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

[Clbraddock]

So is the problem some sort of capacitance/leak into IC1? Should I pull the pcb at this point and re-clean that area and check for solder bridges or is there other testing I should do? It will be a bit of the pain to pull the board out, so Id like to do as much testing as needed/that I can before doing so. Thanks!!

 

[Voyager10]

OK, those traces are very useful.

So - I'm suspicious that R92 has one end left floating when the compressor is out, and this might be picking up noise and injecting it into the IC1.2 input. (This has quite high impedances everywhere so it's totally possible).

I'm wondering whether this is what R90 and CB18 are there to do - filter out high-frequency noise.

My next steps would be:
1) Check whether R90 and CB18 are the right values, and are soldered in properly. (If they are, could you capture a scope trace from each end of R90, in the 'Threshold CC' state? I'm wondering why they aren't having more effect).

2) Is it possible to locate the "left-hand" end of R92 (that connects to U6.2), and ground it temporarily (with a short bit of wire) while looking at the scope on R19? If the "R92 floating" theory is correct, grounding it would make the noise disappear.


(Thanks for your co-operation with this - I'm finding it quite an intriguing puzzle...)

 

[Dreams]

The fact that adjusting the ratio effects things is the weirdest part to me. Try removing the comp in relay, clean the pcb, top and bottom, and jump the ground as shown at U6.3, see if that has any effect. I suspect there may be more than one thing going on, since apparently switching to ext sidechain seems to cure the issue? I'd like to know which stage in the signal path the distortion is actually coming from. Probably the vca, but also could be some strange coupling somewhere else

 

[Voyager10]

What I think we know, at this point, is that signal in the side-chain (between IC5.2 and IC6.2) is breaking through into the "L AUDIO CV" VCA control voltage (as shown by the "R19" scope traces).

The threshold and ratio controls affect the level of signal in this side chain (the threshold pot adds a DC offset to the "L SC CV" and "R SC CV" signals, which will adjust the gain of VCA2, and I'm assuming the "counterclockwise" setting is maximum gain here). The ratio switch affects the gain of IC6.2 (and adds some DC offset? I've not fully figured it out yet).

I'm rather fixating on the U6 relay because it seems to be something in common between the IC5.2 - IC6.2 signal path, and the point where it looks like the break-through signal is appearing.

It might just be solder flux, but if the board has already been cleaned several times it's looking less likely.

Hopefully the things I suggested above won't need an actual removal of the circuit board, if the component side can be accessed.

 

[Clbraddock]

(Thanks for your co-operation with this - I'm finding it quite an intriguing puzzle...)

No need to thank me - You are a total life saver helping me!! I'd be lost otherwise! So thank you!!

Ok, now for your recommendations:

R90 appears to be the correct value 100 ohm (Brown, Black, Black, Black, Brown) and has continuity to IC.1.2, R14, and R15 so I think it is soldered okay (visually also looks fine).

CB18 reads .13uF on LCR. Has continuity to R90 and other side to ground.

R90 has the noise/issue on both sides. See attached. (Both pics are threshold CC)

Will work on trying to ground R92 like you suggest next.

 

[Clbraddock]

EDITED

I tried grounding switch side of R92. It helps, but does not cure the problem. It is hard to tell from the pics but the waveform gets a little smaller and the harmonics in pro tools also get a little smaller.

----

Ok, here's where the story gets weirder (or maybe just further beyond my understanding):

In the process I noticed on scope that there was the weird wave form on switch side of R92 (I must have missed that earlier), but on IC1.2 input (-) side of R92 there was no wave form.

So, I figured maybe the in/out switch was bad all along. To find out for sure, I lifted the switch side of R92 (so switch is out of the circuit completely at that point). However, I still see the noise in pro tools from adjusting threshold (no change by lifting R92).

Next I did a scope reading of the pad for R92 switch side (with resistor lifted). Distorted wave form is present. See pic. Now for the super weird part that I really don't understand: The lifted leg of R92, which is not touching anything) still has a very small distortion marching by. This too goes away when turning down threshold. I attached a pic, but it is easier to see on the live scope as it marches by. With leg lifted IC side of R92 still has no distortion (other than slightly increased white noise - although I guess its also possible the tiny marching signal is hidden under the white noise - all 3 pics are same "zoom" amount)

 

[Clbraddock]

Another piece of information - not sure if relevant though. Pulling Con4 serial cable cures the issue (but can't compress then)

With con4 pulled signal still seems to go through VCA1B because I still get the little bit of second harmonic distortion (as I potentially would expect from the 202c in this circuit). Whereas (with con4 connected), if I engage the mix control and set it to 100% dry there is no distortion at all.

Im not sure what all con4 connects though. It goes between main pcb and the control board with filter, mix, and drive.

Could also be a red herring. Im not sure.

Edit: I'm guessing it breaks the connection at the filter switch, similar to engaging external side chain which we also know temporarily cures the issue. So maybe this doesn't tell us anything new.

 

[Dreams]

So, I figured maybe the in/out switch was bad all along. To find out for sure, I lifted the switch side of R92 (so switch is out of the circuit completely at that point). However, I still see the noise in pro tools from adjusting threshold (no change by lifting R92).

