up by 11C with 1A continuous

But I didn’t actually measure the trace

[quasi flare]

i have the eagle files, i'll look.

looks like 15mil (.4mm)

[fierce elk]

Is it necessary to use a P-FET circuit to safely disconnect a lipo battery from a microcontroller (like this https://www.adafruit.com/product/1400), or can you just hook up a switch in series with the battery terminals?

I've done the latter in the past and it was fine but I'm not too sure how dangerous that is

[unique patio]

a switch is fine, and is straightforward

[fierce elk]

Gotcha, thanks!

[pastel night]

most batt potection circuits are N-Fet (low side).... P-Fet is high side. There are reason either. A contact type switch is also an option on either high or low side. however micro controllers rarely have fingers (that can push switches).

[marsh bear]

eInk Feather friend users - does the FPC connector "open", or is it simply an insertion connection. I have an extension cable coming tomorrow and don't want to mess things up!

Thanks!

[wintry wind]

My two had two fingers that you slide out (not up) trying to get them equal and then slide the cable in and push the two fingers back in to tighten down the cable.

[tulip swift]

You need to gently slide out the retainer clip before insertion (the grey parts) then reverse when you get the cable seated)...

[wintry wind]

Much better -- picture and arrows worth a thousand words!

[tulip swift]

I should add that the dark grey part will only slide out a couple/few mm so basically just pops outward a bit (don't try to pull it further than where it stops normally), slide in your cable, push the dark grey clip back to the left to secure.

[frosty parcel]

hey! could anyone help me troubleshoot a switching mode power supply issue?? I am using the linked 1Mhz IC to step 12V down to 5V and built the typical application schematic as shown in the data sheet but for some reason it's getting quite hot and the output looks like this.

https://www.mouser.com/ProductDetail/Diodes-Incorporated/AP63205WU-7?qs=u16ybLDytRZtkj8PzdWCOw%3D%3D

<@&741361254035095602> <@&617066238840930324>

[distant raven]

I’d have to see a schematic and probably a picture of the board to make any guesses

[latent jungle]

share your PCB design

[frosty parcel]

well, it's on a breadboard

and it's.... not exactly clean, it's a huge switching loop but there's no current being pulled (it's just 12v in and going directly into the scope)

[worldly schooner]

Are your 2x22uF caps in series or parallel? Breadboard is very hard to follow when components are placed tightly.

[frosty parcel]

parallel

[worldly schooner]

If you put a dummy load across the output, does the scope output change?

[frosty parcel]

yea sorry abt that

[latent jungle]

Your breadboard is the problem.

SMPS and breadboards don’t mix.

There’s a reason the data sheets include a detailed PCB layout with ground planes and incredible short traces.

[frosty parcel]

so u think i should just order a pcb w/o prototyping it?

[worldly schooner]

If you want to prototype it, you’ll have to get a bit artistic with soldered connections.

And given the thermal callouts, active cooling is encouraged.

[latent jungle]

You can’t (properly) prototype a SMPS without a PCB.

Especially if it is switching at 1 MHz

[worldly schooner]

I’d advise using a PCB to prototype it, or remaining extremely faithful to the app notes and send it straight into your design.

I’m tempted to try 3D assembly of an SMPS power supply for fun, but yeah there’s more that could go wrong than right

[frosty parcel]

yea... especially since you saw what happpened when i tried to hand solder this tiny ic to a pcb

[worldly schooner]

@latent jungle any SMPS suggestions for a fun challenge? I’m thinking through-hole components only and tying the output to a DC fan to cool itself as a nerdy little art piece.

[latent jungle]

Sorry, no. I don’t find solving SMPS design challenges as fun.

[frosty parcel]

ok ig ill just order a PCB and pray i did it correctly

if i did do it corectly should it really require active cooling tho?? or just proper pcb design

[latent jungle]

cooling requirements depend on your application

[spiral pagoda]

If you’re doing through-hole you’ll want to stay 100kHz or lower. But rather than using the new current-mode synchronous buck regulators, you might have more fun with an 8-bit microcontroller operating a boost converter (low-side switch) to drive a string of LEDs or some EL wire. You can even control with a 555 timer if you’re ok with open loop.

