Help troubleshooting my pcb

[hummusdude]

After populating the board and powering up the arduino nano would not load the sketch if it was in the board socket. I was able to recreate the same error message when it was not in the socket by jumping the reset pin on the nano to ground so like a noob I figured I'd found the fault. But when my hack to isolate the reset pin didn't fix the problem I kept probing and eventually I noticed that the entire 5 volt rail was showing about 4k8 ohms to ground. I also tried jumping the reset pin to 5V when it was in the board but it still would not load the sketch.

I know boards sometimes have manufacturing defects but since this is my first, and my first time working with smd parts, I'm guessing the most likely explanation is that I messed up something. I tested the other boards (they were made by jlcpcb) and they all seem fine. I did some spot checks on the current board before soldering on the parts but it never occurred to me to think that I should check every net for continuity.

At this point I'm not sure what else I can do. I've removed the five bypass caps on the digital components...that seemed like the likeliest place to look for a bad joint or a partial short but I'm still seeing 4k8 ohms to ground from the 5v rail. I've inspected it carefully using my lighted magnifying lens but I don't have a microscope so I can't be 100% about the smd solder joints on the pads. They all look okay but without a microscope I can't be sure. To me, 4k8 ohms seems like it could be microscopic solder bridge or something similar.

[TimNJ]

Do you have a photo of your assembled board?  My usual suspects are components that can be put in reversed, i.e. Check your 3-pin voltage regulators are not in backwards. Check if your diodes are in the right way. And don't trust that the footprint and the schematic symbols are correct and matching up with each other. 

If you had a thermal camera, this would be a nice time to use it. If somethings in backwards, usually it'll get toasty. Alternatively, check the power/current draw of the board in this condition. Is it crazy/above expected?

[hummusdude]

I checked the diodes and regulators and everything looks good there.

I ran some tests on current draw and found some curious things. Without the nano installed, the pcb draws about 25 mA at 25 VDC. The regulators are cool to the touch. With the nano installed, running my sketch, it draws 135 mA and the regulators get quite warm. Since the Arduino IDE was giving a time out error I thought perhaps there was some problem with one of the reset pins. I pulled the nano and was able to recreate the error by jumping the reset pin to ground and trying to load the sketch. But when it's in the board I see 5 VDC on the reset pins so not sure what to make of that.

I installed blink for a reference and I can see the current switching between 55 mA and 115 mA every second on the DMM. Maybe my meter is too slow to track it, but it seems odd that the on-board LED would pull that much current.

Some other possible sources of problems... Pin 10 on the nano socket is grounded somehow. Coincidentally, Pin 10 goes to the chip select pin of the dpot, and it is active low so perhaps not a problem in the end. I tried loading my sketch with the nano off the board and pin 10 jumped to ground and everything worked fine.

At this point, aside from the excessive current draw the main symptom is that the nano is somehow getting stuck during the load process and won't accept data. I did test pins 1 and 2 and didn't find anything obvious there. It make sense to think there is some connection to the mystery current drain and the nano not loading. I've cut the traces of the 5VDC rail to the reset pins but that didn't solve the problem either.

Sorry if the pics aren't that good...just my phone camera. If there's a spot that looks suspicious I can try a zoom pic to get a better look.

[oPossum]

arduino nano would not load the sketch if it was in the board socket.

Don't connect /reset to +5V. That will not allow reset via USB so you will not be able to program it.

[hummusdude]

I don't have a need to reset remotely other than through the usb connector so it sounds like it's okay to leave the reset pins floating. Is that right?

[oPossum]

Yes, leave it floating on your PCB. The nano has all the necessary reset circuitry on it.

[hummusdude]

Another behaviour is that blink will continue to run if installed off board and then the nano is put in the socket, even with no power to the board. But my sketch will not run if I do the same thing.

[hummusdude]

Disconnecting the reset pins from the 5 volt rail solved the issue with the nano not loading sketches while installed on the board. And the sketch now works when the nano is in the board socket as well so all good there.

And I isolated the issue with the cs pin of the nano being grounded. It turns out that pin 1 of the dpot, the cs pin, is itself showing 4.7 ohms to ground. My sketch keeps the cs select pin on the nano (pin 13) high until it is needs to send serial data to the dpot. So I was losing about 80 mA there. After I cut that trace the current draw of the nano at idle is about 50 mA.

Unfortunately, I suspect that the dpot is toast at this point. I've had the worst luck with dpots. When I was breadboarding I lost 3 or 4. Not sure if I'm being careless with them but none of the other ic's that I've worked with have been so sensitive. I assume it's ESD. Are they known for being hard to handle?

[oPossum]

I don't think Microchip's digipots are exceptionally ESD sensitive, but you are running it outside of its operating range. The terminals of the pot must stay within the supply rails. A much better choice would be the DS1882. It is log (rather than linear) and has a bipolar analog supply.

https://datasheets.maximintegrated.com/en/ds/DS1882.pdf

[hummusdude]

The terminals of the pot must stay within the supply rails.

Do you mean that because the input is an ac signal it is outside of the supply rails? The input does reach - 340mV pk, which is less than ground. Is that the issue?

[oPossum]

Yes. The absolute max is -300 mV. As a practical matter it should not go below Vss or above Vdd at all. I can't say if this is what may have caused damage, but it is outside the spec.

[hummusdude]

It looks like I made a  mistake in assuming I could use a dpot in the same way as an analog pot. From what I can tell it might work if I add a voltage divider in front of the dpot to DC bias the input signal. I found this app note from Maxim https://www.maximintegrated.com/en/design/reference-design-center/ref-circuits/161.html that goes into detail about why my original approach doesn't work. I hope my example of what doesn't work is helpful for others thinking about making something like this. Looks like it's back to the drawing board for me!