help-with-hw-design

[spiral pagoda]

That should help with the heat from U1

And then I think you’re ready to clean up the silkscreen.

[frosty parcel]

ok, would that be more of like a fencing via type thing? i dont see a way to do that in the tool

or hsould i just do it manually

[spiral pagoda]

Manually, yeah

Feel free to delete or move a couple vias to make the silkscreen better. It looks like for this design most of the refdes would go on the left or right rather than above

For the mounting holes, did you make sure they fit the screws you want to use?

[frosty parcel]

do you just mean like this?for the stitching vias?? there wasn't much space under the chip?

[frosty parcel]

yea, they're just m2

[spiral pagoda]

Huh. Somehow those didn’t connect to the ground plane

[spiral pagoda]

Make sure they’re clearance for M2, not actually 2mm

[frosty parcel]

looks like i had to set them manually to gnd

does this look ok

[frosty parcel]

yea they're m2 holes so 2.2mm

[spiral pagoda]

Cool. I’ve made the mistake recently

[frosty parcel]

and is the point of those vias around the 5v line for emi stuff or something else? and do they look like enough or should i add more somewhere?

[spiral pagoda]

They’re for heat sinking. I can’t say about EMI, I’ve never done the testing

[frosty parcel]

aah ok makes sense

i just wansnt sure if i could find a good place to put more of them

[spiral pagoda]

Don’t forget to put your name, project name, version, and date on the back silkscreen

Good job!

[frosty parcel]

<3

[frosty parcel]

thank so much for the help!! ive def learned a ton and I really appreciate it!

[frosty parcel]

also when exporting gerber files, should i just leave all the settings and defaults as is?

[spiral pagoda]

Follow the requirements of your PCB house. Some have specific naming and format requirements. They’ll also tell you useful things like minimum trace width and minimum spacing, which you can use to set up your board constraints for DRC

[north finch]

Please forgive me if this is inappropriate for this forum. I am a n00b. I am hoping to get a little help from the community.
I think that I am running into an engineering issue and I am a little bit flummoxed… (yes, flummoxed). Attached is a PDF of my circuit. Call this a board review.
Very simply put, the circuit supports a fan management system. There are two sensors that monitor temperature and humidity in two separate spaces. When there is a great enough difference between the two spaces, the fans are turned on using the relays and then run until the environments are equalized.
I have gotten a fab and assembled board that in the surface, appears and ohms out. Ground is ground where it should be and not where it should not be. The 3 voltage nets ohm out as well. The 12V net attaches where it should, 5v attaches where it should, 3.3v goes to the 3.3V input for the ESP32. However, being able to trace these nets does not guarantee anything but that, nets seem ok.
To test the board, I have hooked the board to a power supply to control the source voltage input. I have the power supply set for 12v @ .5A. When the power is applied to the board, the +12V from the power supply goes to about +6V and there is not any voltage that would seem correct. 5V is low at the output of the LM7805 and no 3.3V on the output of the LM317.
Visual inspection does not uncover any manufacturing errors. The LM7805 does get rather warm telling me that there is power loss. However, it is not too hot to touch and I do know that the LM7805 runs hot.
I am asking the community to take a look. I do not claim to be any kind of an electronics expert but I have played on on TV earlier in my career. Please provide any feedback you believe will help.

[knotty tiger]

have you prototyped this circuit on perfboard or something before getting it fabricated? dropping 12V all the way down to 3.3V with a linear regulator isn’t a great idea. it will create a lot of heat except at very low power levels. can you power supply indicate whether it has gone into current limiting? it sounds like it might be, from your voltage readings

[knotty tiger]

what is the intended purpose of D1 and D6? if they’re meant to be power indicators, they’re in weird places in the circuit, and D1 lacks a current limiting resistor. D1 is also in series with the negative leg of the capacitor, so it will only light up briefly when the capacitor initially charges, and probably fail due to the surge current (open vs shorted is uncertain)

[north finch]

Thank you for your fast response!!
Yes, I have 5 hand soldered boards that seem to work fine. Voltages are stable and drive the rest of the circuit ok. However, I am using an ESP32-S3-WROOM-1 Dev Kit and not a module. Though, I am reading that this should not make a difference.
Instead of the LM317, I have looked at other ways of dropping the voltage but none were as "clean" (fewer components - neat and tidy space) as this. If you have another method to get a stable 3.3V to power the ESP32 module, please share.
Yes, D6 and D1 are to show that 5V and 3.3V are working. D6 does come on when the circuit is powered but there isn't 5V. If you have a suggestion for this, again please share.
I do really appreciate the feedback. Anything else?

[knotty tiger]

i would have done an initial power up at a lower voltage like 7V or so, depending on the expected voltage drop of the particular 7805. if it’s your first time getting a board assembled, maybe double check the polarity of all your polarized devices, especially in the relay section?

[spiral pagoda]

Yeah, linear regulators will create a lot of heat, especially for ESP32 which draws 150mA with peaks of 500mA, plus whatever the microSD card uses.

Moving on, how did you calculate your resistor values? D5 and D9 have 10k resistors, which will only allow 0.2mA through. You can check the datasheet, but 5mA is my rule of thumb for LEDs, and many allow 10 or 20mA. D2 and D3 are blue LEDs (~3V forward voltage) and have a 1k resistor AND a PC817 in series. Are you expecting anything to turn on? PC817 requires 5mA to operate and has a 1.2V forward voltage.

[knotty tiger]

i would put the power-good LEDs on the output sides of the regulators, and use larger current limiting resistors (unless you really need them to be bright). for dropping 12V to 5V, a buck regulator is probably more efficient. you can maybe use a LDO linear regulator from 5V to 3.3V

[spiral pagoda]

Moreover, the output of the PC817 are pulling the base down to ground on a 2n2222. That's the wrong polarity for an NPN

Glad to see there's a flyback diode on the relay. It's a 1N4007 which is notorious for slow switching, but for some reason everybody uses it. Give the 1N5817 some love.

On the USB port, you'll need both CC1 and CC2 to have the 5.1k pull-down to ground. Be careful with your USB routing - it's not possible to get the required impedance on a standard 2-layer board, so you need the leads VERY short. Should also have 22-ohm resistors in series. In fact, you may want to have the serial pins accessible as a backup - they're much more forgiving

Something weird about this design is that USB is tied to 5V, but in no way does 5V power the microcontroller. So you need USB AND external power in order to program the ESP32

[spiral pagoda]

I'm sorry about your broken PCB. Aside from changing all the resistor values, you're going to need to rethink your output circuitry for those relays ever to work.

[knotty tiger]

your hand soldered boards are using the dev kit? and your fabbed board is using just the module? seems like many variables unaccounted for

[spiral pagoda]

Don't be afraid of using the devkit on your finished design. I wish I had done so: only one of three boards I made with ESP32-S3 had functional USB, and even then it was borderline.

I was also trying to measure temperature inside a motor, but apparently thermocouples act as loop antennas and pick up the magnetic fields

[knotty tiger]

could you clarify whether you built your hand-soldered boards from this schematic? or did you draw this schematic from your hand-soldered boards? (or maybe you hand-drew the schematics you used for your hand-soldered boards?)

[spiral vapor]

On the CM4, pins 80 and 82, (SDA0 and SCL0) are a seperate I²C bus to GPIO 2 and 3, right? I have mapped the standard pins, to a header the same as on a Pi 4, and have a display connected to the first 26 of those pins. Can i still use a GPIO expander on SDA0/SCL0?

[fair dome]

I am going through it trying to make a HUB75 display arcade machine. I have a raspberry Pi 5 (3.3V logic) connected to a Matrix Portal (3.3V logic), connected to my Hub75 boards. I am coming across 2 problems. The first is that when I run test code just through the matrix portal (not involving the Raspberry Pi) the data which appears on the screen is incorrect, the wrong color, stagnant etc. The other issue I am coming across is when using ssh to communicate with the Raspberry pi, it is saying that there is nothing connected to the rx/tx port event though the baud rate is the same for the matrix and the pi and they are connected through a common ground. I feel like replacing the matrix portal but I'm convinced it could be a hardware issue where both the Hub75 and Matrix are broken the night before I present at 9am.

If anyone has some advice that would be helpful.

[flat vigil]

hey folks. I'm working on a hardware-in-the-loop board and struggling to figure out how to protect the esp32-p4 3.3v gpio from accidentally being connected to a fixed 5v pin. any suggestions? I've been in ESD land but this is a sustained 5v instead of transient

[spiral pagoda]

How much space do you have? Zener and series resistor might work, or a schottky to 3.3v if you’re sure the 3.3v will sink the current

The smallest PTC fuse trips at 17mA which is probably too high

[flat vigil]

I haven't laid it out yet. it's essentially two rows of 20 pins each. kinda like a half sized breadboard

the 5v is coming from the board itself. so 5v usb -> device under test -> dut pins -> p4 gpio pins

[drowsy drift]

I'd say a 3.3V Zener diode would be your best bet, as @spiral pagoda suggests. If your setup has such unconstrained connections, you may want to put some resistors in front of the P4 GPIOs to limit the current in case you short a pin by trying to drive HIGH a grounded pin

[flat vigil]

ah, true

here is my roughed in layout

[spiral pagoda]

How many of the pins will need this protection? All of them?

Resistor arrays are easy enough, but I wonder about the diodes

Might have to settle for the sot23 dual schottkys like BAT54

[flat vigil]

Ya, I was thinking all 40

[ripe dawn]

Hello! I'm looking to incorporate a tiny boost converter on a project I'm working on, so I've been looking at some of the tiny breakout boards out there. It seems like most use a more "typical" (and bulkier) inductor, but it seems that some have this little grey box instead. Any idea if that is just an incased inductor, or some other component? And how I would go about identifying it?

[latent jungle]

The smaller component is a "metal composite inductor." It is a coil surrounded by a high permeability (metallic powder) material. The larger inductor is an open air "wound bobin" style.

The metal composite usually indicate a higher-switching frequency on the switcher.

[ripe dawn]

Aha! Thank you for those keywords. That should make searching a lot easier 😅

[latent jungle]

Cheap converters like those won't have that kind of detailed outlined.

[fluid dawn]

Hello gentlemen! I'd love some help with this one. I am trying to bring up my custom RP2350 board. This is a flight computer for a model rocket. I can provide the entire project files, but I have a few ideas of what might be wrong (I mean Chat GPT has a few ideas).

When I plug it into USB, my computer makes the USB disconnect sound and then it does not appear as a drive where I can put CIrcuit Python onto it. Chat GPT thinks it's my wiring of the USB lines or Oscilator. However, to the best of my capabilities, I think I followed the RP2350 quickstart guide on this. Would love some help!!!

I know it's getting 3.3V ok (win for test points) but I'm not sure what else could be wrong here.

[spiral pagoda]

Good to see you again! The first things to check are the power (3.3v and also the 1.1v core voltage) and the crystal oscillator (16MHz I think). That at least will tell you that your chip might be running. However, with the USB connection/disconnection you are getting, you probably have an impedance issue somewhere on your USB differential pair. My guess is up near U4 - is that a ESD protection diode? Unfortunately this is not easy to fix.

Also verify that the USB D+ and D- aren't swapped. I've done that before. If that's the only issue, you can make a USB crossover cable with a terminal block and it works fine.

[fluid dawn]

@spiral pagoda Thanks for this info. I did try to impedance match the USB pairs, they are less than 0.01mm length difference and both use the recomended series 27 ohm resistors.

I will look at the 1.1V. I am realizing I wish I had many more test points. Sad to learn lessons the hard way

Oh wow, I actually see that this section is mismatched a bit.

5.6 vs 5.1mm length

Not including the part that goes over the top

I didn't see this before.