I’m actually considering a PIC16F15213 for such a design, because the TPS923650 I had selected was too small to reliably solder

And the PIC is cheaper than a ATTINY85

[fierce elk]

Anyone know why this would be the output of the ov5640 camera module? Using the test code given from adafruit. It's definitely getting some data from the camera sensor because it darkens when I put my fingers over the lens. MC is the rp2350, using the data pins coming from the HSTX connector (not sure if that's a problem). Only other nonstandard thing I did is I'm using the internal 24MHZ clock generator, so the XC pin is left unconnected

(ignore the small image, running at 96x96 for the sake of framerate)

[fierce elk]

Did some more investigating on this - my initial thought was that the 24MHz internal clock on the camera module was too noisy or not good enough. Checking it with a scope, the clock frequency is fine, but should the waveform be more of a square wave? The pixel clock looks very similar (more sine than square wave)

[latent jungle]

where are you measuring and how?

[fierce elk]

Measuring directly at the output of the clock generator (which goes directly to the XCLK pin on the camera module) via a small jumper wire soldered on. Coupled the scope probe to the same ground as the camera module

The other signals like the href and vsync are very solid square waves so I don't think it's an issue with my measurements

[latent jungle]

If you’re measuring with a 10:1 probe switched into the 1 position, that would explain the wave shape.

[fierce elk]

Ohhh I didn't even consider that, hold on

Setting it to 10X this is the new waveform. Still not great

Can definitely see the peaks and valleys of what should be a square wave, so it's definitely better

[knotty tiger]

can you show a photo of this test setup, including the soldered jumper wire?

[fierce elk]

Part of me is wondering if the messed up video is due to the fact that I'm using the HSTX breakout for the camera data, which would be very unfortunate if that's the case since I need consecutive data pins

[knotty tiger]

yeah, that’s marginal for accurately measuring 24MHz waveforms. you need better grounding and shielding

[latent jungle]

1X has less than 10 MHz of bandwidth.

That waveform, in 10X, can be from how you’re connected to ground.

those long wires explain the ringing.

[fierce elk]

As long as the clock's shape on the scope is more due to the way I'm measuring than the clock itself I'm okay to rule it out for now. Do you think that the wire lengths from the HSTX connector on the feather to the camera module could be causing the messed up image?

[knotty tiger]

taking off the hook clip and using the needle point of the probe will help. most probes have a ground ring right behind the needle that you can wrap a small ground probe wire around

[latent jungle]

breadboards and long jumper wires are not appropriate for signals running in the 10s of Megahertz.

[fierce elk]

Yeah this is very much temporary, just trying to get a handle on how it works for now before I do anything permanent. If it's a showstopper to use a breadboard at all then I'll have to bite the bullet and install it on some perf board, but I'll have to deal with that FPC connector no matter what since that's the only way I can interface with those GPIO on this board

[knotty tiger]

is the FPC on the breakout the same pinout as on the camera module?

[fierce elk]

No unfortunately, FPC goes from the RP2350 HSTX connector to an HSTX breakout, which then connects to the camera module breakout

The camera requires 8 sequential gpio pins which this board doesn't have (unused) sadly, unless you pull from the HSTX connector

[knotty tiger]

can you highlight the routing of your clock wire? you might be able to get away with it if you run it in a twisted pair with a ground wire the whole way

[fierce elk]

Which clock, the pixel clock or the i2c clock?

[knotty tiger]

pixel clock. that’s the fast one you have to worry about

[fierce elk]

It stretches all the way across the board, definitely not ideal... I can try to twist it with a ground wire later today and see what happens. That one doesn't go through the FPC, it goes directly to the microcontroller via a jumper

[spiral pagoda]

You made an inductor out of the loop of your probe wire and ground lead. Try the spring clip next time

[frosty parcel]

any chance you could give me some feedback on this pcb design? it's one of my first ones and it's for a 1Mhz buck, esp32, and motor driver

<@&741361254035095602>

[silk lark]

why?

