Console troubleshooting conundrum

[living sounds]

I found elevated THD in a channel on my console and was able to narrow the problem down to the input section. Because of many changes to the circuit board over the years there are a lot of repaired traces etc., so I started resoldering, and also switching out parts to find the culprit. Then I accidentally left out the servos capacitor for a test run and was baffled to find the low end was still all there and the signal was amplified 1.3dB across the audio band. I re-checked all the connections around the servo and couldn't find a fault. I then tried my Spice simulation and even experimentally removed connections in Spice, but couldn't replicate the results. With the cap removed the low end should obviously be gone, the simulation reflects this accurately, of course. With more connections severed I couldn't get anywhere near the +1.3 db difference of the channel without the servo cap to the working channel.

I'm at my wits end now and would kindly ask the wizards here to give me a hint what to do. In the attached schematic C9 is the cap I removed. Thanks!

 

[JohnRoberts]

Put the cap back in and continue troubleshooting.

JR

 

[living sounds]

Hm... why?

 

[living sounds]

Guess I've checked every connection now... could one of the diodes be broken? They measure the same in circuit as on a working channel.

With the servo cap back in I've tried and changed the circuit back to its original setting - R4 and R6 to 75K and a 240ohm resistor between in+ and in-. This lowers the amplitude of the distortion pattern significantly, even with amplitude matched to a good channel by adjusting R9.

Can something usefull be infered from this regarding the location of the problem?

I've also measured DC values, they're the same as on a good channel.

 

[JohnRoberts]

Guess I've checked every connection now... could one of the diodes be broken? They measure the same in circuit as on a working channel.

Probably not...

With the servo cap back in I've tried and changed the circuit back to its original setting - R4 and R6 to 75K and a 240ohm resistor between in+ and in-. This lowers the amplitude of the distortion pattern significantly, even with amplitude matched to a good channel by adjusting R9.

Trying to follow the schematic, this looks like a pretty common topology but with a few issues.

High fixed gain? Looks like mic preamp turned up all the time

DC coupled gain leg. Probably OK if input transistors are matched for Vbe. The DC servo will adjust the current in Q1 to compensate for Vbe difference. 

Can something usefull be infered from this regarding the location of the problem?

The servo is in the audio path so could be suspect.

I've also measured DC values, they're the same as on a good channel.

What is the DC voltage at output of U3? If this voltage is more than a couple volts that could degrade headroom.

Removing the servo cap could peg the output of the servo to one rail or the other. If this actually reduces distortion that suggests the servo is not happy. I am not a big fan of inverting servos while I do not suspect the topology is your issue. 

JR

 

[living sounds]

High fixed gain? Looks like mic preamp turned up all the time

DC coupled gain leg. Probably OK if input transistors are matched for Vbe. The DC servo will adjust the current in Q1 to compensate for Vbe difference. 

This is just in my Spice simulation, the real thing has a pot there. Transistors are very closely matched (and they are OK).

What is the DC voltage at output of U3? If this voltage is more than a couple volts that could degrade headroom.

Less than 10mV.

Removing the servo cap could peg the output of the servo to one rail or the other. If this actually reduces distortion that suggests the servo is not happy. I am not a big fan of inverting servos while I do not suspect the topology is your issue. 

Removing the servo cap should massively attenuate the low end, shouldn't it? This is what happens in the simulation, but not on this channel.
The servo works fine on all the other channels, there is something wrong in this one. Lower values for R4 and R6 with compensated input amplitude results in lower distortion. I thought this might point to a more specific location for the problem.

 

[JohnRoberts]

With the servo cap removed, the servo opamp will try to force the output to be 0V over the full frequency range not just at LF.

Something seems fishy with that part of the circuit.

JR

 

[living sounds]

Something seems fishy with that part of the circuit.

Indeed. I wish I could find out what it was...

 

[JohnRoberts]

Something seems fishy with that part of the circuit.

Indeed. I wish I could find out what it was...

Well there is only one op amp section, one cap, and one resistor...  replace then one at a time, or all at once.

Double check connections... the circuit should seriously misbehave with no cap in the servo.

Think less, do more.

JR

JR

 

[living sounds]

Well there is only one op amp section, one cap, and one resistor...  replace then one at a time, or all at once.

Double check connections... the circuit should seriously misbehave with no cap in the servo.

Think less, do more.

JR

JR

Well yes, it should, but it doesn't. I've triple checked everything. Replaced the cap, resistor, op amp, even the op amp sockets... About the only thing I haven't touched are the diodes, but the voltages are fine so they should be, too...

It would be much faster to build the channel from scratch if I had the PCB...

 

[living sounds]

BTW, I wonder about the role of C8 in that circuit. C7 limits the bandwidth, but changing C8 to an arbitrary value doesn't affect the response in the simulation.

 

[JohnRoberts]

BTW, I wonder about the role of C8 in that circuit. C7 limits the bandwidth, but changing C8 to an arbitrary value doesn't affect the response in the simulation.

I am the wrong person to ask about circuit simulation. Back in the '70s I wrote my own computer program to help me design complex filters, but my early experience with circuit simulation was not worth the trouble (they were not very accurate).

Yes C8 and C7 should "both" affect frequency response. Are you talking about with the servo cap removed? That could shift more of the feedback load to C7 but should also drop final output level significantly.

You can defeat the DC servo to eliminate it as a problem by replacing c9 with a dead short. Your output DC may vary from 0V but the preamp will be working properly.

A subtle distortion could be caused by any active part.

It must be something. Divide and conquer. You can defeat the c7 feedback path by replacing R12 with a short.

JR

 

[living sounds]

I am the wrong person to ask about circuit simulation. Back in the '70s I wrote my own computer program to help me design complex filters, but my early experience with circuit simulation was not worth the trouble (they were not very accurate).

Yes C8 and C7 should "both" affect frequency response. Are you talking about with the servo cap removed? That could shift more of the feedback load to C7 but should also drop final output level significantly.

You can defeat the DC servo to eliminate it as a problem by replacing c9 with a dead short. Your output DC may vary from 0V but the preamp will be working properly.

A subtle distortion could be caused by any active part.

It must be something. Divide and conquer. You can defeat the c7 feedback path by replacing R12 with a short.

JR

Thanks John. I did more testing, cleaned up the board again, replaced all diodes - no change in regards to distortion.

But I was able to solve the problem with the servo. I'm not 100% sure but it seems there was parasitic conductance around the servo for some reason, the cleanup took care of that. The measured frequency response now corresponds to the simulation.

So far I have either replaced or removed every active component as well as every cap in that schematic, but distortion is still higher than on every other channel. Removing U1 and U2 from the schematic and making up the gain by reducing R9 I still measure the same pattern of high distortion vs. a good channel. Can something be deduced from this?

My question about C8 and C7 ist not related to this problem. The original schematic of the console (Soundcraft 200B) has C7 marked with dotted lines, and the component's place was left empty in my console.

But my simulation has an error (and the uploaded schematic shows it): C8 was only connected on one side. It does indeed limit frequency response.

 

[ricardo]

What numbers are you getting for your 'elevated THD' compared to the good channels?

Can you look at the THD residual?

 

[JohnRoberts]

If you remove u1 how does it work?

Have you replaced/checked 4403s?

Of you had conductive smutz on the PCB around the servo, how about elsewhere?

JR

 

[living sounds]

What numbers are you getting for your 'elevated THD' compared to the good channels?

Can you look at the THD residual?

Sure, here's the spectrum of the good channel.

 

[living sounds]

And this is the bad one.

 

[ricardo]

You don't show any decoupling or P48V stuff on your schematic.  Got a link to a pukka Soundcraft schematic?

At that level and that type of THD spectrum , it could be earthing and decoupling.  Have you changed the earthing system on the module?

How are you measuring THD?  What instrument?  Levels?

I take it the servo problems have now gone away and doing stuff to the servo gives JR's expected results.

 

[living sounds]

If you remove u1 how does it work?

Have you replaced/checked 4403s?

Of you had conductive smutz on the PCB around the servo, how about elsewhere?

JR

It's a dual op amp, but I put a single TL071 on a dual adapter leaving U1 free. This is interesting... turning up the volume (= reducing R9 in the schematic) turns up the distortion only. Looks pretty much like the distortion in the analyzer image.

Yes, I've replaced the 4403's. They're fine. And very well matched.

I've soaked this channel in cleaner, and after letting it dry the problem went away (THD is normal now). However, the same problem has now cropped up on another channel. I'll see if cleaning this one will also help it, but it looks pretty good....

 

[living sounds]

You don't show any decoupling or P48V stuff on your schematic.  Got a link to a pukka Soundcraft schematic?

At that level and that type of THD spectrum , it could be earthing and decoupling.  Have you changed the earthing system on the module?

How are you measuring THD?  What instrument?  Levels?

I take it the servo problems have now gone away and doing stuff to the servo gives JR's expected results.

48V is turned off. I've added my own decoupling, 100nf and 10uf from each rail to ground at the leg of each op amp. Removing that didn't change the problem. I've got a very good PSU, too. And there is only a problem on one channel, always the same (only now the channel is fine and a different one has the problem...).

The Soundcraft schematic isn't really different, but mine has some modifications.

THD is measured via a 2k sinewave generated by my Lynx Aurora (converter), the images shown have the signal  a few dbs below clipping of the console input. After the input part of the schematic there are a few more op amps and a transformer in the signal path, but any distortion they might add is dominated by the distortion of the bad channel. The signal goes back into the Aurora and a plugin displays the spectrum. Even after lowering the sinewave amplitude at the converter's output siginficantly the bad channel still produces clearly visible distortion vs. a good channel.