[latent jungle]

usb 1.1 is incredibly forgiving

[spiral pagoda]

It’s not the length that’s a problem. It’s impedance and reflections

[fluid dawn]

I'm just shocked that my old board worked (kinda) and It was not length tuned and also, I forgot the series resistors. I thought this one would be so much better. But it won't even connect

[spiral pagoda]

Unfortunately it’s not possible to get the correct impedance on a 1.6mm 2-layer board. The best alternative is to make the traces short

[latent jungle]

RPxxxx issues tend to be problems with something other than the actual USB lines.

makes sure the Flash is wired right, the clock is running, and the rails are good (on a scope, not a DMM)

[fluid dawn]

This is a 4 layer board. All USB traces are on top layer. If anyone wants to take a look, here's my entire KiCad project

This is my flash wiring, same as the last board which worked.

Not sure if this is helpful, but this is Chat GPT's opinion:

"Your board is most likely failing to start its system clock because the crystal oscillator is wired incorrectly—specifically, the 1 kΩ resistor placed between the crystal and XOUT is damping the oscillator and preventing it from running. Without a working clock, the MCU never initializes USB, which is why your computer only plays a brief disconnect sound and nothing enumerates, even in boot mode. The fix is to remove that resistor and connect the crystal directly between XIN and XOUT, with the load capacitors to ground as recommended by Raspberry Pi Ltd.; once the oscillator is running properly, USB should come up normally."

[spiral pagoda]

Sure, that’s good advice. Have you checked the clock yet?

[fluid dawn]

What I am just confused about is I believe I have followed the quickstart guide in my oscillator design (I dropped my schematic and quickstart side by side above). @spiral pagoda @latent jungle if you guys are willing to double check, that would be amazing. But at this point, I'll try it because I'm not sure what else to do.

On my previous iteration, the USB connection was very unstable, it dropped and reset randomly. I was not using the recomended crystal so I figured that was the cause, but this one doesn't even show up at all.

[karmic herald]

hello, I'm creating a portable project using the raspberry pi pico W, and I want to integrate a battery circuit on my PCB so I can use it on the go

but I'm having trouble finding a good reference circuit

https://www.adafruit.com/product/5612

do you think if I took the design from this product and put it directly on my PCB, it would work?

[latent jungle]

Don't bother asking a random word generator engineering questions. They don't have any actual intelligence.

[latent jungle]

If the MCU isn't running, it wouldn't enumerate, so Windows wouldn't make any noise.

1. The Crystal is wired corrected (in Schematic). Although, the load caps might be a bit high for that crystal.
2. Why do you have pull-up and pull-down resistors on the Flash's SDIO lines?
3. I would have tied RUN high with an external pull-up, or as the datasheet says, directly to IOVDD.
4. Why do you have Capacitor symbols connected to the CC lines of the USB port? I guess you chose a 5.6k resistor. But they should be 5.1k.
5. On PCB, I would have placed more GND vias under the crystal to make sure it is well coupled to the inner GND layer. In fact, I would have used more GND vias everywhere... pretty much anywhere there is a GND pad on top or bottom.

Relying on the ground pours on the top and bottom layers isn't enough.

My experience has been the 1k damping resistor on the oscillator never works as expected. Since you're having trouble, my first step would be to replace it with a 0 ohm resistor (or short the pads.)

I always put a resistor there, because some day I'll encounter a crystal that needs it.

The use of thin traces on the 1v1 switcher concerns me. I'd definitely be checking that rail on a scope.

[fluid dawn]

@latent jungle

1. I picked the values from the Rasperry Pi Design Example Doc
2. On the flash SDIO lines, the genuine answer to your question about why I have pull up and pull down resistors is that I don't know what I'm doing. But I did have this on my previous board and it worked. That being said, I am actually not sure what you are talking about, I don't see pull up and pull down resistors on the flash data lines - see picture:

3. What is RUN high? Is this the hold pin on the W25Q64JVSSIQ?

4. Ah this is the wrong symbol, but I actually have the right part on here. Do you think the 5.6k ohm instead of 5.1k ohm could cause it to not connect/RP2350 to not boot?

5. On ground stitching vias, I tried to use the auto-via tool in KiCad and it kept freezing and wouldn't place anything so I did these all manually. I can put more in though.

Thank you for the tip on the 1k ohm resistor on the oscilator, I'll try removing it and shorting the pads. Maybe it'll help. I'll try anything at this point lol. This is actually the advice of the "random word generator" haha

6. This thin trace? It's 0.2mm thick. Also, I don't have an oscilloscope, but I could probaly borrow one from work. What would I be looking for?

[fluid dawn]

@latent jungle I removed the 1k resistor and there was a change! Now it no longer plays the USB disconnect sound. Just nothing at all happens when I plug it in.

[latent jungle]

You need to short the pads

[fluid dawn]

Haha I did I just forgot to mention it.

[fluid dawn]

I just made a discovery!

The 1.1v rail is not running at 1.1v!

But why?????

I once again copied the design guide - This is on page 7 of the design guide Maybe I have a component backwards or something?

[latent jungle]

1. Load crystal values depend on the crystal plus the stray capactiance from the traces.

2. Pin 3 and Pin 7 of the QSPI flash have pull-ups, why?

3. The RUN pin of the RP2350.

4. These pins and pull-downs tell the DFP to provide the device with a default of 5V (and 500mA?). If they were too far off then VBUS wouldn't be 5V.

5. Look at the example layout in the RP2350 datasheet. It teardrops the traces to the pins of the chip. Yours stay thin.

[latent jungle]

The inductor could be backward. The usually have a dot on them to indicate "pin 1." However, the switcher would still run.

Inductors aren't polarized in the traditional sense.

The "pin 1" helps identify which direction the winding goes around the core. Current in one direction generates less EMI around the component. Which is why it works but you'd probably fail EMI testing.

Again, you'll probalby need to a scope to verify if the switcher is actually doing anything. DMMs lie about SMPSs.

[latent jungle]

@fluid dawn also check if the 1v1 rail is shorted to ground. looking at your PCB layout, I don't see where that could happen, but still worth checking.

[fluid dawn]

@latent jungle

1. Ok
2. The answer here is unfortunately the same as before, because I don't know what I'm doing and data sheets are super hard to read. They're usually hundreds of pages and sometimes don't have an example schematic. However, this is a bad excuse - I need to get better at reading datasheets! The genuine answer is the 'random word generator' said I should pull up pin 3 and pin 7. I can cut these traces if you think it might be explaining the issues I'm experiencing.
3. Ah, ok. I missed this. Thanks!
4. Gotcha!
5. Oh I see, I did not realize this was important. Will update this in the next version.

Thanks for the info on Inductors and the 1.1V rail is not shorted to ground.

Ok, I think we found the biggest problem (so far). Besides all the little things we found here, @winter sonnet pointed out that the capacitors on my oscillator are not 15pF, but appear to be larger. The schematic specs them at 15pF, but the actual part is 15uF....

So off by a bit! I'll toss the correct values on there and see where we go from here! Maybe it'll boot!

[latent jungle]

Haha yeah, that'll give you some problems!

The pull-ups only apply for a device used in SPI mode. The RPxxxx processors put the Flash into QSPI. So just remove the resistors.

[fluid dawn]

Thank you for this additional info! The new caps are on the way and I'll install them this weekend! Hopefully with the QSPI resistors removed and the correct cap values, my board might boot - fingers crossed.

I still don't know if this will fix the 1.1V rail issues, but we'll see.

Another random question, how would you recommend connecting the shield pin on USB-C? Chat GPT said Is should connect it to the system ground through a capacitor.

[spiral pagoda]

Capacitor in series? No. Tie it into the regular ground with thermal relief to make soldering easier

[fluid dawn]

I read that large voltages like from ESD could disturb the system if tied directly to system ground.

[latent jungle]

The actual answer is very complicated. Some sources say direct, unconnected, or a combination between of RCL.

Since you aren’t going through FCC testing then you can tie it directly to ground.

[fluid dawn]

That's easy. Thanks @spiral pagoda @latent jungle !

[charred merlin]

I'm working on a personal project for a simple IV6 VFD nixie tube display (https://www.ebay.com/itm/223870481211) and I'm looking to use 5 total tubes using the HV5812 display drivers. (basically just a 20 channel high voltage shift register)

Working on the board layout, I noticed the datasheet has a **startup **and a **shutdown **sequence to follow for the HV5812s. (see screenshot for snippet, full datasheet here: https://www.digikey.com/en/products/detail/microchip-technology/HV5812P-G/4902558)

The plan is to use one of the inexpensive MT3608 boost converter modules for the 30v but I'm not quite sure how to handle the sequences for power-up/down. Current idea is ground is permanently connected together, making VDD consistent 5v in from USB then using a MOSFET switched on by a RC delay + Schmidt trigger circuit (or MCU pin) to enable/disable 30v input to VPP via that MT3608. (the boost boards don't have an enable pin, or would tap into that)

Startup seems easy enough, just curious if anyone has any ideas on an easy way to ensure 30v gets shutdown before 5v in case of power loss - is a simple pull-down on the MOSFET gate going to suffice? Thinking not because 5v is powering both MCU and the boost converter, so they'll likely drop at the same time? Maybe I'm thinking about this the wrong way and there's an easier option? Wouldn't be the first time 😁

Thanks in advance!

[spiral pagoda]

Can you run some simulations in LTSpice or something? It’s weird that they don’t give timing requirements on the sequence. Maybe simultaneous shutdown is sufficient

Also we’ll need to know the current consumption at 30V and 5V to estimate how quickly the power drops

My guess is the power-down sequence is cautious but not necessary. But I’ve never used Nixie tubes before.

[pale grove]

Question:

Assuming I create a flex PCB that sandwiches around a coin cell battery. (top and bottom)
What would be the best way to make sure the contact is good between the PCB surfaces and the battery?
Ideally I would want to keep the thickness as low as possible

[unique patio]

a spring clip? C-shaped thing

if it's in a thin enclosure, a bit of springy foam plus the force of the closed enclosure could provide pressure

[pale grove]

hm yeah maybe I try the C shaped spring clip.

Do I need to give the PCB a specific surface so it has better contact with the battery? Or not necessary?

Something like this perhaps? Not even sure if that's an option for a PCB fab

[unique patio]

that I have no idea about. I don't know of other examples that do this. I'd make the contact area somewhat smaller than the diameter so that it won't accidently fold over and short the battery on its shoulder.

may want to put some tape or something on the spring clip so it doesn't damage the flex PCB

[merry folio]

Closest I can find is this thing. In this case, they're just using a thin strip of flex PCB with arbitrarily-shaped contacts, held between thumb and forefinger. The main take-away seems to be that the shape of the contacts probably doesn't matter as much as the mechanical connection to the coin cell.
My first instinct would be to 3D print a thin pocket where the ends of the PCB slip into slots in the closed end and the coin cell slips into the open end, sticking out just enough to facilitate easy removal (see rough sketch).

[pale grove]

Yes that is roughly what I have in mind for the flex PCB shape

[unique patio]

there should be some kind of downward force to make good contact: C-spring, spring or spongy tape in a cover, tight fit in plastic, etc

velcro strap and maybe in a case like above, but that might be thicker than you want. The velcro would be a little compressible and springy

[novel flame]

Possibly stupid question, but I’m a complete newbie when it comes to LiPo and have some cautious fear of battery fires… I’d like to have some way of powering down a battery powered Feather RP2350 completely, via a switch. Would the proper way be to ground 3v3en, or to put a switch on the positive leg of the battery wire, or some other way I’m unaware of?

[tulip swift]

Grounding the EN line usually just disables the 3.3V output LDO regulator. For many boards that may effectively cut the vast majority of power. But if you want to have a real lipo battery disconnect, put the switch between the battery V wire (red normally) and the board, so not even the LDO is seeing the LIPO input. Note that for boards that have a built in charger, this will also effectively cut off the battery from being able to charge. So keep that in mind.

[ripe dawn]

does anyone know where I can buy a bare reversible usb connector like this (without the breakout)? I was hoping to put it on a project I'm designing and it feels silly to buy these and desolder them
https://www.adafruit.com/product/6001

[silk lark]

lcsc?