[spiral pagoda]

There’s a lot of issues. Start with the layout around the buck regulator. AP63205 is a great chip but you need to keep your inductor loop tight. There’s a recommended layout near the end of the datasheet - try to emulate that

Increase your trace widths for durability and current handling. I usually do 10mil traces unless I need to fit in tight spaces, but you might need more for the regulator. KiCAD has a nice trace width calculator that can help you decide.

Please pick a smaller inductor, too. That through-hole part is enormous and will be the heaviest and tallest thing on your board. Actually the main benefit of using a 1MHz buck regulator is that you can use a physically smaller inductor, saving weight and cost.

What really puzzles me is all those resistors between the ESP32 and motor driver. Are you expecting them to serve as a kind of level shifter?

[frosty parcel]

i kinda already ordered it and have it so that’s the only one i can use

[frosty parcel]

no, they’re just from the application schematic

i can check what they do rq

ill take a look at the recomended layout and trace width tho, that's a good point

do u think i need to increase all the traces or just the ones that'll be driving the motor and stuff

[spiral pagoda]

Increase all the traces to 10mil and then adjust as needed

[frosty parcel]

is there a good way to change them all at once in kicad or is that more of a thing i do to each track individually

[spiral pagoda]

I think there’s a way? But you’ll be rerouting a lot of this stuff, so better to set your default netclass

If you programmatically changed all the trace widths without moving them, it would result in a lot of shorts and DRC errors

[frosty parcel]

this seem like a more reasonable current loop?

[spiral pagoda]

Pin 5 is a sense pin. The current loop is actually through C3 and C4. So you need the ground of those capacitors to be near the ground of U1, but the 5V on those capacitors to be near the 5V of L1

Here's a great tutorial on this https://www.powerelectronicsnews.com/power-supply-design-tutorial/

[frosty parcel]

aah makes senes

also for some reason it seems like some of the traces on the blue side are cutting the ground plane on the red side and im not sure why

any ideas

[frosty parcel]

wait arent C3 and C4 just the Vin and GND pins?

`oh nvm ic what u mean

wait but C3 and C4 are just the output bulk caps for smoothing right

[spiral pagoda]

I guess we need to talk about what kind of capacitors they are. Did you pick MLCC? Are they still capacitive at 1MHz?

[frosty parcel]

im using these https://www.digikey.com/en/products/detail/kemet/C1206C226M9PACTU/2215548

[spiral pagoda]

Datasheet have the impedance curve, but it's MLCC so probably ok for 1MHz.

[frosty parcel]

ok cool

[marsh bear]

thank you!

[frosty parcel]

so basically, should i swap C1 and C2's locations with C3 and C4 and then bring them closer to the gnd

[spiral pagoda]

The hbridge also has important layout and heat sinking requirements. See pg 23 of that datasheet

I think you will need more space to work with. An easy way to get this is to flip your through-hole components to the bottom side

Anyway. Read that tutorial. It's very helpful

[frosty parcel]

sounds good

i’m also gonna change the gnd pins to solid for thermals

[spiral pagoda]

@silk lark Have you ever seen series resistors on the inputs for an hbridge? They’re in the datasheet but not clearly explained. Inputs are 3.3V compatible but 20kHz max

[frosty parcel]

i mean i’d breadboarded w them and it worked fine

[silk lark]

I've seen that on inputs for hobby servos. Presumably to protect the pins?

inside the chip it's mosfets, so maybe to limit the base current?

[spiral pagoda]

Mosfets don't have base current. The gate is a capacitor

I thought maybe it was to reduce slew rate? But at 20kHz I don't see why it would matter

[frosty parcel]

i mean it said to "limit the current sunk into the IO pins"

idk

[silk lark]

perhaps a precaution for very delicate old microcontrollers?

[frosty parcel]

also im a lil confused here for the buck

like there isn't any resistor in the application schematic so like... what are R1 and R2?

i just have mine set up like this

[frosty parcel]

should i just ignore the resistors

also not clear on what C4 is

[knotty tiger]

where's the datasheet for the switcher?

