DIYRE G Bus 202c VCA Weird Noise. Please Help.

[Clbraddock]

I recently finished building a DIYRE G Bus compressor. Very fun build with good documentation and the compression sounds great.

For reference for my question, schematic is here: https://cdn.shopify.com/s/files/1/0698/2265/files/G_Bus_1.0_Schematic.pdf

Unfortunately, I have an issue I cannot figure out. I used 202c vcas (the build manual provides for using them and I did the resistor swaps indicated in the manual). The G Bus has a switch for compression in/out. With compression in things seem to work as they should. Per the schematic, with compression out the audio signal still passes through the audio VCA (VCA1B in the case of the 202c. VCA1A is not used in my build), but the side chain is disconnected (shorted to ground) so that no compression occurs. This also works correctly, i.e. no compression occurs when compression switch is in out position.

Now for the issue: With compression out, turning the threshold (lowering the threshold) and/or increasing the ratio cause an increase in noise/saturation. Using I sine wave to test there are additional upper partials. With threshold most of the way down and ratio on 10:1 the upper partials are very audible. I've trimmed the VCAs for the lowest distortion I can get (which actually doesnt look horrible on REW RTA - about .11 )

I've attached pics showing the upper partials with compression out. The upper partials can actually get worse then this with and ratio higher, but I dont have a pic of that at the moment.

I cannot understand why the threshold and ratio have an effect on the distortion when the sidechain is out of the circuit. I have tested to confirm that the input of the side chain connects to ground when compression is out, so that is not the issue. Is this just the nature of an SSL style circuit or 202c vcas or is there something else I can check/test that may be incorrect in my build?

 

[roofRob]

Hey! You will probably hear(d) this and I'll only be the first one to mention:
Why wouldn't you go ask for support from the manufacturer themselves in the first place?

DIYRE- as you probably know- is one hell of a cool company

I did have a support case when having built my OLA5 when it suddenly after a few weeks behaved strangely.
Peterson is an ultra nice guy, snappy responder and awesome remote debugger.
Helped me out real quick.
Awesome support, highly recommended!

No offense and best regards!

 

[scott2000]

Could also ask Peterson to post response here...

 

[Clbraddock]

I did e-mail DIYRE first and Peterson did respond quickly and even tried to recreate my problem and on his bench. He was very helpful and generous and probably the best support I’ve received from a diy company. So, in no way do I mean to disparage DIYRE they are awesome!

Their bench unit uses 2181C vca (that’s what is stock in the kit) so he couldn’t reproduce my issue exactly. Technically the 202c is a “mod” although it is a mod with instructions on the DIYRE site.

I’ve always had great help from members here and just thought someone “group think” might be able to help me solve it instead of continually bothering Peterson.

Again, however, everything in my experience with DIYRE has been absolutely top notch.

Edit: I’m not sure if I can delete the post chain, but if the mods can delete it that would be okay. Like I said I 100% didn’t mean to give a bad impression of DIYRE

 

[scott2000]

Maybe try browsing some of the gssl threads here. There's a thread dedicated to mods also which seems like the 202 could be considered as...

https://groupdiy.com/threads/gssl-add-on-help-thread.65232/

 

[Ricardus]

Maybe swap out those 202c's and swap in the ones that come with the kit to see what happens.

Peterson did a post about all of the measurements he took experimenting with every possible VCA, and they all tested the same. There should be not "magical" outcome using those VCAs you chose.

 

[Voyager10]

Does the distortion change/disappear if the filter switch is set to 'external side channel'? From the schematic, this will remove the input signal from the IC5.2 input.

If the sidechain signal is somehow leaking into the main VCA's control voltage, this will give nonlinear amplification, which may produce the results you're seeing. If so, we may be able to narrow down where this crosstalk is occurring by cutting the signal at various points along the sidechain.

 

[Clbraddock]

Does the distortion change/disappear if the filter switch is set to 'external side channel'? From the schematic, this will remove the input signal from the IC5.2 input.

If the sidechain signal is somehow leaking into the main VCA's control voltage, this will give nonlinear amplification, which may produce the results you're seeing. If so, we may be able to narrow down where this crosstalk is occurring by cutting the signal at various points along the sidechain.

Thanks!! That is a great idea! I will test that tonight when I get home from work.

In the meantime, I did find that turning the threshold (regardless of compression in or out) morphs the side chain signal that feeds back to the Ec of the Audio VCA turning it from a straight line into a saw wave read by oscilloscope at R19. I think that is causing the harmonics in the audio path. Still not sure why the control signal morph is occurring though. I e-mailed those results to Peterson and he already e-mailed back that he was going to try to recreate it on his bench unit.

 

[Voyager10]

If there's an AC signal of any significant level at R19 that will certainly cause distortion artifacts at the output.