[ripe dawn]

have a link? 🙂 things like "reversible usb" and "flippable usb" don't turn them up there, and on mouser the only thing I'm finding is the same breakout

[silk lark]

https://www.lcsc.com/product-detail/C840351.html

there are probably more, but they are not well described

[ripe dawn]

aha, thanks! yeah I'm guessing they're on all the various suppliers, but just hard to search for

[silk lark]

it's as if they didn't care if you buy them or not

[rigid lantern]

yo what why are those not standard that's sick

[ripe dawn]

I think they're pretty fragile 😬

[cunning flint]

They're also not suitable for higher amperage in my experience too, they get hot

[pastel night]

no spec.... so 5amps is probably out... do you mean 3 amp rating "higher" amperage?

[cunning flint]

Depends on how much voltage drop you can tolerate; they get warm at just 1A

[agile turret]

hi y'all, I'm pretty inexperienced in circuit design but am trying to understand the Adafruit ADXL343 board. I was wondering if anyone knows why D1 is in this direction in the schematic? It seems like it's stopping the host device from sending a signal to the CS pin for no reason? (I can provide more info if needed, I'm never sure how much to frontload for questions like these)

[tulip swift]

The Pull up resistor to the left of the diode ensures the CS pin is high and then a host device can pull it low as needed. Note that this pin is a dual purpose pin where setting it high at power up puts the IC in I2C mode where as pulling it low at power up puts it into SPI mode. Adafruit supports I2C mode. I would think it would work fine with just the pull-up resistor and the diode not needed, but I assume Adafruit found issues getting it to reliably boot into I2C mode for some reason and the diode fixed that, although I am not sure how.

[agile turret]

Oh I see, in my mind I hadn't realized that the pins of the host device could be used to pull a line low while the line had a pull-up resistor. Does this mean any GPIO pin can be used a GND? I'm sure it's not wise for a reason, but I'm assuming the current going into Host CS has to go somewhere and that has to eventually end up going to GND.

Thanks for the help though! I guess I'll add the diode in my schematic then, the manufacturer's test board didn't include one which made it feel weird, but I'll trust Adafruit

[loud mauve]

generally io pins can't source or sink much current at all, but if you just need a digital low or high against a strong enough pull-up or pull-down, that works

[tulip swift]

As @loud mauve mentions, while you can use a GPIO pin to as a GND or V pin the limiting factor is how much current that pin can sink or source. As an example, I am running an ePaper display off a battery operated "dashboard" which get data from the cloud and displays it then sleeps for 20 minutes before the next update. Since ePaper displays use so little current, I am powering the ePaper from a GPIO pin so I can turn it off completely between updates. For anything requiring more current you would be wanting to put in a relay, transistor, mosfet, switch, or whatever can handle the current needed.

But back to your comment about pulling a line low, with a 10k resistor, when your microcontroller pulled the line low, the current the micro would have to sink is only 3.3V / 10k = 0.33mA which is well in the capability of any GPIO pin, so your micro only needs to be able to pull the node down to GND then letting go of it (and letting the "pull up" bring it back up to 3.3V).

[agile turret]

I see. Gotcha, thanks!

[upper grove]

hey guys, I was looking to use a transistor as a thermometer in my design. I've tested it on a breadboard using the EMC2101 module from adafruit with a through hole NPN transistor and the external temparature was read fine by the IC. However, I've recently manufactured a PCB which uses a much smaller SOT-23 transistor on a standalone EMC2101. This setup does not work and external temperature is read as 127C. What am I missing?

[unique patio]

3904 is a bog standard transistor. Was the breadboard one a 2N3904? I would suspect a pinout error or some other wiring error.

Wire it up on the breadboard again and check the voltages at various points

... https://www.eevblog.com/forum/beginners/transistor-as-a-temperature-sensor/ is interesting. First post sounds like your issue.

[wintry wind]

I’m curious on your pin-powered epd setup. Is the pin directly connected to the Vin with no additional components?

[upper grove]

yes the breadboard one was a classic 2N3904. I've tried swapping collector and emitter but I don't think wiring is the problem. measuring forward current gives me ~700mV on the SOT while ~400mV on the through hole.

[unique patio]

since there are no resistors in the transistor circuit I think you are depending on the typical characteristics of the transistor, which might vary across batches (not necessarily by part number). The link above suggested adding a resistor. Can the EMC2101 use some other kind of sensor on those pins?

[upper grove]

That's a good starting point, I'll dive into the datasheet see if I can find some clues

[unique patio]

Maybe the issue is that the tiny package is causing higher temperature in the '3904 itself

[upper grove]

uhm that would be strange, the reading is constant at 127C

[unique patio]

is that the max allowed?

[upper grove]

just reading the datasheet now, it seems these are actually error codes

[unique patio]

LLM notes a possible pinout error:

Pinouts differ. The 2N3904 (TO-92) and MMBT3904 (SOT-23) share the same die but have swapped base/emitter pin
assignments:

• 2N3904 (TO-92, flat facing you): pin1=E, pin2=B, pin3=C
• MMBT3904 (SOT-23): pin1=B, pin2=E, pin3=C

If the MMBT3904 dropped into a footprint or wiring scheme designed for the 2N3904, base and emitter are swapped. The
EMC2101 then sees a reverse-biased junction (or near-open) instead of the diode-connected B-C-tied-to-collector
arrangement it needs, and reports the diode fault / open code — which on the EMC2101 reads as the full-scale +127
°C.

so 127C is full-scale

you swapped collector and emitter but not base and emitter

so you are maybe shorting the emitter and the base rather than the collector and the base

[unique patio]

This says an MMBT3904 (when wired correctly 🙂 ) should be fine: https://ww1.microchip.com/downloads/aemDocuments/documents/OTH/ApplicationNotes/ApplicationNotes/00001838A.pdf

[upper grove]

it's looking very likely. I trusted the default footprint in easyEDA

[unique patio]

that does match picture I printed. I'd think EasyEDA would not be wrong ??

[upper grove]

well the datasheet has pin 2 as the emitter (bottom left pin) but the footprint places it top left so looks like they've swapped them

In fact, I've realised what I've bought is not the MMBT3904 but this https://www.lcsc.com/datasheet/C41369493.pdf?spm=wm.sxq.inf.ggs___wm.ddx.3.xqy&lcsc_vid=QFcIAwIFQwcLVgJSRVINAVBUFQBeUwYHEVUKV1xRRlAxVlNRT1RcVFBTQFVYUTsOAxUeFF5JWBEPFBcWGBMaSQgFBAJABAsLWA%3D%3D

it was probably out of stock and I've picked a random one

[unique patio]

its a darlington! well, that would explain thigns

[upper grove]

I knew it was going to be something stupid... thanks for your help!

[unique patio]

no problem! good luck !

[agile turret]

Wondering if anyone here can give any last-minute feedback on this design? The pico is to be hand-soldered and the rest assembled by the PCB manufacturer. I am using M2 screws and the holes on the pico/pcb are 2.1mm in diameter, will this be an issue or should it be fine? Thanks!

[spiral pagoda]

Can you tell me more about the purpose of this PCB? I don’t see any benefit over an accelerometer breakout board

Which, if you haven’t tried the breakout board and written your software, you really should.

Maybe you will find the accelerometer doesn’t meet your needs, or that there are other things you also want on your PCB

[agile turret]

I've already made a prototype of my project with the Adafruit breakout board, however I wish to take this to the next step which seems to be making my own pcb. I didn't really even want to this at this point in time, but the fact that the Adafruit breakout board is currently out of stock forces my hand

as far I can tell right now, either I stall deployment of my device in order to find and develop everything all over again using a different breakout board, or I make my own pcb since the accelerometer chip itself is not out of stock

this just feels like the faster and easier approach, and cheaper too

but I'm not an expert, this is not my domain of knowledge. If you know of any accelerometer breakout boards under $6 which support SPI with at least 1000hz sample rate and a FIFO buffer, by all means, I would be happy to hear about it!

[spiral pagoda]

Ok, but what are you trying to do? Surely you want some way to power the project? Or some kind of indication?

I’m also not sure why SPI is a requirement? MPU6050 has a 1kHz sample rate…

In fact I made a spreadsheet of available accelerometer breakout boards and sorted by sample rate, if that’s helpful

[agile turret]

ah, right, apologies. I'm making a collision detection device, the basic idea is that the accelerometer is constantly sampling data, if some threshold is passed it sends an interrupt to the pico at which point it wakes up from deepsleep, grabs the data, and sends it to a server wirelessly, then goes back to sleep.

[spiral pagoda]

As in car collision?

[agile turret]

I guess SPI isn't a real requirement, would just be nice to not need to change the code base and wiring much to account for a different protocal like I2C

[agile turret]

windows in urban buildings! I'm attempting to detect when a bird collides with a window

[spiral pagoda]

Thank you! That clears it up a lot

[agile turret]

oh, yes, as for power, I'm going to solder some wires onto those test pads J1 and J2 and connect it to a 5.2V battery supply

[spiral pagoda]

You probably know that the RP2040 deep sleep isn’t very deep?

[agile turret]

yes!

I switched to the RP2350 because of that as it has improvements on that regard (I am aware that there exists other boards that are even more energy efficient though)

[spiral pagoda]

If this is all you need, increase the trace width, add a ground pour, and call it good

[agile turret]

after reading the RP2350 datasheet and MicroPython source code, I've gotten my deepsleep code to run at about 0.6mA which at my battery capacity is about 6 months If I recall. Which is good enough for now

[spiral pagoda]

Just how big is this battery???

[agile turret]

2800mAH at 5.2V

roughly

[spiral pagoda]

Huh. Cool

[agile turret]

I can't tell if that's astonishment at how small that capacity is or if it's a large capacity

[spiral pagoda]

It’s a large capacity

I have some 3000mAh batteries at 3.7V nominal

So yours is basically two of those

But I never got months worth of battery life

[agile turret]

it takes A LOT of effort to reduce that power consumption. There were many months of hairs being pulled out

[spiral pagoda]

I did a solarimeter project with ESP32 and got about a week. Probably it waking every five minutes was too much

[agile turret]

I recall how flabbergasted I was when I first learnt that due to how the pico is wired, anytime you turn on the LED the Wifi chip also turns on and eats away your power at around 20mA. Maddening!

[spiral pagoda]

Ouch. WiFi is 200mA?

[agile turret]

Using the chip is about 50mA

just having it on and idle is 20mA

[spiral pagoda]

But doesn’t the MCU use 30mA by itself?

[agile turret]

even awake but idle I don't believe the pico draws that much

[spiral pagoda]

Hmm. I’ll have to measure sometime

I see you’re using a four-layer board. Are you going to do a ground pour on the outer layers?

Also, have you considered adding a reset button? This feature is conspicuously missing on the pico

[agile turret]

the implementation of deepsleep in MicroPython is actually pretty good and follows the datasheet very well (and turns off the PLLs!) which is nice. But there's more power gains to be made by disabling entire components like RAM

[agile turret]

having never had one I don't fully see the utility in it. Is it simply to force a hard reset on the code?

[spiral pagoda]

Ah. Maybe not needed with micropython. In Circuitpython it’s helpful to get into safe mode

[agile turret]

my terminology is very juvenile, I'm not sure what a ground pour is to be honest

[spiral pagoda]

This is old, but I don’t think the pour tool has changed much https://www.wayneandlayne.com/blog/2013/02/26/kicad-tutorial-copper-pours-fills/

[agile turret]

oh I see

[unique patio]

Our ADXL343 breakout is in stock at Digikey and Mouser. There are 50-60 at each

[agile turret]

unfortunately I will likely need more than a 100 at this moment

(also, if jlcpcb's quotes are anything to go by, this is MUCH cheaper at my scale)

[unique patio]

oh, sure

[agile turret]

it does feel just a little bad to blatantly steal the adafruit design and print it myself, but I swear I still support adafruit quite a bit (and have the many free coasters to prove it)

[unique patio]

we don't mind, that's why it's open source, just don't use our logos

all kinds of people prototype with our stuff and then make their custom board

[unique patio]

follow the license terms: https://github.com/adafruit/Adafruit_ADXL343_PCB. See README.md, license.txt

[spiral pagoda]

Sometimes I find insightful things that aren’t in the datasheet. For example, the wiznet chips don’t say they need pull-ups but the outputs are open-drain and won’t work without them.

