CV issues

[mezzman]

I am building a bunch of RadioMusic modules (rev2 PCBs) and I am having odd issues with a few of them. I am seeing the Reset LED and bank indicator LEDs flash while a sample is playing. Sometimes this happens with CV into either Start or Tuner position inputs. Sometimes running audio out direct into an external mixer will cause this (ground loop?), but going into a VCA before the mixer fixes it. Other times changing voltage going into Start actually triggers reset on voltage change. And other times just a constant voltage into the tuner position input causes these to flash very rapidly and audio starts stuttering. Video of issue: https://vimeo.com/117084179 password: radiomusic

[TomWhitwell]

I've started a troubleshooting page in the wiki
And added a few things that might be relevant to you.
Because you're the first - everything on that page is relevant to your questions!

You can get stuttering 'between' stations - either from knob position or from CV input - I haven't yet found any way to avoid that

I'm going to close this issue now, considering it a question answered, but be sure to re-open it if you need more help.

I'll leave issues open once we identify a hardware or software issue that needs to be fixed

[mezzman]

I ran the test sketch and I am seeing an identical issue across several boards. What I am seeing is Channel CV and Time CV are noisy with nothing plugged into the jacks. LEDs 3&4 are flickering. Channel CV ranges from 1 to 280 and Time CV is 1 to 400. Introducing CV into Channel jack makes both LEDs 3&4 change intensity, but only Channel CV changes in value. It still is noisy, as it jumps around 20-30 points. CV into Time jack has no effect on LED 3&4 intensity, but value does change and is also noisy.

[TomWhitwell]

Hi, take a look at issue #13 re: values for R21 and R10 https://github.com/TomWhitwell/RadioMusic/issues/13

Also, if you have Rev1 boards, there is a ground error: https://github.com/TomWhitwell/RadioMusic/wiki/PCB-Version-History

See the fix here regarding the legs on R21 and R10: https://github.com/TomWhitwell/RadioMusic/wiki/Building-the-Radio-Music-module

[mezzman]

I am using rev2 boards and the following resistor values: R10 47k R21 47k I tried 22k at R10 and R21, noise slightly less.

In test mode should Channel CV only affect LED3 and Time CV only LED4? I have one board that is not exhibiting the noise problem, but when testing Channel CV, both LED 3&4 react to voltage level change, and when testing Time CV, both LEDs still change together, but are much dimmer. Seems like this happens across all modules, it was just more difficult to see with the noisy ones.

[TomWhitwell]

Those values work fine with my setup, so I don't think that's the problem.

[mezzman]

Are you getting individual LED control for each CV input on the test sketch?

[TomWhitwell]

Hi, I've updated the test sketch (there was a bug in the LED display, sorry) And made a demo video to show how it's supposed to work: https://github.com/TomWhitwell/RadioMusic/tree/master/Collateral/ModuleTest

[mezzman]

Thank you. I downloaded the updated test sketch and now I am narrowing the problem down to the Time CV input. It is noisy with nothing plugged in, but the problem goes away when I input 0v via external source. I am measuring -0.1v between pin20/A6 and Gnd on the Teensy. I am seeing the same issue across 10 modules and I have gone through all the solder points and checked the usual suspects. It's very odd, maybe I have a bad batch of TL074?

[TomWhitwell]

I sounds a lot like a hardware grounding issue; a bad connection somewhere. The ground path goes all the way through the Teensy itself, so it could be anywhere on the boards.

I think I did that test video on a patched rev.1 board, I'll see if I can find a rev2 board to check.

@monobass Have you seen anything like this?

[TomWhitwell]

On an unplugged module, put a patch lead into one of the CV inputs, and use a multimeter to check the resistance between tip and sleeve. It should be 47K

[mezzman]

I am getting around 43k on both CV inputs for all modules.

[Thonk-Steve]

I'm seeing 47k on both Station and Start.

[mezzman]

I removed the TL074 and connecting pin 8 on the socket to GND at the CV jack and it stops the noise. This pretty much rules out bad grounding. I am suspecting the opamp is the culprit, will try to find some locally today and swap it out.

[mezzman]

I tried a different manufacturer of TL074 and also a TL084 with no improvement. back to square one. I just do not know why this is happening with all boards of mine. Have you tested any v2 PCBs yet? Are DGND, AGND, and GND all tied together? Is it possible for a ground loop to happen? Shorting GND pin with pin20/A6 or pin22/A8 on the Teensy attenuates the noise but it does not go away as I would expect. I'm truly at a loss here.

[mezzman]

I am going to do a separate build with PCBs from a different source to test with, I will close this issue now.

[Thonk-Steve]

mezzman what value pots did you use? and also did you have a panel on the module as you were testing?

[mezzman]

50k linear Alps (from small bear) on both pots. Yes I had grayscale panel installed. I removed panel and there was no difference

[Thonk-Steve]

Ah I wondered cos I ran into some cv drifting issues when using metal-shafted 100k pots, particularly on the Start pot... And even more so when the panel was off and they weren't grounded. Once I swapped back to 50k and put the panel back on the problems disappeared again.

[mezzman]

That's interesting. The issues I have had were only on noise on the CV inputs, the pots were rock solid.

[Thonk-Steve]

Yeah they are rock solid now. I updated the bom to warn people off using 100k pending further testing.

[mezzman]

I would also say stay away from Itead as a PCB fabricator, I used a PCB made by Seeed and it worked perfectly with no noise issues at the CV inputs. I have also had confirmation from others that Itead has had quality control issues.

[Thonk-Steve]

Oh yeah iTead are definitely worth swerving! Their boards have been pretty poor for a couple of years now. Seeed protos on average much better from a manufacturing standpoint, the silkscreen on protos is pretty hit and miss. This issue should be closed then?

[mezzman]

Yes, this issue has been resolved as far as I can tell. Progress has been made.