[Solved] Very strange problem. Both SIDs go mute until I change patch?

[ytsestef]

I have one last problem in my MBSID build (last for now at least :P )

Under two conditions both SIDs go mute (no sound at all, not even background noise).

If I change patch or do an "Init Patch", the sound comes back.

These condiions are:

1) I ground the audio gnd to 0V. I mean the moment the contact is made, the sound is gone. If I change a patch while the audio gnd is connected to the chassis of the build, the sound comes back until I disconnect and reconnect it. But that doesn't really bother me practically but I think it's a symptom of the same problem as the second *really annoying* instance of the symptom:

2) I have a modulation wheel connected in a controller that is configured as a voltage divider. It's a 10K linear pot that has 0V on pin1, 5V on pin 2, and the output of the wiper goes to the MIDI controller. When this pot reaches high values (close to 5V - ie close to value 127) the sound disappears (same way as before). If I change patch without touching the modwheel, it comes back on until I move it again. Midi messages work OK, values change, everything else is fine.

Notes:

TOTAL WEIRDNESS #1: If I move the modwheel slowly, it works as it should (the sound doesn't disappear), if I move it fast in that high value zone, the mute SIDs strike :P

TOTAL WEIRDNESS #2: In bassline engine, it works perfectly even if I move it fast. It just works. Actually I think the problem exists only in Lead engine.

#3: The pitch-bend wheel uses the exact same setup (voltage divider 10k lin 0V-5V = 0-127) and works without any problem.

Any ideas? I'm really at a loss. I already tried putting a 3k resistor in series with the pot to reduce the maximum voltage (even if it reduced modwheel range/resolution), but the symptom remains... Next step is to change the pot, but I won't be able to do that soon. because I'm going on vacation on Monday.. I'm not in a hurry to fix the problem, I just can't find peace of mind until I FIND it :)

Edited September 6, 2012 by ytsestef

[ytsestef]

Update: I was playing around, making some patches, made one where an LFO controlled VOL L/R for tremolo effect, moved the modwheel and the sound did not disappear completely, instead got very very low in volume (the LFO made it modulate between nonexistent and faint). This may have a lot to do with the problem only existing in Lead engine. Is there any fixed assignment in all the patches? Maybe a weird bit is toggled by the controller and the SIDs volume drops?

The VOL in CFG menu seems to stay in value 120 after the sound is gone, but if I change the value to anything else (like 121 or 119) jumps back in as loud as normally. If I move the modwheel the sound disappears and value stays where it was. I change it again, it works. Is this parameter listening to CC#7? When a new value is sent via MIDI, does the indication on the menu change? Could it be that the MIDI controller sends CC#7=000 due to whatever reasons when the modwheel pot is close to MAX?

Edited August 11, 2012 by ytsestef

[Imp]

The pot of the modwheel may be faulty. If the wiper looses contact, values will float and generate lots of garbage data.

I'd try the pot of the pitch-bend on the ain of the modwheel. Don't forget to clamp the ain of the pitch-bend to ground, while nothing is connected.

[ytsestef]

Right, thanks a lot! I'll try it :)

[ytsestef]

Changed the pot, same happens... :frantics:

[ytsestef]

Update: It also happens when I plug/unplug another device on the same power outlet. For example I unplug the soldering iron from the wall while the SID is playing, BOOM, SID "stops". I go in the CFG menu, tweak the VOL value by an increment, SID is back on...

I suspect power/grounding scheme errors here...

[Imp]

I suspect power/grounding scheme errors here...

Yes, sounds legit.

I'd check the sid's supply voltage for dips, while you provoke the error.

It depends on your Multimeter though, if short dips can be detected.

[ytsestef]

I don't think my multimeter is that fast.

Could it really be a drop in voltage? Seems like the Core is getting a CC#7=0 message every time this happens. Maybe garbage data are produced?

Oh, and something else (not sure if related). I power my build using a +5V/+12V/-12V switching PSU from Meanwell. So, since it's already regulated, I have omitted the 7805 in the Core module as TK suggests. If I leave it like that, it works, if I short the outer pins of the pads where the 7805 used to sit (in order to connect the cap directly to the input), the whole build becomes unstable and does several restarts until it works ok.

[Imp]

What is the power rating of your PSU?

Which pins did you connect it to?

[ytsestef]

The PSU is the Meanwell RPT-60B, it is 60W: http://www.google.com/url?sa=t&rct=j&q=&esrc=s&frm=1&source=web&cd=1&ved=0CFwQFjAA&url=http%3A%2F%2Fwww.meanwell.com%2Fsearch%2FRPT-60%2FRPT-60-spec.pdf&ei=UCYoULLlOpKT0QW8zYH4DA&usg=AFQjCNH8gKBBjoU0jYk0eDxtkwIVbZjEbw&sig2=gZ56aSt-8cs9HB-d1LGOAA

I have connected it to J2 of the Core Module, since the power is already regulated.

[Imp]

Ok, there's no need to bridge the pads on the core where the regulator used to be. It would only give you some more caps between gnd and +5V anyway. Maybe these caps overload your PSU at startup, when connected. Can you supply +5V from a different (linear) PSU for testing?