There are two parts to that relay, lifting R92 is only taking it halfway out. My suspicion is that anything happening at R92 is more of a symptom than a cause

 

[Dreams]

Could also be a red herring. Im not sure.

Edit: I'm guessing it breaks the connection at the filter switch, similar to engaging external side chain which we also know temporarily cures the issue. So maybe this doesn't tell us anything new.

Yes this breaks a ton of connections, nothing really to be gleaned from it

 

[Dreams]

Since switching to ext sidechain seems to solve it, I'd suggest you probe the filter test point and switch through the sidechain settings and see what changes, might be worth looking at the distortions each setting generates also

 

[Voyager10]

I've just seen (from the build guide at G Bus Assembly Guide |DIYRE Manuals) that there are a whole load of ribbon-cable connections between the main board and the front panel (not all on one PCB as I'd just assumed).

This expands the possibilities for mayhem quite a bit... I don't suppose you could ask Peterson at DIYRE nicely for a list of what signals go over con2 - con4?

 

[Clbraddock]

Since switching to ext sidechain seems to solve it, I'd suggest you probe the filter test point and switch through the sidechain settings and see what changes, might be worth looking at the distortions each setting generates also

I tested the filter test point under same test conditions as the other tests (1 kHz sine -20 dBfs). Looks like a perfect sine wave at the test point even with threshold counter clockwise.

Changing filter setting makes no real difference on the scope except maybe changing amplitude a couple millivolts. Settings for off, 60, 80, 120 are all right around 3.44 volts peak to peak. The curved line setting is 3.88 volts peak to peak. Ext Side chain is no signal at test point.

 

[Clbraddock]

I've just seen (from the build guide at G Bus Assembly Guide |DIYRE Manuals) that there are a whole load of ribbon-cable connections between the main board and the front panel (not all on one PCB as I'd just assumed).

This expands the possibilities for mayhem quite a bit... I don't suppose you could ask Peterson at DIYRE nicely for a list of what signals go over con2 - con4?

I will ask and see what he says

 

[Meathands]

Hi all! I didn't realize we had parallel email/forum threads going until yesterday. I'll just move my responses here.

At this point, it seems like we're chasing our tails a bit looking for an issue that I can't recreate with our stock units. So the elephant in the room is the 202C VCAs. Chris, how do you feel about reversing the mod for now and using the 2181Cs? This way we can at least make apples-to-apples comparisons of measurements.

 

[Clbraddock]

Hi all! I didn't realize we had parallel email/forum threads going until yesterday. I'll just move my responses here.

At this point, it seems like we're chasing our tails a bit looking for an issue that I can't recreate with our stock units. So the elephant in the room is the 202C VCAs. Chris, how do you feel about reversing the mod for now and using the 2181Cs? This way we can at least make apples-to-apples comparisons of measurements.

Sure, at this point that seems reasonable since there doesn't seem to be an obvious/easy answer as it is now. Might take me a few days to disassemble a swap the relevant parts, but I will report back once that is done.

Thanks for all your help!!

 

[Meathands]

Sounds good! BTW in response to your email question:

Lastly, I notice on the website description its mentioned that the g bus LF351 in the side chain instead of TL074. Mine looks to have LF347N ics instead. Was this changed between the bench model and second run of production models? Could it partially explain a discrepancy between the bench unit and mine?

The LF347 is the quad version of the LF351, so no discrepancy there.

 

[Dreams]

I've been able to sort of recreate what i think is happening here, but I'm still pulling on threads, so nothing really to report yet. Stock VCAs, I'm betting those aren't the issue here, at least not entirely. I'm still thinking there are several things going on, and that the ratio changes the behavior despite being "disconnected" is the most important clue. The distortions appear as soon as the rectifier in the sidechain starts conducting, so I'm focusing my efforts around IC6, i think everything else is a red herring, a symptom not a cause, or at least a different problem to be sorted later

 

[Script]

If I had to guess, I'd say the noise comes from the 202c itself. How does a 218x perform, as suggested above?. Looking forward to problem solved

 

[Clbraddock]

I tried really really really cleaning the main pcb with ipa again. Unfortunately made no difference.

Digging through ancient Gssl threads I did find some possible clues though.

https://groupdiy.com/threads/fault-in-ssl-compressor-design.10812/#post-127729
https://groupdiy.com/threads/improve-your-gssl-for-free.34219/
Sounds like the exact problem I have. If I understand those threads correctly, it might be the floating grounds in bypass. I don’t think I can fix it so simply by rewiring the bypass switch since the g bus uses relays though. It’s late here though so I will try to look at this again with fresh eyes later.

Edit: Before resorting to dumping the 202c vcas I might try grounding R92 again like Voyager had mentioned at one point.

 

[Voyager10]

This looks a lot like the area we were poking about in earlier

If grounding the 'loose' end of R92 makes a difference when 'out', a simple permanent fix would be a decoupling capacitor (say 100n - 1uF) between the R92 / U6.2 junction and ground. VR2's wiper only has DC on it, so a capacitor to ground would be harmless to circuit operation, in or out, and it saves having to cut any PCB tracks.

i.e. this: