{SOLVED} Help me solve this riddle: Op-amp circuit path and mixer

[JJanus]

Hello all, I'm writing here again in the hope some of you might help me solve a problem I have been trying to solve since 1 year and a half.
I own a Koma Field Kit, for those who do not know this device, it is a box that puts together a 4 channel mixer and several useful blocks, like a DC block (to control motors, a CV radio, interface inputs, LFO etc… (more https://koma-elektronik.com/?product=field-kit).

Attached are the full schematics of the unit.

Basically, one of my mixer channel ended up dead all of a sudden and I have been trying to repair it ever since.
Here are the schematics of the channel:

I wrote riddle in the title of this thread because this channel behavior is a big clue in what might be wrong there.
here is what is happening:

If RV-Gain2 is totally down and up to 5% of the pot resistance and if RV-Tone2 is all the way panned to the high pass filter (pin3 of RV-Tone2), I get a signal coming from Prein-2. If I turn RV-Gain2 to more thank 5% (and thus add gain with op-amp U2B) OR if I pan RV-Tone2 a little to the Low Pass Filter (Pin 1 of RV-Tone2), I get no sound and a bleed signal from the LFO block.

The guys at Koma Electronics have been very helpful in the past by helping me understand the logic of this circuit and i cannot bother them more.
I swapped a lot of components around, some SMD for thru hole (which made my board look like a mess, especially since I lifted some trace in the process, replaced by wires… ).
Here is what I have done so far:

—First i thought something happened with the op-amp itself (U2) since I can trace the signal up until pin 6 (inverting input of U2B): No improvement
—This made me think that there might be a problem with R14 which links the inverting input to the output of the op-amp, I swapped it for a new resistor: no improvements
—I then went ahead and switched C6, the cap part of the High pass filter: no improvements
—I did the same with R18 (the resistor part of the low pass filter): no improvements
—I probed everywhere on this channel strip for shorts on traces: nothing suspicious was found

I have measured between 2,5 to 5v between the Inverting input and the non inverting input of U2B (Pin 5 and 6) so it seems that the Vin+ = Vin- condition is invalid.
I also noticed that when I measured between pin 5 Of U1B (the non-inverting buffering op-amp at the beginning of the strip) and VV (virtual ground) I get 50K of resistance. I should get 100K as the other channels give me this result.

Sorry for the long post and description, I have the feeling that everything is there under my eyes and there are many clues but I could not find anything and i'm banging my head against the wall…

thanks for your time!

[magic]

Disconnect all inputs and compare DC voltages at corresponding nodes in channel 2 and channel 1. They all should be more or less at ground and equal between the two channels. I suspect that there is DC offset appearing at some point.

Check if R6 is good.

[mikerj]

Check capacitor C2, it may have become leaky or short circuit.

[JJanus]

Thanks gyus for your input.
I forgot to mention that I already changed R6 and desoldered C2 to check its capacitance (which was fine).

[StillTrying]

All the opamp inputs and outputs should be at Vv, +/- a few mV.

If you listed each Vin and Vout it might help, and might as well include the supply voltages.

[floobydust]

The entire section is DC-coupled, it could be an offset problem. Take voltmeter readings from in to out, and see what output U3B pin 7 is doing with no signal and you turning the pots.
Not sure what "w" is, assuming it's 1/2VCC and connects to many other sections like the LFO which is why crosstalk can happen. "w" needs to be very well filtered and low impedance.

The design is missing any filters for noise and interference, which exist in the real world. You don't want AM radio or RF getting in. Slight chance it's oscillating and I would try add 10pF or so across R14 which should make no difference- if the stage is not oscillating. It might be. U1B also has no filter, like a 1,000pF cap across R2 to block RF. If PreIn-2 can get zapped with ESD than the op-amp U1B might be damaged.

[JJanus]

Hey guys, thanks for your messages.
@StillTrying All opamp inputs and outputs are indeed a few Mv except the one for the channel that is dead.
I get -1,3V on pin 6 (inverting input) and 3,3V on pin 7 (output). That with RV-Gain2 halfway and RV-Tone2 also halfway and measured to virtual ground of course.
So yes, not looking very good.

@floobydust
The vv is the virtual ground in fact, in these schematics, C4 and R8 make up a passive 1st order high-pass filter biasing the input signal in reference to "virtual ground", half-way between the 9V supply and GND. This is marked as VV throughout the schematic. This is needed since the whole circuit runs with a unipolar 9V. (quote from one of the designer of the unit).
The thing is that, with and audio probe, I can trace the audio signal up to pin 6 of U2B so doesn't that mean that the previous Opamp is fine? According to the schematics, U1B buffers the input signal, protecting the downstream circuitry (Classic Opamp 'voltage follower'). Would U1B's failure influence the circuit downstream in the way it does at the moment?

[magic]

You didn't say which opamp it is

I guess U2B, in which case something is probably wrong near U1B and U1B outputs negative DC.

[floobydust]

It's easiest to just ground the multimeter (-) and measure all the op-amp voltages and VV. At rest they should all be near +4.5V

All the op-amps are DC-coupled to each other, voltage gains U1B +1, U2B -10, U3B +20. The pots do lower it. If U1B was stuck way off 4.5V then U2B or U3B would saturate.
You should not hear audio at U2B pin 6 as it is a virtual ground if U2B is doing its job.

One huge disadvantage of virtual ground VV is any one op-amp connected to it can short and upset it, such as a shorted U2B pin 5, or even other stages elsewhere.

[fzabkar]

The virtual ground circuit is on page 7 of the circuit diagrams. It's a 1:2 potential divider buffered by an op amp configured as a voltage follower.