[ytsestef]

I tried a different PSU (linear) for +5V supply. Nothing changed. I was able to try a few experiments as well. This is what I found out (everything that follows happens with both PSUs):

1) When earth is connected to chassis and SID outputs are only connected to PCB, as soon as the wiper of the modwheel pot is close to 0V, the sound stops. If I connect it directly to the ground or chassis (exactly 0V), the sound is stuttering and SIDs click between zero volume and normal volume (this happens about 5 times/sec - imagine something like a square LFO assigned to volume with 5Hz rate and max depth). This happens ONLY in lead mode, in every other mode the modwheel behaves normally, even when the wiper is connected to ground (even though the modwheel has mechanical stops at 2k and 8k resistance, so MIDI value 127 is something above 0V). After a while (sometimes sooner than others) a high A (maybe A5 or A6) is triggered, and can't be stopped (probably if I send a MIDI NOTE OFF message it will stop, but I only have a 3-octave keyboard (with no octave selection buttons), so I can't test it. Note: the stuck note thingy does not happen with the wiper floating or high values of resistance between 0V and 5V. As soon as I turn the power off the stuttering sound accelerates (while the caps discharge) to maybe 10 or 15 stutters per second until it finally runs out of juice and turns off.

2) When earth is connected to chassis and SID outputs GND are connected to chassis as well (ie wall socket earth meets 0V at one point), the sound disappears every time the wiper changes resistance in the lower values (2-5k). Example: I load a patch with wiper close to 0 ohms (close to 0V). As soon as I move the wiper to the other end (towards 5V), the sound stops. I leave the pot there (high resistance = high voltage), I change the patch. Sound works OK, I move the pot to the other side (close to 0V again), sound stops again. Only in Lead Engine. Stuck notes don't occur.

3) When both earth and audio gnd are not connected to chassis (whole build is "floating"), same as case 1) happen. Stuttering sound etc.

4) In ANY grounding configuration, as soon as I plug or unplug anything else in the same power socket (for example, my soldering iron), the sound stops. Again, until I change patch, or do an INIT patch.

5) Probably irrelevant, but no harm in mentioning it: I have connected a dual gang log pot across the outputs of the SIDs and the ground, for analog master volume control. Doesn't seem to make a difference, though.

PS: I suspect a wrong grounding scheme in the MIDI controller PCB (the one I ripped off the master keyboard). MIDI is opto-isolated, maybe I need to do something to that board?

PS2: Could the CORE module have any problems? Can anyone explain this behavior?

Edited August 16, 2012 by ytsestef

[Imp]

You could monitor the midi data of your controller.

Midi-OX or Mios-studio can do that.

That "square-LFO-phenomenon" could be related to the controller sending incremental volume changes. Maybe faulty ones.

[ascarrar]

What exactly does your MBSID look like? Is it mb-6582 or a custom build? Make sure your power supply grounds are all tied together. If your gnd reference for +5 and +12 are not the same then the sid and core wont be able to communicate properly.

[ytsestef]

I'm building a custom build with an integrated keyboard. The PSU I'm using (Meanwell RPT-60B) already has common ground for both rails.

Also I can't at the moment monitor the MIDI output of the controller (lots of unsoldering / re-soldering to do) BUT I tried not even connecting the MIDI output of the controller to the MIDI input of the Core Module and the problem (4) where the sound stops as soon as I plug or unplug anything else in the same power socket (for example, my soldering iron). Let me strain that the SIDs don't actually stop, they actually go mute. If I do a Patch Init sound comes back on. So, even if the controller has issues, this symptom has nothing to do with it.

Side note: I haven't built the optimized PSU circuit that TK recommends, mostly because I am not using the C64 power supply, therefore I don't need the +14VDC line etc. Do I still have to put the extra caps at the end of the SID chain (actually I have only one core)? Do I still have to power the SID Module directly from the PSU and not from Core's J10?

[ascarrar]

The power supply you are using will work fine since it provides the necessary 5v and 12vdc. If you are using 6582 chips you will need to install the regulator to make 9VDC out of the 12VDC from your power supply. Either way you will also need decently sized capacitors on both rails to filter out voltage spikes and glitches, without these your circuit will be unstable.

[NorthernLightX]

Can you attach a few pictures of the modules / pcbs / prototype board you are using, and list any components you deliberately left out? The Mean Well PSU delivers a good, stable voltage, but maybe your electrical installation at home is a bit quirky. Maybe you can pick up your set-up and try it at a friends house?

One other thing that comes to mind: MIDI grounding. MIDI IN should not be grounded, where MIDI OUT should, even though MIDI IN has a 3 pin connector.

[ytsestef]

I solved it! I'm so happy, after weeks of headaches!

The Keyboard Controller had a midi volume control pot (actually slider) and I had it thrown away. But I forgot to tie the input to the ground. As soon as I put the slider back in and turned it all the way up, the problem disappeared.

:D

Thanks everyone for your help!