[agile turret]

everything I do is with the plan of being open source. Although I will give that liscense a good read when I finally have to sit down about that issue

[spiral pagoda]

I didn’t find that out until after I made my PCB, so now I use the adafruit boards as reference designs

[spiral pagoda]

I like the idea of the adalogger PiCowbell, but I'm not happy with the battery covering up half the pins, so I'm looking for ideas to change it. So far there seem to be three options:

• Smaller coin cell battery. Do these get much smaller? This is already the smallest diameter lithium battery, so this means two alkaline or zinc-air batteries
• Capacitor charged via diode from the 5V bus. I would need 6.8mF per day of timekeeping, which is at the edge of Tantalum, so we're talking ELDC.
• Connector to external battery. This might not provide reliable power, and seems redundant anyway.

Hmm. This looks like a good option https://www.digikey.com/en/products/detail/cda-zhifengwei-technology/CBS5R5334HF-ZJ/22461731

[silk lark]

there are also vertical battery holders

call it a time snail

[spiral pagoda]

Ooh, I can get 1nA reverse-leakage diodes for cheap! FDH300A looks perfect

[slender warren]

Hey guys, I am having trouble wrapping my head around the matrix schematic in KiCad for my project. It's a music tool but essentially just a macro pad mechanical keyboard controller in topology.

I am using 36 PB-86 Switches - https://www.adafruit.com/product/5502?gad_source=1&gad_campaignid=21079227318&gbraid=0AAAAADx9JvTRG8RgG7qqPbtx2YK5YsEux&gclid=CjwKCAjw2rrQBhBuEiwAarLWHTgykH6R7kf_mc0Pgzjvw4hMR05J8pBw1WQ2rlqf8ubQ4DJc_d-14RoCEmsQAvD_BwE

I am addressing the switches with an MCP 23017
https://www.digikey.com/en/products/detail/microchip-technology/MCP23017-E-SP/894272

and the LEDs with with the IS31FL3731 https://www.digikey.com/en/products/detail/lumissil-microsystems/IS31FL3731-SALS2-TR/14308358

I learned about matrices and started setting up my schematic as shown (it's not complete, needs decoupling caps and other bits and bobs)

As I'm looking through the IS31FL3731 datasheet.. It seems to want to a very different style of matrix than my MCP switches do.

I guess it's not a problem, the switching portion and led are separate circuits, but I'm having trouble reconciling both on the same schem. This is the datasheet example.

I'm using the IS3 far below it's limits. Can I address it in a dumber way for the 36 switches I need via the coding? Or is it always going to want to see a routing similar to what's shown?

Any other LED drivers that might be a bit simpler for me? I do need individual dimming though.

Thoughts on just splitting the schem into two matrixes? I looked into it and it doesn't seem worth it to split the schem up and try to make the pcb footprint a werid combination of them later.

[drowsy drift]

what do you mean "in a dumber way"? Like not use the IS31FL3731? You have a lot of free pins left on your RP2040, you could drive the LEDs directly from the GPIO pins, if you're programming this in Arduino or Pico-SDK where you can set up a timer interrupt

[slender warren]

By dumber I maybe meant a simpler matrix style.

Rows on one A, columns on B. I only need 36 total leds. They happen to housed in the switches is all.

I'll look up going direct to mcu. I think I'd need resistors for each led but maybe it would work well.

Thanks

[drowsy drift]

ah well if you're worried about "wasting" some of the power of the IS3, I don't think you should be so worried. There are other LED matrix drivers I think, but this one is very well supported. Even Adafruit doesn't use the whole thing sometimes: check out the 15x7 Charlieplex Featherwing (https://www.adafruit.com/product/3134), it leaves off both a row and a column https://learn.adafruit.com/assets/32561

[slender warren]

Nah not too worried. Mostly I'm just having trouble displaying both matrices on a schematic.

I think I'll probably have to use two separate footprints for my switch and led even though they are connected.

[drowsy drift]

that's what I've done when I use these switches 🙂

[spiral pagoda]

Time to try it out

[spiral pagoda]

Looks like I’m only going to get one day of unpowered timekeeping. I wonder where the leakage is?

[drowsy drift]

The Pico is not a particularly good board for low-power use. You can adjust the CPU clock speed down and get a pretty linear drop in power and also use the watchdog to sleep. But even then you’ll be struggling to get lower than 1mA

[spiral pagoda]

This is with the pico unpowered. Should only be 1nA leakage back through the diode and 150nA to the RTC but I’m seeing more like 10uA

I’ll breadboard the diode and see if it’s doing what the datasheet says. I wonder if I have a 100Megaohm resistor?

[chrome bear]

If I wanted to buy some stuff to make a toy (for my grandson) that played sound when buttons were pressed etc. what hardware would I get? Ideally I'd like to play wav files or mp3 or some format thats not super esoteric

[spiral pagoda]

Propmaker feather is a good choice with the built-in audio amplifier

Then add an SD card for more storage

Or you could start with your favorite feather and add the musicmaker featherwing with built-in amplifier

https://www.adafruit.com/product/3436

Lotsa ways to go. I just like the feather footprint for the small size

[chrome bear]

I'll give all of that a Google in a bit. I think between that and our 3d printer we could make some cool stuff

[spiral pagoda]

Absolutely. Post it in "show and tell" when you're done - this may turn out to be a popular project

[chrome bear]

So that music maker would be used alongside another feather?

[spiral pagoda]

I think it requires a feather, yeah.

Looks like the musicmaker is intended for low-power feathers that can't do the decoding themselves, so it might be a good choice if you need low-power sleep

Do you know if you're programming in Arduino or CircuitPython?

[drowsy drift]

Do you want to program it with custom logic? If not, then there are these "sound trigger modules" that are pretty cheap and have 8-12 inputs that can be hooked to buttons. Adafruit sells a few https://www.adafruit.com/product/2133 & https://www.adafruit.com/product/2217. if you need more sound storage there are ones that take an SD card https://amzn.to/3PEarvV (those are harder to use than the Adafruit boards but you can find some manuals with a little searching)

[flat vigil]

Do you want it as a custom shape? You could buy a yoto mini player and hack it

[chrome bear]

Im looking for: the ability to write custom programs, load and play custom sounds, and a standard set of io pins to drive 2 to 4 servos

Ill look around at the thing you all have suggested in a bit and see what I can piece together

[unique patio]

RP2040 Prop-maker + motor/servo Featherwing:
https://www.adafruit.com/product/2928 or
https://www.adafruit.com/product/2927

https://www.adafruit.com/product/5768

there is a Prop-Maker FeatherWing and there is a Prop-Maker Feather

i posted the wrong one initially

[spiral pagoda]

Adafruit is out of stock on the propmaker feathers but Mouser has 9 available

The accelerometer with motion detection is a nice bonus

[unique patio]

digikey is oos

it has a single set of servo pins

[spiral pagoda]

@danh Have you used octopart before? Makes it super easy to check stock

[unique patio]

and you could use some other pins for servos

i have used it but forget about it -- thanks!

[wintry wind]

Those are interesting headers. I'd guess two sets soldered from opposite sides in a dual row header board. Are the round versions swiss pins?

[spiral pagoda]

The round headers are a low-profile BBL-120 from Samtec which mate with SL-120 sockets on the pico

https://suddendocs.samtec.com/catalog_english/sl.pdf
https://suddendocs.samtec.com/catalog_english/bbl.pdf

Mill-max also makes a 2.1mm mating height product, but I found the Samtec to be slightly less expensive. I want minimal height so it doesn't run into things if it needs to take a ride on a wafer-handling robot.

[wintry wind]

Thanks, they look pretty cool. I don't know if I would have ever found them online.

[spiral pagoda]

My only trouble so far is the insertion force is pretty high with 40 pins. I have to lever them apart with an awl if I want to change something

[wintry wind]

Good to know. Sometimes the standard headers from Adafruit are hard enough to separate even with the feather layout.

[native plover]

Does this look like a reasonable approach? The geometry and connector placement is based on external requirements. I tried to make groups (which are probably too tight right now) and place them sensibly, but this means that I will be routing several voltage levels and lots of I2C/UART/SPI lines all over the board.

[nova tiger]

I want to design a board that has two Adafruit boards plugged into it (one is a Feather M4, the other a DAC). I'm using KiCad. I see the Adafruit Eagle components for Feather M4 - but I want a part that is just the Feather as a monolith block (since I'm not copying the electronics onto my board, I'm plugging the Feather into it).

Do part / schematic / 3d files exist for Adafruit boards used this way... that I can load into KiCad?

[tulip swift]

You can try using the File -> Import -> Non KiCAD files and they will usually open Eagle files, so you may be able to start with the Eagle files imported.

[tulip swift]

The Feather "standard" can be found here: https://learn.adafruit.com/adafruit-feather/feather-specification

[spiral pagoda]

Yes, there’s a few in the default library

Yes, this is too tight for a two-layer board but could work with a four-layer board

[novel flame]

Is there an easy way to detect in software if a Feather RP2350 is running on USB power vs battery power?

[silk lark]

depends on how it's connected to that power

[spiral pagoda]

You could measure voltage. USB is 5V, battery is 3-4.1V

Or if you’re connected to a computer you might be able to look at the state of the USB endpoints

Otherwise no

[novel flame]

Feather has a LiPo battery installed, and also occasionally receives USB power to charge. I’d like to implement a sleep mode while not on USB power. Guess I can just go check, but I don’t suppose the USB/VBUS pin is powered if USB isn’t plugged in. Perhaps I can read that.

Though the 5v tolerance of the rp2350 gpio pins is a bit of an exaggeration, so I probably don’t want to go down that path.

[spiral pagoda]

Is there any reason you don’t want to sleep while on USB power?

[novel flame]

I was simplifying to avoid confusion, but apparently caused some instead 😅

The LiPo is connected through a controllable board, which can cut power if I send it a High signal on a pin. So, not sleep, but power off. Apologies for the mix-up.

When powered by USB, I’d like to retain the connection to the LiPo to allow charging, so no power-off desired.

In my testing, it appears to do nothing if I send the power off signal to the LiPo power board while charging, so I suppose it’s not a huge deal. I just don’t want to waste the cycles on it if I don’t need to.

[smoky terrace]

I wanna say a FET to hold the line low and using a resistor to pull the signal high, then a button to pull the line down to wake the board, (either through pulling the FET high or seperate) I'm not at home or I'd sketch up a possible circuit in KiCAD, thinking similar to an ATX power line though

[agile turret]

oops, nvm, deleting my message because I just realized I misread your comment! Didn't realize you were using a feather oops

gimme a sec, this should still be possible...

[novel flame]

I’m not seeing VBUS connected to any GPIOs on the schematic I found on the Adafruit site. Looks like WL_GPIO2 is a Pico W variant thing.

[agile turret]

yeah, I've been desperately looking at the docs and it doesn't seem promising :<

sorry for getting your hopes up! Next time I'll make sure to double-check what I read

[novel flame]

No worries, thanks for looking into it!

[agile turret]

okay yeah, I don't see an avenue to go from (unless you wanna look at D+/D-, but this isn't very elegant). Was hoping the USB's CC pins might be used but it seems to be pulled-down and not connect to anything. Ig they're not really needed since the Feather has power-management chips anyways, and while the chips have the functionality (and suggestion in their datasheets) of outputting a signal to a microcontroller to indicate what state they're in (charging or not), it was not implemented on this board it seems

[novel flame]

So, I dug into the Pico 2 data sheet. Looks like on the Pico 2, GPIO24 is set up as a VBUS sense pin. It’s connected to VBUS via a voltage divider (5k6 and 10k) to bring it down to ~3.2V, and pulled down to GND when not present. Should be simple enough to do if I end up needing it 🙂

Yeah, that worked like a charm.

