Ambika voice card (SMR4) muffled sound [solved]

I just finished building the third voice card for my Ambika. But something isn’t working right because the oscillator signal is being too much filtered. Everything seems to be working except that the cutoff of the filter goes too low. I tried adjusting the trimpot to see if that had some effect, and it did, but not enough. When I play the faulty card with filter cutoff turned all the way up (127) and env mod at full with full sustain it sounds the same as one of the fully functional voices with cutoff at around 45 and no env mod.

I’ve checked all the resistors and they all seem to be right (I don’t think I have swapped any around). What else should I check for? What can be causing this?

Maybe an incorrect cap value somewhere?

Thanks for such fast reply! I’ll check the capacitors! Is there a cap it’s more likely to be, that has an impact on the filtering (except for the film caps)?

Ok, so I’ve visually checked (almost) all caps and all seems to be the right value. There are 4 caps that I can’t read the labeling - the 100nF caps that sit right beside the film caps. They all have the labeling facing against the film caps so it’s impossible to read. But I know I have been very thorough and they are exactly the sames size and shape as the other 100nF caps that I soldered at the same time on this card.

I measured resistance over a lot of the resistors on the faulty card and compared the values I got with measurments from a functioning card. I found that the resistance was slightly higher on all the measures on the functioning card, f.eks. three of 10K caps sitting close to the film caps gave 3.20K on the functioning card but only 3.04K on the faulty card. Unfortunately I haven’t the slightest idea if this gives any clue to what could be wrong.

What should I check next?

Have you tried following the signal chain to see if the muffling appears at a particular filter pole? After the VCA?

I think I have to build my self an audio probe! DIY to fix DIY problems! :slight_smile:

I’ll get back with my findings. Thanks for helping out!

The tip of a jack lead is usually the signal, you can use a 3.5mm mono jack?

In the DIY probe I’ve seen you just use a regular mono cable, but you put a cap in the signal line. I’m not sure why, but I think it has to do with a risk of the signal otherwise being too hot. And you connect the shield to ground.

Can some one with knowledge explain?

The cap is to remove the DC component.

The cap only let’s AC signals pass. It prevents loading your amp or mixer with DC voltage. Though i think most are protected from that since i have accidently done this but it didn’t damage anything.
But to be sure i made this probe for future use. The 100n cap is inside the jack.

@shiftr That probe looks pro! Inspired by yours I built one very similar today. Here is a close up of the inside. I put some electrical tape over the solder joints to make sure they stay isolated from each other before I put the cover of the contact back on.

I tried probing the audio after the four poles today. Just to make sure I measured at the right spots: first pole on pin nr 8 on IC3, second pole on pin 8 on IC2, third pole on pin 9 on IC2, and fourth pole on pin 9 on IC3. Right?

I compared what I heard on the faulty card with what I heard on one of the functioning cards. If I measured at the right spots it is clear that the sound is muffled (filtered too much) already after the first pole (IC3 pin 8) on the faulty card.

Does this mean it’s likely something is wrong with/around the first pole? Or could the sound be muffled by some (wrongly placed) components even before it hits the first pole? So maybe I should measure before the first pole?

Yeah, just trace it back until you find where it muffles.

And check all the capacitor values…

I’ve measured some more and this is what I found:

The raw sound coming out of the DAC (IC6 pin 6) sounds just like it should.

If I’m reading the schematics right the osc signal then goes via the 4,7µ cap and a 10k resistor to IC3 pin 4. I measured on pin 4 and 5 on IC3. Both on the bad card and a working card.

With the filter fully open (cutoff at 127)
-Bad card, pin 4: weak highpass-filtered signal.
-Bad card, pin 5: weak unfiltered signal
-Working card, pin 4: very weak highpass-filtered signal
-Working card, pin 5: strong unfiltered signal

With filter half closed (cutoff at 60)
-Bad card, pin 4: weak highpass-filtered signal
-Bad card, pin 5: signal too weak to hear
-Working card, pin 4: (very) weak highpass-filtered signal
-Working card, pin 5: slightly weak unfiltered signal

I’m guessing this means that the “corruption” of the signal is happening right before IC3 or at IC3. But I need expert help here. Is it possible from these finding to draw any conclusions about what could be causing the “corruption”?

Should I do more measurements?

All help very much appreciated!

It is not possible to draw any conclusion from the measurements at IC3.

Compate the voltage on IC1 pin 7 for different values of the cutoff, between a working and a dead card.

Ok, so I’ve measured on IC1 pin 7 for four different values of the cutoff (127, 80, 40 and 0) on the bad card and on two working cards.

On the bad card I got a reading of -51,9mV on all cutoff settings. On the first working card I got 5,4mV on all cutoff settings. On the second working card I got 8,8mV on all settings.

I’m guessing the slightly different values on the working cards depends on the settings of the trimpots.

But what could explain the big difference and negative voltage on the bad card?

Are you sure of your measurements? Because this voltage should change when cutoff is modified… Check all resistors around IC1, and the soldering joints of the trimmer.

I realized this morning I forgot to activate the cards in the settings. That’s probably why the values didn’t change with cutoff. :slight_smile:

I’ve measured pin 7 on IC1 again.

On bad card
Cutoff: voltage
127: -102,5mV
80: -19,2mV
40: 0,04mV
0: 0,04mV

On working card
Cutoff: voltage
127: -41,2mV
80: 35,7mV
40: 53,8mV
0: 53,8mV

Can any conclusions be drawn from these values?