Dead Voicecard Atmegas

Came across a problem today and thought I’d share it here in the hope that someone can shed some light on the cause. Good to learn from your mistakes and all that.

My Ambika with 6 svf’s was built and fully functioning a couple of months ago, but I had used an encoder shorter than specified and hadn’t installed a volume pot. Decided to sort these little issues out, so replaced the encoder for one with the correct length shaft, soldered a volume pot and put it all back in my metal case.

Upon testing I noticed only voicecards 1-3 were functioning, nothing out of 4-6 at all. I opened the case to check and the red leds for the dead voices weren’t lit up. (on close inspection they were lit up very dimly). Assumed I must have shorted something on the motherboard when changing the encoder so started checking that. Eventually I realised voicecards 1-3 worked in any location, set to either A or B, and the 3 dead voicecards wouldn’t work anywhere, so something was wrong on the 3 voicecards.

I now know the Atmega chips on the 3 malfunctioning voices are not working. I’ll try reflashing them and if that fails replace them so problem should be solved. I’m just interested if anyone has any thoughts on what could have caused this? I only removed the voicecards closest to the encoder when replacing it, could the heat from desoldering have damaged the chips? Find that a bit weird as its the chips furthest from where I was working that got damaged. Alternatively could I have a sketchy joint somewhere that could short temporarily and fry the Atmegas in the voicecards?

Any thoughts welcome, like I said I’m always keen to learn whats gone wrong.

Cheers,

Jim

> I’m just interested if anyone has any thoughts on what could have caused this?

When a voicecard cannot complete the boot sequence, it locks itself in firmware update mode (and thus expects you to reinstall the firmware from the SD card). That’s a safety measure to allow to recover easily from a failed firmware update.

The problem is that this safety measure can be triggered if the ambika is powered off immediately after having been powered on (eg when you fiddle with the power connector).

That could well be it, thanks!

I definitely noticed it flash on and back off quickly once, but that was after the 3 voicecards went and I was testing things out. It could well have happened before and I didn’t notice though.

I’ve got the kit to reinstall the firmware so no big deal!

Hi Olivier , Jimmi, i noticed here the following. One of my customer had build an Ambika . He makes some mistakes and change the 4050 against the 74HC138 (IC4). Than he insert 3 voicecards and switch the power on.
All Voicecards are death ! We find the mistakes by phone, he put the IC in the right place now , but , all voicecards are death – no reaction.For testing i had in my studio an Ambika - all works.
I remove the 74HC138 (IC4) from my MOBO , switch Ambika on , yes , all voicecards are death. But , if i insert the 74HC138 again and switch the power on, the voicecards are death again
No reaction I take a look, nothing . I had to flash all new !
Hmmm, Olivier , how can i flash the voicecards direct with the SD card ???
Greets

Andre’

I might have the same problem with two of my six SVF boards. On these two boards no LED are lit (I checked the LEDs) whatever the place 1,2,3…6 they are installed and configured for. I will try and swap the processor with a working board to check if it’s a firmware issue…

@tubeohm search for ‘firmware update’ on this page manual

OK I swapped the processor of a working SCF boards and the “bad” SVF boards are resurrected meaning either the processors of these boards need a firmware reinstall or they are dead.
Well it looks like I have to read the “firmware update” chapter in the manual then…

……

OK I managed to reinstall the voicecard firmware on the two silent processors. Now the 6 boards work… Still one board is exhibiting an abnormal behavior : first I could not install the firmware on this board, I had to place the processor on another board to flash it successfully. On the faulty board the installation was aborted immediately. With a freshly flashed processor the faulty board has a different filter tracking range compared to the others, it cannot be corrected with the frequency tracking trimmer. Furthermore the filter cannot oscillate while the others whistle like a charm. I already tried a swap of the DAC IC. Any ideas ? I swapped all the other ICs except the SSM2164, I must do that… Any other ideas ?

@ tubeohm thats an interesting observation.

While replacing the encoder I removed IC4 from its socket. You’re saying powering up the motherboard with ic4 removed kills any voicecard attached. Its possible I powered up before putting the chip back in but I cant remember that happening. For my case I think Oliviers suggestion of a partial boot locking the chips is the likely culprit. I had 2 voicecards removed to give access to the encoder, but only 3 of the 4 attached voicecards died.

Anyway reflashing the chips solved the problem for me.

@yusynth was your faulty board ever working correctly?

@yusynth I would check the joints on the trimmer and on the ISP header for that board as a start

Thanks jimmy_p for your suggestions.
Concerning your first question I finished to assembly the voice boards yesterday as a matter of fact.

I will pay attention to these solder joints at least those of the trimmer (still this would not explain the resonance problem). I don’t think the ISP connector is involved for the flashing problem since I used the upgrade procedure based on the SD card.

Sounds like a misplaced part. I had this same issue with a card. I don’t have the schem in front of me so I can’t remember which part was the culprit, but I do remember I put 100n caps where the 100p’s should be. I also had to replace the 18k with different ones as the tolerance difference had become an issue. I really should write this stuff down as it happens and what the fix was…

I swapped all ICs, checked all resistor marks, all capacitor marks, and tested zeners. I also set the trimmer value similar to the other five boards, no success, the board sounds duller (filter don’t open correctly) and does not oscillate.

Ideas ? Testpoints ?

I will do some voltage measurements at the F_CONTROL and RES_CONTROL points…

Could you post a photo of the board?

Sure Olivier, I will post one this evening. It is perfectly similar to the five others… but may be fresh eyes might see what I am blind for :wink:

PROBLEM SOLVED, I took photos and check the solder joints and although they did not look bad I re-soldered the legs of the DAC socket and shazam now it works, therefore thanks jimmy_p it was a bad solder joint after all.