[novel flame]

Was my last free pin, so I’m kinda stuck now 😅

[agile turret]

were you not using the feather? Or did you switch to a pico to solve this

[novel flame]

No, it’s working on the feather. I connected VBUS to a GPIO pin on the feather through a voltage divider and it worked. I just used the Pico 2 schematic as a guide.

Obviously not a pure software solution like I was hoping for, but it was easy enough and I had a pin to spare.

[agile turret]

oh ic, gotcha

glad to have maybe helped!

[spiral pagoda]

Yes, it’s analog input 3 on the pico, which lets you measure battery voltage

[chrome bear]

What is this LED telling me? Pro Micro nRF 52840 on battery power. with circuitpython nicenano bootloader

[ember laurel]

Autoroute

🙂

[cerulean crest]

Friends dont let friends use autorouters

[ember laurel]

Enter the AI router, drawing its models from a vast database of best practice routing.

[ember laurel]

Microsoft Electronics 360

[inland quest]

oh no

[silk lark]

Visual PCB Studio

[ember laurel]

Express

With PCB Teams

[supple pollen]

Add one character to the silkscreen and your BGA chip is pushed off the board, all its traces deleted, and other parts rotated 45°

[still sphinx]

Sorry, that ic is exclusive to office 365 members.

[pale grove]

Has anyone here tried the 0.4 mm thick PCBs?

Are they still rigid PCBs?

Or can you sort of bend them

[unreal flax]

I haven't tried them, but PCBs are fiberglass, so they would probably just snap easier when they're thin instead of bend.

[misty escarp]

If I wanted to create a plated hole, is a large via the best way?

[distant raven]

@misty escarp that’s what I do

It may not be best practice

[misty escarp]

I just added tstop to my plated hole trying to import an SVG TT

[distant raven]

But fabs see giant via as played through holes

[frigid seal]

@pale grove Yes, You could probably bend it to about a 100mm radius (if I had to guess) before breaking.

I wouldn't do it a lot because traces would probably break.

[misty escarp]

parts could pop off too right

[frigid seal]

Yeah, depending on the part, orientation and more.

[misty escarp]

weird, polygon import "worked" but it's not showing up. EAGLE shows a polygon on tstop but nothing shows on the board 😦

I remember now trying this before and it being a headache

[distant raven]

Could make a custom part for what you’re trying to do

Then add the part to your design

[misty escarp]

hmmm

I thought about that

Can you do the SVG import into a library?

Ah you can do DXF

[distant raven]

Yeah and bitmap import too

Bitmap requires inverted BW bitmap though

[misty escarp]

Looks like it's working, the scale is just way off though

I resized my svg but it seems it unscaled it when I saved as DXF

thanks for the suggestion @distant raven worked great

[distant raven]

💪🏻

[misty escarp]

Me: Tries to convert a complicated SVG to DXF and import into EAGLE.
EAGLE: Hold my beer

[distant raven]

loool

[misty escarp]

I guess you're NOT supposed to use EAGLE to make tarot cards for your weird friends

[distant raven]

hahahaha

cursed PCBs

😉

[misty escarp]

that's probably why EAGLE didn't work, didn't want me to summon baphomet or something IDK how that works

[distant raven]

haha

just ordered some 8MB (64Mb) SPI Flash

going to use them to test and add support for larger SPI Flash in CP 🙂

[tough matrix]

and i found a bunch of old pcbs i created as mounts for HC-SR04 sonar sensor, and put them up for sale, too

[misty escarp]

"good" sign, eagle is taking forever to import a DXF so maybe that means it's more detailed?

seems it is, it's just scaled terribly!

how does one apologize to a computer program

[silk lark]

with reverence, as to not anger it

[misty escarp]

I got on my knees and offered my CPU more thermal paste, hopefully that helps

[tough matrix]

@misty escarp this way you just encourage it!
You have to be firm, show it that there are rules and if it doesn't play by the rules, there will be consequences. You may have to show your computer who is the master - maybe even ground it for a couple of days.

[misty escarp]

https://tenor.com/view/star-wars-the-mandalorian-mando-this-is-the-way-armor-gif-17546902

[silk lark]

@tough matrix practice on your cat

[misty escarp]

should work about as well

[tough matrix]

with the cat... I have this poster on the wall, next to their bowls: https://cdn.dopl3r.com//media/memes_files/how-to-negotiate-with-your-cat-step-one-give-your-cat-whatever-they-want-step-two-end-negotiation-D0ymz.jpg

[misty escarp]

at least they let you give them scritches

that's a decent bargain

[tough matrix]

so in this case it is quite clear who is the master in the house
they are

[inland quest]

0.000001% chance i'm routing this by hand

[ember laurel]

is that really the optimal placement of those components?

[inland quest]

nope

it's a rough draft

when it comes to PCB design, i really have no idea what i'm doing

[ember laurel]

put them in a smileyface config

[inland quest]

LOL

[tough matrix]

your schematics was more orderly - you cna try and follow the same placement as much as possible

[inland quest]

yeah that's what i did

i created a second layout that's based on how i set up the breadboarded version

and i based this one off of that

[misty escarp]

waiting for Eagle to import this DXF is a real lesson in the meaning of the word asymptote

[inland quest]

lol

[misty escarp]

What'sthatpokemon.gif

[inland quest]

fancy design

what's it for?

[misty escarp]

Just realized that gif has a flashing something, i hope it's not something rude

[inland quest]

you might want to remove that

[misty escarp]

It's gonna be a PCB with a Tarot card on the tStop layer so it comes out copper/gold in the white spots

It's flashing too fast for me to see, is it something awful?

[inland quest]

if it's who i think it is, the context is somewhat messed up

[misty escarp]

Oh well I'll remove it

[inland quest]

👌

[reef flicker]

I have a slight worry about Eagle here. I made both sides ground pour, and obviously there is no connection to ground here in this picture, yet it does not create an airwire or flag that it isn't connected to the net

This time I noticed it but I'm worried that maybe there's something I might've missed that it's not flagging

[misty escarp]

I assume you've pressed rat's nest?

[reef flicker]

Rat's nest?

[misty escarp]

I think that's the official name?

[reef flicker]

Hm... can't seem to find it

[misty escarp]

It's here on mine

[inland quest]

same

[reef flicker]

Couldn't find the button for it, but I'm using fusion so it may be a hidden

[misty escarp]

ah

[reef flicker]

Regardless, I've run the command, but it didn't do anything

[inland quest]

hmmm

what version are you using

[reef flicker]

Fusion 360 2.0.9313, I'm not sure what version of Eagle is underlying

But should Eagle be able to flag this as not connected?

[inland quest]

i'm not really sure tbh. sorry

i'm going to try to autoroute the PCB even if my computer turns into a smoking pile of goo

[misty escarp]

I'd need to see more of the board I think

Is there a via somewhere we're not seeing?

EAGLE is pretty good about catching these things

[reef flicker]

I triple checked and I can't see one

See if maybe I'm blind?

[misty escarp]

Where is "here" in the picture?

[reef flicker]

The whole of GND's net

[misty escarp]

In blue?

[reef flicker]

See how it's basically isolated from the rest of the net?

Correct

The easy fix is just to throw in a via I assume, but I'm concerned it's not putting an airwire or something

[misty escarp]

Weird

I'd place a via to the top layer and then run ratsnest again and see what happens

[reef flicker]

Didn't do anything special

Could it be that somehow the pads go through the board to the other side?

[misty escarp]

Shouldn't be

[reef flicker]

Maybe it has to do with the fact it's not removing the material from the plane above, but it is under?

[misty escarp]

I feel like I'm going crazy, unless I'm missing something there should be an airwire there?

If you're worried about it just drop vias to the top

[reef flicker]

I will, but the problem here is that I can't trust the software

Unless I'm missing something

[misty escarp]

I think we're both missing something tbh

[reef flicker]

No airwire on the GND net of this specific sensor

[misty escarp]

dumb question, but did you name the polygon GND?

[reef flicker]

This is what I would expect

Yes, it's named GND

[misty escarp]

Maybe try closing and reopening eagle?

[reef flicker]

The classic 1-2 huh 👀

[misty escarp]

yup off and on again

[reef flicker]

Looks like the airwires are back...

[misty escarp]

hayyyy!

Where you expect them?

[reef flicker]

Yes, but they're behaving weird again

[misty escarp]

how so?

[reef flicker]

🤔

[misty escarp]

that looks normal?

[reef flicker]

Bottom right pad and middle pad aren't connected to the net

Event though they're connected to each other

[misty escarp]

hm

I have literally never seen this behavior before

what do they look like on your schematic?

[misty escarp]

if you drag the GND symbol that's on SDO/SA0, does it have a green wire?

[reef flicker]

It does

[misty escarp]

Ok cool I've made that mistake before, and I didn't think it was happening here, just wanted to be sure

[reef flicker]

Yeah I really got no clue here

[misty escarp]

Only thing I can think of is that the integration with fusion360 is buggy

hopefully someone else here can figure it out

[reef flicker]

Could be

[misty escarp]

you don't have vanilla eagle right?

[reef flicker]

I do, but it was a pain to use 😅 found this easier

[misty escarp]

Ah haha

[reef flicker]

Because eventually I'll have to integrate it with the 3d design workflow of fusion

[misty escarp]

Are the files the same type?

[reef flicker]

They're cloud so I don't have the files, but I can export them as .brd

[misty escarp]

Ah IDK maybe try opening the .sch and .brd together in vanilla eagle and see if that fixes it? Then we'd know if we're missing something or if it's a F360 problem

[reef flicker]

Opening the .brd file in Eagle leads to the same behavior

This is so weird lol

[misty escarp]

for real

I was just watching the twilight zone so it might be my fault for bringing that juju into the chat

[reef flicker]

For some reason it looks like it's not detecting that part as orphan

I must be missing something right?

[unreal flax]

Does the "orphans" checkbox in the polygon properties change the behavior?

[misty escarp]

that was my next question!

[reef flicker]

Kind of, it properly removes the planes with absolutely nothing connected

There is a property called "Airwires hidden", but toggling that on/off doesn't fix the issue

It really thinks it's connected.. somehow

Is there a way to make it show me the route a signal takes?

[unreal flax]

You can highlight a signal with the 'show' tool, but it'll probably just consider the whole polygon to be the ground signal.

[reef flicker]

For some reason, setting the ground plane as "Hatched" displays the expected airwires

Bottom left GND pad displays the expected airwire

[misty escarp]

This seems like the kind of thing autodesk would want to know about

[reef flicker]

Can a pad... connect to itself?

[misty escarp]

hm?

[reef flicker]

I know it sounds crazy but

Nevermind

Yeah I really have no clue

Welp. Guess I don't get nice things like ground planes 🙂

[reef flicker]

@misty escarp Just for fun, if you try wrapping a pad connected to a pour plane with another signal as to cut it out, do you get an airwire or anything indicating that it's a problem?

[misty escarp]

I would think so

but my heart hurts looking at that make it stop

[reef flicker]

I just wanna make sure 👀

Actually in this case it does work for some reason... Running ratsnest adds the expected airwire

[misty escarp]

Did you ping me? I got a notification sound but nothing is showing up

A second time that is, i got the one above

[reef flicker]

Nope

[misty escarp]

huh

ghosts in the machine

[reef flicker]

Many machine ghosts tonight

[misty escarp]

https://tenor.com/view/cartoon-ghost-haunted-gif-5446914

[reef flicker]

I see online that people say you should route the GND anyways

[misty escarp]

Is there a way to make the top/bottom polygons fill more of the board?

[reef flicker]

Are you asking me?

[misty escarp]

No sorry the channel, unless you know!