[StillTrying]

After 20 mins studying this thread there's not enough info to solve the riddle, so I've had to give up.

[fzabkar]

After 20 mins studying this thread there's not enough info to solve the riddle, so I've had to give up.

All the OP needs to do is to provide the voltages at pins 5,6 and 7 of each of the op amps (with respect to VV). Then the solution should be easy.

[StillTrying]

All the OP needs to do is to provide the voltages at pins 5,6 and 7 of each of the op amps (with respect to VV). Then the solution should be easy.

I know, it wouldn't even matter where they where measured from as long as they were all measured from the same place and done at the same time.

[goaty]

Maby dead LM324 ? Try replacing both U1 and U2, it´s just a few cents.

Also make sure the board is clean and you have no leakage by  carbon residue or dirt.

If you measure 50k (you have a good meter do you?) from U1B-5 to vv, then there is leakage or U1B is dead or the resistor is bad (unlikely)

P.S. Sorry the german engineering isn´t the same anymore 

[JJanus]

Ok thanks for your answers. Sorry for not providing all the readings of the 3 op-amps.

Here they are, in regard to Virtual Ground (vv):

U1B
pin 5: -3,8V
pin 6: -3,8V
pin7: -3,8V

U2B
pin 5: 0V
pin 6: -1,3V
pin7: -3,8V

U3B
pin 5: 1,7V
pin 6: 150 mV
pin7: -3,3

And in regard to Ground (AGND):

U1B
pin 5: 0,7V
pin 6: 0,7V
pin7: 0,7V

U2B
pin 5: 4,6V
pin 6: 3,2V
pin7: 8V

U3B
pin 5: 6,36V
pin 6: 4,7V
pin7: 7,9V

I did not think about U1B as a problem, i was really focused on U2B as the signal seemed prefectly fine until U2B… but those readings on U1B ?!?!

[StillTrying]

That's better.

U1B
pin 5: -3,8V
pin 6: -3,8V
pin7: -3,8V
but those readings on U1B ?!?!

Are you sure the bottom of R6 100k is connected to Vv and not the -ve supply.

[JJanus]

There is continuity between the bottom of R6 and vv.
Not sure what you are reffering as -ve supply ?

[JJanus]

Also, I just wanted to mention that on U1, the other working channel of the mixer are checking out with +- 0mV to vv and 4,6V to AGND.
These makes more sense than on U2B…

[magic]

If you are absolutely sure that only R6, C2 and U1B are connected to U1B noninverting input then one of them is blown. The 0.7V kinda suggests it's U1B because that's the voltage you would see if there is a diode connected from U1B input to negative supply and only an opamp should reasonably contain diodes inside.

[StillTrying]

"Not sure what you are refering as -ve supply ?"

Whatever powers the -Ve supply of the LM324s.


If all the connections around U1 are as on the schem, and R6 is 100k-ish, it can only be C1 C2 or U1 then. (famous last words )


"the other working channel of the mixer are checking out with +- 0mV to vv and 4,6V to AGND.
These makes more sense than on U2B…"

Yes, U2B will never work right until U1B is right, because U2B needs both of its inputs to be very close to Vv.

[JJanus]

Yes, the bottom of R6 is NOT connected to the V+ input of the LM324 (U1).
As I already checked C1, I guess my next move is to change U1!

I'll do that right away and check back with you.

thanks!

[JJanus]

Success!
The culrpit was indeed U1 ! The channel is working fine again!

It is so stupid from me to have only focused on the audio path and especially on the point where it disappeared (U2).
So I guess something happened with U1, maybe something that I plugged in once? This unit is designed to take contacts mike inputs or electromagnetic coils, could one of those, while shorted, fry an op-amp? Is there an improvement I can make to protect this unit in the future?

Thanks a lot everybody, i really needed fresh input on this, sometimes you get so focused on something that you forget the basic stuff, like measuring voltage on every op-amp  !

Also, I'm curious to hear what caused this behavior, like LFo bleeding in the signal path and especially only when I increased the gain on U2 or switched the fader from the hi-pass filter to the low-pass filter?

[GeorgeOfTheJungle]

Here they are, in regard to Virtual Ground (vv):

U1B
pin 5: -3,8V
pin 6: -3,8V
pin7: -3,8V

The input is AC coupled => you should be seeing 0V @ U1b pin 5, but instead you've got there something much closer to AGND: the first thing I'd do is double check that R6 (100k) is indeed connected to VV and not to AGND: remove u1b and check that first. If it's zero with u1b removed, u1b is very likely borken.

[StillTrying]

Marvelous. Now that U1B's output is at Vv, all the other op amp outputs should stay close to Vv, even when the pots are moved to the ends.

[magic]

So I guess something happened with U1, maybe something that I plugged in once? This unit is designed to take contacts mike inputs or electromagnetic coils, could one of those, while shorted, fry an op-amp? Is there an improvement I can make to protect this unit in the future?

Not a short but rather some high voltage pulse.

ESD, unearthed equipment being plugged in using the braindamaged RCA jacks, power-on transient in upstream equipment, that kind of stuff.

Also, I'm curious to hear what caused this behavior, like LFo bleeding in the signal path and especially only when I increased the gain on U2 or switched the fader from the hi-pass filter to the low-pass filter?

The DC offset was amplified by either U2 or U3, depending on which pot you tweaked. The final output ended up saturating to the positive rail and you heard noise that exists on virtual ground. Normally it's not heard because all signals are referenced to virtual ground and they go up and down following any changes of virtual ground potential. This is impossible when the output is saturated.