If you've got access to a scope it's probably worth taking screenshots of the signal at all the test points ('FILTER', 'S/C', 'A/R' etc.) down the sidechain signal path and sending them to Peterson, for comparison with a known good unit.

(Wild guess: something wrong with the rectifier around IC6.1 and IC6.2? Are all those diodes the right way round?)

 

[Clbraddock]

Voyager10 you were on to something for sure! Switching to external side chain cures the problem. Step in the right direction at least

Edit: Double checked diodes and they are all facing the correct direction

Edit 2: More progress. Following the signal path from input down through side chain, BAL IN, GAIN1, GAIN 2, and FILTER all are perfect. Test point at SC is a mess though. So something between the filter test point and SC test point is the problem perhaps.

Edit 3: Signal is good all the way to the signal-in side of R47. Signal-out side of R47/signal-in side of R48 are a mess. So I think Voyager10 is correct it is the rectifier. Im not sure I fully understand how the rectifier in that part of the circuit works though (Im not used to a rectifier with ICs/JFETS)

Edit 4: Tested the nearby diodes with DMM. D6 and D10 show voltage both forward and negative bias - so normally I would think they are shorted, but it is the same for both the left channel and right channel, so seems unlikely that the exact same diodes are shorted on both sides. Also double checked all resistors in the area and all are correct resistance.

 

[Clbraddock]

Ok problem is definitely detectable in side chain rectifier. With threshold clockwise (no upper harmonics) its pure DC at R50 coming out of the precision rectifier. With threshold full counter-clockwise I'm getting .29 VAC.

Where do I go from here?

 

[Voyager10]

OK, so if I understand correctly you're seeing an AC signal on the 'L Audio CV' and 'R Audio CV' lines, even when the compressor is 'OUT' (i.e. switched out with SW1 and the U6 relay).

When it's switched out, all the following should be at 0V:
* The input to SW1_CB, the 'attack' switch
* The A/R test point
* The output of IC1.1 (i.e the R89/R88 junction)
* The output of IC1.2 (i.e. the R91/R90 junction)
* The outputs of IC1.3 and IC1.4.

(Again, if I've understood the circuit correctly...)

It's a question of tracing through the chain finding where the signal starts leaking in.

Is there any possibility of a short, a solder bridge or even just dirt around the U6 relay contacts? Some coupling between the U6.1 signals and U6.2 / R92 might explain it.

 

[Clbraddock]

Comp Out (Threshold Counterclockwise)

SW1_CB 0VAC
IC1.1 Out1 (pin1 and/or R89/99 junction) 0VAC
IC1.2 Out 2 (pin 7 and/or R91/R90 junction) 0VAC
IC1.3 Out 3 (pin 8) 0VAC
IC1.4 Out 4 (pin 14) 0VAC

With comp out there is no continuity between pins of VR2-CB (makeup gain) and either side of R92

Trying to trace from the threshold pot, I don't see any AC on either DMM or Oscilloscope on the outputs of IC7 (feeding into the control of the Audio VCA)

 

[Voyager10]

Are you still able to observe this (from an earlier post)?

In the meantime, I did find that turning the threshold (regardless of compression in or out) morphs the side chain signal that feeds back to the Ec of the Audio VCA turning it from a straight line into a saw wave read by oscilloscope at R19.

R19 is at the output of IC1.3. If the sawtooth you saw earlier has now disappeared - but the distortion is still there - we're back to square one

 

[Dreams]

Have you thoroughly cleaned the solder flux off? This is the first thing i suspect with diy builds when there's possibly weird leakage and the easiest thing to try first

 

[Dreams]

I may have missed if you said it, but is the problem the same on both channels?

 

[Clbraddock]

Are you still able to observe this (from an earlier post)?

R19 is at the output of IC1.3. If the sawtooth you saw earlier has now disappeared - but the distortion is still there - we're back to square one

Ok, Im thinking maybe my DMM isn't accurate enough to pick it up. Here is R19 on the scope:

I will retest the other points with scope instead of DMM

Edit:

Retest:

IC1.1 looks good
IC1.2, IC1.3, IC1.4 all look like pic of IC1.3 Out below
SW1_CB Looks okay

 

[Clbraddock]

I may have missed if you said it, but is the problem the same on both channels?

Yes, more or less the same on both channels

 

[Clbraddock]

Have you thoroughly cleaned the solder flux off? This is the first thing i suspect with diy builds when there's possibly weird leakage and the easiest thing to try first

I hit it 4-5 times with 99% IPA. I don't have an ultrasonic cleaner, but it's about as clean as I can get it without one.

 

[Clbraddock]

Ok, Ive got progress:

Starting at IC1.1 Out (where signal was good), Signal is good up to the left side of R91 (side that connects to IC1.2 In -) Other side of R91 (connected to IC1.2 Out) is distorted.

Swapping the IC1 LF347N makes no difference though. So its not a defective chip