I have a small board with a design in tStop and I would like the polygon to go closer to the edge of the board

Ah I think I found it, I was googling the wrong terms again

[reef flicker]

Should I be worried about traces being near/on top of each other?

Crosstalk & whatnot

[cerulean crest]

Im not following exactly what you are doing, but probably not

If you work with high speed signaling, RF, etc, then lets talk 😜

[reef flicker]

Alright

All it is is a bunch of i2c sensors with an IC

Idk if that counts as high speed?

[misty escarp]

I2C is relatively slow

[cerulean crest]

Not at all 😆

Im talking like, the RAM on your computer type fast

Many of Ghzs

[reef flicker]

Oh I'm very far from that

Glad to hear I have one less thing to worry about 😛

[misty escarp]

I'm excited to see this project it sounds super cool, no pun intended

[reef flicker]

God me too

Spent a lot of time on this, hope it works out

[misty escarp]

That feeling hits close to home

[reef flicker]

I'm still waiting on my first test boards, so I'm guessing we won't have a result on this till 2021

[misty escarp]

that's a long time, is OSHPark slower to CAN?

[reef flicker]

I don't know 😦

It shipped nov 10th

[misty escarp]

dang

I'd follow up with them they are pretty quick for me in the states at least

I guess with the Covid of it all

[reef flicker]

Yeah, that's what I assume

Last UPS update was nov 29th

https://tenor.com/view/mr-bean-waiting-still-waiting-gif-13052487

[misty escarp]

Bummer

[misty escarp]

Keychain for a friend who is a doctor

[cerulean crest]

Make sure to conformally coat it or somethin, I have a PCB ruler on my keychain and it wore out rather quickly

[misty escarp]

Oh good idea, I have some polyurethane I think

The copper wore off?

[cerulean crest]

The gold finish, then the copper discolored

[misty escarp]

Hrm

I was hoping to ship directly to them but I guess I can coat them first

[cerulean crest]

Ah, yeah thats a tad annoying

[misty escarp]

oh well, I can ship in an envelope for the cost of a stamp

I'm loving this Eagle-as-tchotcky-maker method

[cerulean crest]

Huh, Ive never seen the word tchotcky before, and the US has its own definition... weird

[misty escarp]

it's my favorite loan word

[cerulean crest]

In North America, in general, its a decorative object, in the US its a pretty girl, elsewhere in the world its seen in jewish communities which is how it eneded up in the us

[misty escarp]

I've never heard it used to mean a pretty girl but I'll have to be careful talking about collecting them

I figured it was Yiddish but good to know

[reef flicker]

Welp this looks like it's about it

[misty escarp]

looks great!

[reef flicker]

Gotta fix the silkscreen, breadboard test the design, design the 3d print and then we're off to the races

[misty escarp]

You know this but make sure to BB test before ordering! I've made that mistake once

[tough matrix]

BB?

[misty escarp]

Breadboard

[tough matrix]

oh

[misty escarp]

my hands are cold so I decided to save a few characters and then spend them now 10 times over

[inland quest]

I'm back and the autorouter mangled everything

so i need to do it all by hand.....

😶

i'm really not looking forward to this, but i know it's going to be worth it in the end

i might even turn it into a kit!

[misty escarp]

I don't envy your position either

[harsh pendant]

deos this look ok? any routing issues?

[inland quest]

looks good

is it a flight computer?

[harsh pendant]

Yes!

[inland quest]

do you watch BPS.Space?

[harsh pendant]

yes! In fact this is a tvc rocket just like his. I had my first successfull fligh this month!

[inland quest]

oh congrats!

[harsh pendant]

thanks!

[inland quest]

last year i tried to put a small flight computer on a model rocket. issue was it wasn't running fast enough and it oversteered

i might try it again sometime

[harsh pendant]

oh what processor did you use?

BIG F

[inland quest]

Attiny85 (thats why it failed lol, a 1mhz AVR isn't fast enough)

it was before i knew how to make a PCB, so it was all done on a tiny protoboard and stuffed into the top or the rocket

[harsh pendant]

I suggest using teensy 3.2 or LC

[inland quest]

i also didn't really know what i was doing either

[harsh pendant]

also are you on the amateur tvc discord server? they are a very helpful and supportive community

[inland quest]

nope! i'll have to check it out

so, are you creating a custom design for your rocket?

[harsh pendant]

yup! well this one is a scale model of the Falcon 1.

[inland quest]

dudeeeeeee

thats awesome! will it be able to land itself too?

[harsh pendant]

falcon 1 never could land itself... so no unfortunantly

[inland quest]

OHH

sorry i thought you were calling it falcon 1 because its small

[harsh pendant]

check out my instagram if you wnat more info!

[inland quest]

i don't have an instagram, sorry 😦

i don't really have any social media

[harsh pendant]

oh thats cool

AHHH just pinged random person

[inland quest]

👀 uh oh

does the rocket use vectored thrust?

[harsh pendant]

yup!

[inland quest]

mine used fins

[harsh pendant]

oh that is rly cool! canards are much more dificult then tvc

[inland quest]

oh yeah totally

the reason that (i think) mine failed was the servos i used just weren't powerful enough to deal with the high air flow

[harsh pendant]

just sent you invite to amatuer tvc

[inland quest]

so they got locked in the maximum position they could go to

[harsh pendant]

ahh i see. THat could be a big problem

[inland quest]

and i ended up with a electronically controlled lawn dart

[harsh pendant]

lmaooo

[inland quest]

what's the height of the falcon 1 model you made? also what's its highest recoded altitude you got it to?

[harsh pendant]

Havent flown it yet. THat is what this pcb is for. It is gonna be about 1.2m tall

The one i flew was not modeled after anything in particular and went to about 60-70m

[inland quest]

60m seems pretty good

mine went like 10m before loosing control

i have some cool stuff to talk about, but i'm going to say it in the TVC server cause it has nothing to do with PCB designing

[harsh pendant]

op well 10m is better than none

oh ya sounds good

[inland quest]

👌

[harsh pendant]

THis got kind barried. Anyone else see any things that may cause problems?

[cerulean crest]

Looks good to me, buy the champagne

Although, they dont do that to spacecraft do they? Lmao, wack a billion dollar spacecraft with a glass bottle

I was thinking about flight computers the other day because I was watching someone on youtube and they were having issues, interesting set of problems I wish I had time to make Arduino friendly.

What do you do to make the rest of the rocket?

I assume you make it yourself not a kit, just wondering what materials and processes you use. And when buying things like nozzles, where do you go for that?

[unreal flax]

No champagne, I think, but they do put them through various torture tests in thermal-vacuum chambers and shake tables...

[reef flicker]

@harsh pendant that's sick, what are the PCB dimensions? What size motor? How do you protect it from the ejection charge?

I've designed a few 3d printed rockets, but they were all meant for very small motors, I doubt event that MCU would fit inside

[harsh pendant]

Yes it is not a kit. Most of the parts are caded and then 3d printed like the TVC mount, flight computer mount, nose cone, and ejection system. THe rest of the parts like the body tube ( ultr light carbord tube) and couplers and stuff is bough from either apogee components or madcow rocketry. I do not make the motors, as that is pretty dangerous and there is a massive sullection of comercially available ones.

[harsh pendant]

the pcb is 100/35mm. THe aiframe i am using is 74mm. and for this one i am planning to fly the lower stage on an f20 and the upper stage on an E6, both long burning 24mm motors.

[reef flicker]

Oh wow, that's a lot of power

[harsh pendant]

Yup

[tough matrix]

@harsh pendant can you provide higher-resolution images? There is one spot I wanted to check out

[reef flicker]

How would I go about importing an adafruit board in my design? I'm looking at the itsybitsy and want the footprint to attach the board through some headers, but I want the dimensions to be accurate

I have found the .brd & .sch files

[harsh pendant]

@tough matrix maybe this one?

[misty escarp]

@reef flicker In the past I have made my own footprints using the create library tool

Based on the board files from adafruit, with vias for each pin

Just make sure to upsize your vias 🙂 I have made that mistake twice!

[reef flicker]

I'm thinking of using a jumper object or smth

[misty escarp]

that should work better

[reef flicker]

I'm just worried about the dimensions idk

[misty escarp]

18 hours later

[misty escarp]

You can get the position from the adafruit .brd file and then do some simple geometry to translate them to your file

[reef flicker]

Will do

[misty escarp]

should be a fun process and not too hard. Both EAGLE and Adafruit have guides for making parts

[pearl tapir]

Digikey and Mouser have links to load footprints that have worked well for me. But I have only used them for Mentor.

[reef flicker]

Yeah, didn't find them for this particular board

[tough matrix]

@harsh pendant not much better - 220x562 is not nearly enough to see details
Anyway, my question was about the component circled in yellow here:

what is it? looks like both contacts are connected to the same signal....

[misty escarp]

So I spent over a day importing a DXF into EAGLE for a lib part only for it to be 0.15 in tall

[silk lark]

meanwhile you can just import SVG files in Fritzing

[misty escarp]

How does that work?

[silk lark]

fritzing uses svg internally for everything anyways

[misty escarp]

I'm trying to put something on the tStop layer to have a design on the board

[silk lark]

hmm, I don't think you have direct control over the soldermask, though

[misty escarp]

I think you can?

[silk lark]

in Fritzing I mean

[misty escarp]

Ahh

[silk lark]

I wish that Inkscape could simply save gerbers

[misty escarp]

Eagle lets you import an SVG into a .brd but it doesn't work well at all

skerr pointed me to this method and it's much better but also clunkly

SVG - > DXF via an online converter (Inkscape's doesn't work well) and then import into .fpt

[misty escarp]

Aesthetics opinions? Thicker lines maybe? Gonna use black soldermask

This is only barely PCB design, apologies

[tough matrix]

would it pass DRC? 🙂

[silk lark]

there are no traces, so yes

[misty escarp]

it passed EAGLE's basic one

Nah I hand drew all that in traces XD

[silk lark]

uh

[misty escarp]

I kid

It's just a polygon and tStop

[silk lark]

that's heavy humor

[misty escarp]

about 1oz heavy

[harsh pendant]

@tough matrix nope it is a voltage reg

[tough matrix]

what is the red outline under the voltage reg?

oh, I see, a capacitor - it is juts that in the low-res picture, all signals seemed to merge into each other

[harsh pendant]

ahh i see

[misty escarp]

@harsh pendant which reg did you use?

[harsh pendant]

buck 5v 2a reg

[misty escarp]

ah cool, was just curious

[tough matrix]

with buck voltage regulators it matters where you place the components - e,g. you don't want to put the inductor far from the IC. It is best to closely follow whatever design they have in the datasheet

[elder peak]

Yeah, you really don't want to buck the trend on layout.

[frigid seal]

groan

[harsh pendant]

I fallowed the datasheet

[reef flicker]

Look what I got in the mail today!!

[misty escarp]

wooot

time to break out the sandpaper

[distant raven]

💪🏻

[reef flicker]

These pads are TINY

Lord knows how I'll get this soldered

[distant raven]

Hot air

Or hot plate

[misty escarp]

yeah

[reef flicker]

Scared they'll merge the pads together

[misty escarp]

merge?

[distant raven]

Use flux

[misty escarp]

https://tenor.com/view/delorean-mcfly-fluzo-condenser-gif-9780783

[reef flicker]

I have paste, idk how it's not going to solder the pads together

[vapid grove]

Idk, I’m going to solder mine on the frying pan lol

Also update on my MacBook: It’s fixed now. The hinges were broken, but now they are replaced with the new ones!

Only thing is I can’t afford to buy my components now ((

[reef flicker]

I have access to a true reflow oven but yeah

[misty escarp]

I'd do that

[reef flicker]

Can it really be that precise? Don't have a stencil

[misty escarp]

I've seen people use syringes for applying paste

[reef flicker]

Yeah, it came with a syringe

But still, the distance between these pads is like... 2 human hair

[misty escarp]

I wonder how adafruit does it then

I mean they do have a really fancy setup but it's gotta be doable

[reef flicker]

Stencil-less?