[spiral pagoda]

Those are to move the poles on the frequency response, improving step load response and stability. Basically the capacitor advances the feedback so the regulator responds to droop before it gets very far

[knotty tiger]

they're probably a filter for the feedback node, but you need to see what the datasheet says about them

[spiral pagoda]

https://www.diodes.com/assets/Datasheets/AP63200-AP63201-AP63203-AP63205.pdf

[knotty tiger]

which particular chip are you using? the fixed-output ones seem to not need the filter. the variable-output ones need a resistive divider, with the capacitor to adjust the frequency response

[spiral pagoda]

They're using the 5V one. You're right that the chip works fine without the filter. I usually add it anyway and haven't bothered to compare.

[frosty parcel]

ok sounds good so ill just ignore those

[spiral pagoda]

I mean... how much do you want to learn about buck regulators? You could always leave the footprints in and not populate

But if you leave the footprints out, it's sure hard to fit an 0805 onto that sot23-6

[frosty parcel]

i mean id like to learn about htem

i thought i didnt need that filter since im just using it at 5v tho

i might be misunderstanding what you're saying

id need to populate tho or else FB wouldnt be connected to VOUT

unless i also bypassed the footprints

[spiral pagoda]

You probably don't need the filter, but we don't know for sure.

[frosty parcel]

i mean, when i breadboarded it it seemed to work fine w/o the filter

[knotty tiger]

Figure 20 shows the application circuit for the variable-output versions, which does have the filter and the resistive divider. you probably don't want a divider if you're using the fixed output voltage

[spiral pagoda]

Yeah, you're right

Just leave them out then. Sorry for the confusion

[frosty parcel]

no worries

do u think i could get away with just beefing up the traces that'll actually be carrying substantial amounts of power and leaving the IO ones as is?

[spiral pagoda]

Are you particularly against redoing your whole layout?

There's a lot to be improved

[frosty parcel]

i mean not really

lemme make a quick list of the things to change:

• increase general trace width to 10mil
• substantially increase power carrying
• use buck and motor driver layouts from data sheets and make a power loop on the buck

is that about it?

or are there other big things

this is where im currently at

btw red is a GND plane and blue is a 12v plane

[spiral pagoda]

No, there's a lot more

[frosty parcel]

oh ok

[spiral pagoda]

All those vias and routing bends around R1-R6 are curious. Why is the layout like that?

[frosty parcel]

i was trying to do it to keep the ground plane solid

[spiral pagoda]

Or D1 and R7. Why go to the bottom layer rather than repositioning your components?

While I'm routing traces, I often find that I can make a trivial change to the layout (or even the schematic) that makes routing cleaner

In this case, I wonder if you could remove R1-R6 entirely?

[frosty parcel]

do you think i should instead just use two ground planes so i dont have to worry about breaking it when routing on top

and then just carry the +12v in traces

[spiral pagoda]

What do you mean "two ground planes"?

[frosty parcel]

ground plane on top ground plane on bototm

ig "filled zone"

idk what the proper terminology is

[spiral pagoda]

But you also have traces on those layers

You could stitch them together. Better if you can minimize the traces on the bottom

You'll need it not just for current carrying but also for heat

[frosty parcel]

oh ok

[spiral pagoda]

How much current can your motor take?

[frosty parcel]

i think it's 5.5A stall

but it shouldnt get anywhere near that since the gearbox cant handle it

i can send link

[spiral pagoda]

The gearbox breaks if you stall your motor??

[frosty parcel]

https://www.pololu.com/product/3207/resources

note 2 basically explains it

so it's recomending operation at about 1.25A

even though stall is 5A

[spiral pagoda]

Yeah, DC motors don't like to stall. But they always start from stall

[frosty parcel]

oh so on startup i would actually be pulling 5A?

[spiral pagoda]

Ok. So your hbridge traces need to handle at least 1.25A, but ideally 5A

Yeah, for a fraction of a second. What's it turning?