I mean I can give it a shot

[misty escarp]

You've looked a bunch, time to leap!

[reef flicker]

But I'm pretty sure those pads are just gonna merge

[misty escarp]

yeah you're in a tough spot

Fortunately you can test for continuity using your breakout header points

[reef flicker]

The pads are under the sensor, so I won't have access to those for testing continuity to the pads

Only detecting shortage

[misty escarp]

Yeah

[elder peak]

Well, in your case I wouldn't want you to be padding your chances of success.

[harsh pendant]

https://www.instagram.com/p/CIXFjsHhXdu/?igshid=16qt4y1hp7wfk

Check it out I made pcb vid

Oh 😂

[ivory jasper]

I've started laying out my hall effect keyboard PCB but I have a few questions for the experts... I've got per-key RGB LED via ws2812b-b (5V/12mA current version) which means the LEDs sit right next to the analog hall effect sensors. Am I going to run into (serious) trouble with the LEDs messing wth my analog signals or is that not going to be too much of a problem as long as I follow <insert simple rules>?

(BTW: Only reason I went with the full-size ws2812b-b is because JLCPCB is out of stock of basically everything else right now! Also, that particular version doesn't require caps or resistors which is sweet... Still would rather have the mini version though)

[supple pollen]

The current flowing to the LEDs will tend to create a magnetic field, but a small one, and should largely cancel out. However, there will be some induced noise (especially with PWM). How relevant that is depends on the strength of the magnetic fields you care about.

[ivory jasper]

The plan is to have the whole back side be a ground pour... That worked fine for my numpad PCB (mV wobble wasn't too bad that I couldn't deal with it in software)

[supple pollen]

Fortunately, the path the return current takes through the ground pour will tend to follow the path the supply current takes, and they should largely cancel.

[ivory jasper]

Right now I've got two 5V lines: The one coming out of the LDO and the one coming over USB (or optional external USB C and barrel jack connectors). The external 5V power goes right to the LEDs but the hall effect sensors are getting power from the LDO. Should I just use a bigger LDO that can handle more current and put them all on the same 5V source?

[supple pollen]

Like most engineering decisions, that depends on what you're trying to optimize.

[ivory jasper]

Well, it's a two-layer board and I'm thinking that unless I go four layer it's probably best (from an analog signal standpoint) to put all 5V on the same LDO. But I'm not an EE so I'm not so sure if that's true

BTW: The board also features a USB hub. Still haven't laid that out yet though (it'll be at the top edge--away from everything else)

[supple pollen]

My feeling is just have one LDO for lower parts count. However, to separate analog domains for lowest noise, individual power supplies might be useful. I'm guessing super low noise isn't a requirement here, so I'd probably go the simplest route with one shared LDO. The number of layers doesn't really impact that choice, although it can affect things like noise coupling.

Note: I'm not an EE either, so treat my thoughts as the guesses they are.

[ivory jasper]

Well, with a four layer board I could have a ground pour between the different power sources and between the digital signals which supposedly is "what you're supposed to do"

[supple pollen]

Exactly: when going for lowest noise, that sort of shielding and separation become more important. However it seems to me you're more concerned with magnetic noise than electrical noise, which changes the landscape somewhat (it generally simplifies things).

[ivory jasper]

...but I'm probably vastly overthinking this design. My numpad PCB didn't get this much thought and it works fine with an attached-via-headers microcontroller hehe

[supple pollen]

I suspect to: I'm also often guilty of overthinking. In this case, since you've built a similar device and the noise was manageable, I don't anticipate it being an issue with this one.

[ivory jasper]

Oh I don't care about magnetic noise. The hall effect sensors aren't that sensitive. It's the end millivolt value coming out of the sensors that matters... If it wobbles too much it's a real pain to deal with in the firmware and makes it so you can't set as fine an actuation point

(well, more coarse control of the actuation point of the switches)

[supple pollen]

Ah, there are other sensors and millivolt signals involved: that's where noise can be a real problem.

[ivory jasper]

The numpad didn't have per-key RGB LED. That's really the new thing. It also got all its power via the USB bus

...but since it was powered by the USB bus and I know the Black Pill board doesn't exactly have a high quality LDO (hehe) I think this board design should in theory be better from a noise perspective--assuming the RGB LEDs don't impact much

Can I just use multiple 1000mA LDOs or do I need to pick one big one?

Only reason I ask is that the LDO I've chosen is a "basic part" at JLCPCB hehe

I can have as many as I want and it doesn't impact the price much! 😄

I'm already at 9 non-basic parts so I do have a free "slot" as it were

The built-in USB hub goes into "powered" mode when USB C or the barrel jack is connected but I don't have an LDO for that. I suppose I should just eat the cost and pick a beefy LDO that can handle it all

(reading article now about paralellizing LDOs... Interesting!)

[supple pollen]

Ah, that's relevant too: it does make a lot of sense using the basic parts, in multiples if need be. And having separate regulators for your LEDs and sensors will give the sensors cleaner power

[ivory jasper]

These AMS1117-5.0 LDOs are $0.14/each so I can "afford" to give each USB port its own LDO and also give the WS282b-b LEDs their own LDO as well (they should take up a max of 852mA)

YOU get an LDO and YOU get an LDO! Everyone gets an LDO!

[supple pollen]

Yup. Sometimes that's the way to go

[ivory jasper]

Yeah, pricing and logistics reality have already resulted in me modifying this design many times haha

I originally had a super sweet USB 3.1 hub IC that had built-in UART so it could provide the same function as a USB FTDI breakout. It even supported that same sort of functionality for I2C and SPI! The plan was to have this one awesome universal keyboard that was infinitely repairable that had header pins on the side for all your embedded development needs. Turns out JLCPCB doesn't have that part in stock (anymore) and it was also like $12/chip. Though, it would've let me do USB C Power Delivery on two ports. Man, what a beast that'd be!

[supple pollen]

That does sound like an amazing part

[ivory jasper]

OMG it's friggin awesome. It's like my dream USB hub IC haha

It's been a while since I looked at it but I think it was this one: https://www.microchip.com/wwwproducts/en/USB5734

There was another big caveat to that part that I found out later though: If you plan to use it you're supposed to contact Microchip and work with their consultants to develop the firmware for the thing. Apparently it doesn't just work out of the box.

(that might have been their 8000-series Smarthubs though--I forget)

[supple pollen]

I had an issue like that with some other chips, where the configuration details were hidden. I've avoided that vendor since.

[ivory jasper]

Yeah I'm never touching anything like that

Not only that but first order of business if I ever had to work with hardware like that would be, "Well, as long as you're OK with us posting all this code publicly to Github..." 😄

Just found this: Microchip did a video (which is kinda low production value haha) where they demonstrated the features of their Smarthub ICs. They demonstrate controlling I2C stuff from a tablet that's connected to the hub: https://www.youtube.com/watch?v=6NxoAQYn4OQ

[supple pollen]

I still have a beef with Microchip from the days when they wouldn't document the protocol for their chip programmer

[ivory jasper]

The IC has its own GPIOs and is essentially a fully-featured Cortex-M0+ ARM device. When you order 5000 of the things they're only like $4/chip but if you order one from a PCB assembly service it'll end up being 3x that 😄

Yeah, they kinda suck from a big-corporate-still-doesn't-get-it standpoint

Do I need ferrite beads for each LDO or can I just split off from one for all of them?

Never mind: They're $0.01 each. Ferrite beads for everyone!

[pearl tapir]

@ivory jasper Static LEDs won't create much noise but dynamic(multiplexed) will. If you are switching the LEDs, stop that process while doing analog measurements.
You don't always need a ground plane. It depends on what is underneath the analog traces. Low level analog signals can be run as a transmission line where a ground trace equal in width trace and parallel to the signal is run.

Using transmission line concepts, I've designed RF boards in single sided with 40 dB return loss.
Sometimes people over complicate a design without measurements to justify extra cost.

Verify that the ferrite beads have significant impedance at the frequency of interest. They are not magic bullets. Sometimes at a low enough frequency, they have little or no effect and a low value resistor is a better choice. And on both sides, close to the R or L, use a small value ceramic capacitor. If I'm really worried about noise, I'll parallel 100pF, .1uF and 10uF. That is probably overkill but it tends to insure high frequency dielectrics. If you really need high frequency capacitors, which I doubt, I'd recommend Johanson Technology.

[ivory jasper]

@pearl tapir Thanks for that info. For reference, these LEDs are addressable (not a matrix). They pass the data signal from one LED to the next.

[pearl tapir]

Like Neopixels?
Those will generate a lot of uncorrelated noise. Uncorrelated is good as the noise doesn't add directly. But I don't know if you stop the transmission how long it would take for them to stop switching. A dual trace scope on the input and the final output could tell you the delay time.

[cerulean crest]

I have a beef you spend all that money, then when you ask for their free compiler, they give you a version knowing full and well it has issues

For instance, in a if statement in their latest version, the bitlength of integers was changing even though both variables were the same length, and 16 bit on a 16-bit MCU

[elder peak]

I feel like Microchip really loathes their customers.

[cerulean crest]

Actually no I take that back, it was a 8 bit integer, let me see what the micro was

LMAO the processor I was using has a 8 bit data path so that is funny to me

Wait.. my cpu had to do extra work to give me a wrong answer? 🤔

[ivory jasper]

@pearl tapir Yeah they're basically the same thing as Neopixels... I suppose I can stop sending data to the LEDs while the microcontroller spends ~300µs reading all 80 analog multiplexer pins. The LEDs won't turn off during that time though so I'm not sure how much it would impact the total noise

[pearl tapir]

@ivory jasper My guess (and it is just a guess) is that they are static once they have received data.
That can be checked by stopping everything from switching and using an "EM-probe" basically an antenna hooked to a high impedance scope probe laid along the LED signal path while the LEDs are lit.

Possibly, if you are running the LEDs at less than full intensity there will be PWM noise.

[tough matrix]

regarding Microchip: I had little experience with their tools, but I like the Atmel/Microchip hardware - e.g. SAMD chips

[supple pollen]

Amusingly, I jumped ship from the Microchip PIC to the Atmel AVR. But now Atmel is part of Microchip, who has cleaned up their act some and now documents their protocols and their tools run on more than one platform.

[elder peak]

Yah, I'm just annoyed because STM32's come with burned-in bootloaders. atmega32u4's come with burned-in bootloaders. SAMD21's don't

[supple pollen]

TI's SBW technology makes it easy to install code on their chips

[misty escarp]

silly question but does anyone know if JLCPCB's gerber viewer shows more than your board outline? I have a polygon that extends just a bit past the rounded corners of my board and it's showing in the gerber viewer. I don't recall this happening before..

It also doesn't capture all of my soldermask design 😦 I could use OSH which seems to but it's literally 5 times the price

[supple pollen]

There are other Gerber viewers out there

[pearl tapir]

OSHPark recommends gerbv, which can be installed locally, and claims that is the same as their online viewer. OSH requires a board outline file and they cut the board to that.
Somewhere you have to tell the board manufacturer what size your board is and the CNC router normally cuts outside that outline. Anything that hangs over is typically cut away and discarded.

[distant raven]

Their tool also works with the .brd files

[misty escarp]

I tried PCBWay's gerber viewer and it also captures the whole design. I confirmed that the CAM job is outputting the soldermask as expected too. Will probably just pull the trigger on JLCPCB and hope they get it right

[distant raven]

Which is nice that I don’t have to generate every gerber file

Even oshstencil gets the stencil file when you upload the .brd file

Eagle also have a gerber viewer built in the cam window

[misty escarp]

Oooh I wonder if I could get a stencil of this design

Then I could make prints!

[pearl tapir]

I prefer to generate the gerbers myself and look at them to be sure what I'm going to get.
I typically have text on each gerber file, well outside the board that tells the layer and part number. On a more complex board like multilayer or aluminum core, I include the layer stack up including dielectric, solder mask color, silk screen color and any testing, etc.