[frosty parcel]

it's turning a wheel that's pulling the cord on some window blinds lol

[spiral pagoda]

Ok. Did try it and see how much current it actually needs?

Most benchtop power supplies will tell you, or an ammeter

[frosty parcel]

havent tested yet but i checked the force required and it should be about 25% of the stall current

maybe a lil less once it starts moving

[spiral pagoda]

Ok

Torque is proportional to current for the brushed DC motors. You'll have a little extra friction from the gears, but your math checks out

How much power will the hbridge be dissipating when it's delivering 1.5A?

[frosty parcel]

lemme check

[spiral pagoda]

Also... what happens when you shut off the motor?

Will the blinds go back down?

[frosty parcel]

no it has a brake

also it's one of these things so itll stay where it is

[spiral pagoda]

Oh cool. I was worried it was venetian blinds where you have to pull diagonal to let it back down

[frosty parcel]

i think it'll be about 1/3 of a watt

listed resistance for low and high side combined at temp is 145 mΩ

probably less tho

since at 25C it's only 72mΩ

[frosty parcel]

wait so in the redesign are there any big things i should change other than these+making the bottom layer ground w/ minimal routing+routing R1-R6 more cleanly?

[spiral pagoda]

I thought you would want some heat sinking for the hbridge. If it really ends up being 1/3W it won't be a big deal, but if you want to use this board for other projects too you might not want that constraint

By heat sinking I mean vias to the bottom ground plane to suck heat away

[frosty parcel]

aah

ok that makes sense

[spiral pagoda]

One more thing that might be nice is a series diode on the 12V input to prevent accidental reverse-voltage

Not all barrel jacks have the same polarity

[frosty parcel]

aah ok good to know

would that cause a voltage drop tho?

[spiral pagoda]

Yeah. Schottky diode has low voltage drop though

[frosty parcel]

would this be an ok choice?

https://www.digikey.com/en/products/detail/smc-diode-solutions/SB160/6022605

[spiral pagoda]

Here, I sorted by voltage drop so you can pick one https://www.digikey.com/en/products/filter/diodes/rectifiers/single-diodes/280?s=N4IgjCBcoKwOwAYqgMZQGYEMA2BnApgDQgD2UA2iACwIIDMcAHCALrEAOALlCAMqcAnAJYA7AOYgAvsQCcYKshBpIWPEVIUQAJgRgYMhcR1gAbDpDEwcMI0aHwcLYxl0LDujJP2rVTxEtwJghw3nCOJjBuVoxaEVFwMggwkZaMtFpwbloZYDJaWUl0VMxGCEHJBSZp9johvgUxdCXaCHmIWWC6YP7anVX5Rt10VVlUYWVZdGAMmUZTiSnadFpTJpNOJrNLRZtZJmYmEGwgXDz8wuJSljLOisqqBMRkkJRILNIgALQyd1CCAK7qZ6UI4fT75aBKP4CQFPTSRd4fNaQoQAEx42UOblOkBAAGESP92Nh8KiAKoiIScADy6AAsvhMLh-gJ8G4AI6cACePE6bm57DZuKZaEkkiAA

[frosty parcel]

oh i was just asking since i already have the one i linked

and you just mean something like that right

[frosty parcel]

or am i misunderstanding

[fierce elk]

Ended up trying this and it didn't make a difference. I'm now convinced my problem is all my data pins are passing through three sets of breadboard links and an FPC breakout board before they reach my microcontroller which is scrambling them like crazy. Rewiring it all to perf board (which was my plan for the final implementation) might help but I kind of doubt it at this point, I may just have to rethink my implementation of this with a microcontroller with a more accessible set of 8 sequential GPIO pins

Didn't realize just how susceptible those camera data pins were to noise

[knotty tiger]

these high data rate interfaces are really unforgiving. looks like this particular module doesn't use the pixel clock pin. it does have some differential pairs, so keep them next to each other when you wire it up on a perfboard

[fierce elk]