One maxim of quality control is that you get what you inspect not what you expect. You have a better chance of getting what you want if you look at the gerbers.

[misty escarp]

Using a gerber viewer?

Thanks btw

I'm just a little uncertain if JLC's gerber viewer is making a mistake or is showing me what they are capable of fabricating

For example, PCBway's gerber viewer shows angled corners where I have rounded. But I believe PCBWay is capable of rounded corners

[pearl tapir]

http://gerbv.geda-project.org/

Most PCB routers are Excellon or Hitachi. Both can do outside mitered corners or radiused. To avoid breakage, the largest diameter that consistent with the board specifications is used. More often than not, in the USA, that's 0.125" diameter. So inside corners won't be square, they will have a radius of the router bit.
The manufacturer should state the size of the router bit or minimum diameter for inside corners somewhere.

[misty escarp]

cool thanks I'll see if the windows version works, I believe JLC is Excellon. Fortunately I just have outside corners, which are rounded so should be OK. Good to confirm about inside corners, I had an idea for a board with a cutout

[distant raven]

I email OshPark to get my designs panelized in a way to save $$ and get all the gerbers that way

They are super helpful

[misty escarp]

I love them they just cost so much

This board is 5.8x3.5 so it's real expensive thru them

[distant raven]

Yeah

[misty escarp]

I'm gonna use them for my M4 board since I don't need a large number of them and I like their purple mask

[distant raven]

Medium run brings cost to $2/sqinch

But you pay ~$200 for a medium run

[misty escarp]

that's more feasible if these were to be sold. So far my friend doesn't think there's a market for these in their store, but wants one as a present

I will probably end up making a bunch with a laser and some found wood

[distant raven]

Set up your own tindie *

[misty escarp]

Eh I don't have the contacts to the community they do, I'd be an interloper heh

[distant raven]

My phone tried to autocorrect tindie as to die

[misty escarp]

Time is setting up my die for me 😉

[distant raven]

I’m kind of unknown in the maker community but have done okay so far not selling on tindie

[misty escarp]

they have a small Etsy store selling crafts and jewelry. I thought I had such a great idea here but it was panned as a product heh

[distant raven]

ahhh a new set of beta boards order 🙂

I think it was @elder peak who mentioned that most of the USB-Serial boards don't have PTC fuses and other safety features. I'm probably going to sell one for about $9

FTDI with PTC resettable fuse, and whatnot

[misty escarp]

wooot

[elder peak]

Ooooh, yeah that was something I mentioned

[distant raven]

the two other boards on that are a high precision temp/humidity sensor and a 7 key capacitive touch breakout

[elder peak]

Sometimes I say helpful things instead of puns

[distant raven]

😉

[elder peak]

But, yah, I have a CY7C65213 breakout from SparkFun (largely because FTDI's antics with bricking devices that left me with a dead Monoprice serial cable means makes me resistant on knowingly buying their stuff) and it's got some protective diodes, but nothing on the USB D+ / D- lines and nothing on the power lines.

Mostly, the PTC fuse has saved me from zapping my stuff the most.

[misty escarp]

would this be better than the arduino - board method of programming

[elder peak]

A non-USB AVR-based Arduino tends to be a serial port (originally a FTDI but they switched to an ATMega32u2 or ATMega16u2) permanently affixed to the side of an AVR prototyping board. Obviously, the newer ARM-based Arduinos with native USB are going to work very differently.

So, when making arduino-alike boards, you could make it all as one piece, or you could make it with a 6-pin FTDI-ish header and just use a serial breakout.

[misty escarp]

Ah gosh I was wrong in my terminology. I was actually planning on using a feather m0 with SD Card to program CP onto a custom board. That's what I meant

so not an actual arduino brand board

[elder peak]

I feel like I could set up a Tindie to sell larger quantities of the tastefully odd breakouts that I make for myself except that I really don't want to actually run a Tindie and be accountable for delivering orders on time.

[distant raven]

I sell boards to fund other board projects lol

[elder peak]

I ration my delivering-things-on-time energy for my dayjob so I can do board projects.

[misty escarp]

post when you launch your FTDI board, I may order one

[elder peak]

Well, I've got a ATSAMD21 on the way to experiment with bootloaders and I'll probably use a Feather M4 for that.

[distant raven]

That’s a pretty good idea

I use the feather m4 express for my dap for boot loading

[elder peak]

Would you say you use your Feather as an a-dap-ter?

[misty escarp]

@distant raven when you did your build for CP, did you have to add any pins to the board file? I think I will need to

[elder peak]

Mostly I feel wronged by Microchip because they could have just put a perfectly reasonable USB bootloader on the part but didn't, probably because they hate me.

The ATMega32u4 can totally be DFU'd without any accessory hardware.

[misty escarp]

I wonder if they are just big enough that they don't have to, market forces and all

like they can force customers to deal with it

[elder peak]

Yah, there is a strong "you will beg, peasant" vibe from most electronic parts manufacturers.

[misty escarp]

https://tenor.com/view/shrek-lord-farquaad-smug-walk-gif-4571303

Microchip: Some of your projects may die, but that is a sacrifice I am willing to make

[distant raven]

@misty escarp yeah, I build a custom uf2 using the Adafruit uf2x-samd GitHub fork

[misty escarp]

Specifically I want to add to board some pins on port B

[distant raven]

Different than the circuitpython uf2 process which I’ve got a build for

2 actually

[misty escarp]

interesting

This is totally new territory for me to I'm excitiscared

[distant raven]

I feel that

It’s much easier than it seems

[misty escarp]

Eventually I want to write some actual c-micropython-circuitpython so I can do port masking

[distant raven]

Port masking in C is fun

Though I’m not sure it’s supported yet in CP

[misty escarp]

Idk if it's really simultaneous but if it isn't it's probably fast enough that makes no difference

I want to make it possible!

I'll start with my application and then see if I can't write something that makes it possible in vanilla circuitpython

[distant raven]

That would be interesting

Add support for ISRs natively in CP

[misty escarp]

hahahahahaha I wish

they are working on async capabilities

but that's beyond me

I think micropython style interrupts are never coming to CP

[distant raven]

Probably not

[misty escarp]

I used to be sad but now I'm excited to see what adafruit does to replace the functionality

[elder peak]

I mean, interrupts are, by definition, strange fae beasts.

So trying to do an ISR in anything other than assembler, C/C++, or Rust seems questionable.

[misty escarp]

It's been a while since my micropython days but IIRC they have some serious limitations

[distant raven]

Yeah, it’s how you handle the interrupt calls tends to be the issue

[reef flicker]

Assembled my first PCB!

Surprisingly the spacing between the headers is perfect for a breadboard

It's a shame though because the board is slightly too long and doesn't leave space for jumper wires

[misty escarp]

nice!

[reef flicker]

Yeah yeah it is

Now just gotta figure out how I'll test it

[misty escarp]

You might be able to use wire under the board to the rows that are available. If it's thin enough

They also make female headers with long pins

you could buy and cut those

could also solder directly to the pins

[tough matrix]

"what big logic gates you have!" - "To better understand logic, my dear"

this is for my new project, a demonstration kit to explain logic gates to kids

(Beer bottle is not part of the kit)

[misty escarp]

That's a notsober gate

[tough matrix]

"Friday night logic gate"

Beer NAND chips

[misty escarp]

"gettin thru quarantine one at a time"

Just checked, and my board are almost here! So excited. I just got my parts from digikey

[tough matrix]

but I was surprised that something like this hasn't been done long ago
actually it was, but rather poorly

[misty escarp]

Oh I thought those were printed (ha) on the paper, they are actual boards. Cool!

Where's the actual gate? is it on the back?

[tough matrix]

they are actual bords - with electronics on back

on top, there will be just three indicator leds

[reef flicker]

@misty escarp I have some female headers, think I'll just use those on one side

[tough matrix]

haven't soldered them yet

[misty escarp]

@reef flicker are they the long ones?

[reef flicker]

They don't look particularly long

[tough matrix]

@reef flicker one trick is usign two breadboards

[misty escarp]

Again, another use for a wide breadboard!

[reef flicker]

Omg why didn't I think of that lol

I have a second tiny breadboard that I don't use

[tough matrix]

you might need to break off the part with power rails - it is usually detachable

[reef flicker]

As for the female headers, they're just the same size as the male ones

Wow it is detachable

Why is it?

[misty escarp]

Ah totally get it if you want to get to prototyping asap but you can buy long leg female headers that let you plug into a Breadboard and have female on top

[tough matrix]

just for cases like this

[reef flicker]

Perfect! Thanks for the idea I'm dumb 😂

[tough matrix]

no, it is actually non-obvious idea
but I didn't invent it - saw it on another forum long ago

[reef flicker]

It also means I could've used it even if the pins didn't accidentally match up the dimensions of the boards this is great

[misty escarp]

and you still have power/gnd rails!

[reef flicker]

Yay!

[tough matrix]

yeah, you know that the spacing between pins of Arduino Uno on one side is 0.05 inch

[reef flicker]

Time to figure out what's gonna make it blow up!

[tough matrix]

instead of standard 0.1in

[misty escarp]

these are your level shifters right?

[tough matrix]

to make life interesting for everyone

[reef flicker]

It is yeah

It... works? Both ways!

[misty escarp]

ay!

that's great

[reef flicker]

Doesn't appear to heat up or generate any significant noise in the signal

This is perfect!

[distant raven]

Nice nice!

[distant raven]

Me: orders boards
Also me: checks email 200 times waiting for panelization email

[misty escarp]

For anyone curious, in case they get the same issue. I heard back from JLC. They have an issue with their online gerber viewer, it sometimes doesn't show gerbers correctly. They are working on it

[distant raven]

Oh gotcha

I ordered from OshPark

But I order tiny boards 🙂

[misty escarp]

Hey I try to make mine small! usually

[distant raven]

Well, a lot of mine are usually around 1sq inch

Lol

JLCPCB is definitely a good choice for large boards

[misty escarp]

for sure, I'm gonna use OSH for my "smaller" board. I tried to get it smaller than it currently is but it's got some parts that are rather large

[distant raven]

Also boards panelized and are off to fab

[misty escarp]

woot!

[distant raven]

Same day 🥳

Should be back around Dec 15th

So I should have the PCB before Christmas

[misty escarp]

woot!

I should get my tarot card before then too and my keychain. I need to resin or polyurethane them per iotpanic's great suggestion

That gold leaf plating is thin

[distant raven]

Yeah 🙂

I’m excited to see your board 🙂

[reef flicker]

OSH was by far cheaper than jlc for my PCBs, especially given that shipping is free and they don't charge extra for ROHS compliance

[distant raven]

Yeah

[misty escarp]

I can't wait to see how it comes out either. My project boards should be here wednesday

[distant raven]

If more people order from Osh, they can do more. They told me their 4 layer medium run is just above breaking even. But if you order enough volume on 2 layer, they can give you some discounts

[misty escarp]

IDK why but JLC always seems to have a discount deal going when I order from them

5 boards was $4.60 with ROHS

[distant raven]

Overseas PCB manufacturing is heavily subsidized

[misty escarp]

that makes sense

Yeah if I were to sell these I'd probably do a "buy american" thing but since it's a one off gift I'm ok with using overseas labor. It would have been over 100 dollars to use oshpark

[distant raven]

They have to be to undercut local fab companies

It’s kind of insane how much Chinese electronics fab is subsidized

Like trying to do DoD certified PCBA with like AdvancedCircuits is $150+

The $150 I think is the base tooling fee

It’s one time, but still crazy

[misty escarp]

Their labor is probably also really cheap relative to ours. Although I suspect PCB fab is really automated

*ours being america's, sorry non american friends, i am working on not being so US centric

[distant raven]

It mostly is

But again, Chinese government heavily subsidizes manufacturing activities

They do it to make it almost impossible to be able to choose any other services