You're referring to the OV5640 module? It does use the pixel clock pin, it (optionally) doesn't use the XCLK pin if you cut one of the solder jumpers in the back

[knotty tiger]

pin 17 on is listed as NC on the datasheet

[fierce elk]

Oh wow you're right, that makes everything WAY more confusing

Maybe they put it on the breakout to give it backwards compatibility with other camera modules? Very strange that it's there and is required to be assigned within the ov5640 circuitpython library

[knotty tiger]

maybe other modules use it?

but my guess for the observed failure modes is splitting some of the MIPI differential pairs (pixel clock (which is on one of the differential pairs) can go up to 192MHz in some modes)

[fierce elk]

Could be, I think this breakout is only sold with the OV5640 by adafruit but maybe they had some intention of using it for other camera modules

Data pin D2 is also connected to a NC oddly enough

But either way I think my plan going forward is to try it on a perf board with extremely short wire lengths, if that doesn't work then it's most likely the added overhead from the HSTX FPC connector. At that point I gotta just bite it and either design a board myself or buy a new microcontroller

Very insightful day though, learned a lot from this. Thanks for the help everyone!

[spiral pagoda]

The one you have is rated for 1A only. Nevermind the diode I guess

Sadly you can’t parallel diodes because they don’t share current

https://tenor.com/view/gandalf-share-power-does-not-share-not-share-power-gif-16080343783123063712

[frosty parcel]

ok so i should get something beefier

[frosty parcel]

could i also use a p channel fet or is that a bad idea

[spiral pagoda]

That’s a fine idea

[frosty parcel]

you think the p channel in this pair would be ok? https://www.digikey.com/en/products/detail/alpha-omega-semiconductor-inc/AO4629/3152394

just asking since it's something i already have

[spiral pagoda]

Cool chip! But no, leave it out. It’s not made to dissipate 1.5W continuous. Forget I said anything about the diode

[frosty parcel]

ok cool

is this looking better

i was trying to figure out how to ground the mounting holes but other than that

[spiral pagoda]

YES. Much better

[frosty parcel]

yay!

is there a way i can make the moutning holes grounded? /would that be wroth it?

[spiral pagoda]

Kicad has a different symbol for grounded mounting holes

They have to be wired up in the schematic

[frosty parcel]

oh ok

[frosty parcel]

is it just this?

wait not it's this right?

[spiral pagoda]

You got it

And now you can wire it to ground

Btw, you can “exchange symbols” rather than redrawing

That way the footprint and placement won’t change

[frosty parcel]

aah ok cool, so are there any major things i should change on here or should i just leave it at this

[frosty parcel]

ok i also shrunk it a lil

[frosty parcel]

also do u think this button footprint would work for this? https://www.adafruit.com/product/1009

[spiral pagoda]

I’m not going to answer that one. You have to read the datasheet and confirm it yourself, or create the footprint if you aren’t sure

[frosty parcel]

fair enough. you think the layout looks ok tho?

[spiral pagoda]

I don’t like the 5V going under the regulator. Can you route that around on the top layer and add some vias to bring the regulator heat to the bottom layer?

Likewise, you have 12V hopping to the bottom layer just before a capacitor. What is this about?

If you swap the position of C5 and C6, you can stay on the top layer

[frosty parcel]

the 12v is a good catch

as for the 5v, that’s just what the data sheet said to do

[frosty parcel]

should i connect the vias to anything or do anything special to them or generally just place them

also should i be using stitching vias to connect the top and bottom ground pour

[spiral pagoda]

Connect the vias to the ground plane. Yes, stitching vias are a good idea. There’s a plugin for that

[frosty parcel]

ok i feel like i did this wrong. is this what u meant w stitching vias

also for some reason it generated a bunch of vias in keep out areas, is that normal?

[spiral pagoda]

Stitching vias look fine

The only keep-out I see is C7

[frosty parcel]

yea i deleted them manually before i sent

is that normally how it works?

[spiral pagoda]

Depends how good the script is. I don’t remember

Please add a via on both sides of your bottom-side 5V trace