Bucha 275r - noise culprit found?

Discussing some incredible modules that don't quite fit into the other forum categories.

Moderators: Joe., lisa, luketeaford, Kent

Post Reply
levelhead3
Wiggling with Experience
Posts: 263
Joined: Mon Jun 11, 2018 8:53 pm
Location: Olympia

Bucha 275r - noise culprit found?

Post by levelhead3 » Mon Mar 09, 2020 12:36 pm

So amongst other problems I know folks have battled with noise and apparent grounding issues on the Roman 275's.

While working on an adapter PCB to allow currently available RCA jacks to work with the black PCB version (which has a pin-out for some completely obscure and apparently unobtainable jack) Bill Lynes noted something and mentioned it in passing which made zero sense to me. It also raised all sorts of alarm bells. I promptly pulled my green PCB version and discovered the same thing.

So while the "from tank" jacks are connected to quiet ground as one would expect, the "to tank" jacks are connected to noisy ground! With the specified tanks having input and output grounds connected, it means that this module directly connects NGND to QGND through the reverb tanks. And if I understand the 275 circuit correctly, the tank returns are then directly followed by an amplification stage. Seems to me like this would be a great recipe for introducing noise...

Can anyone think of a rational reason why the audio signal being sent to the reverb tanks would be connected to NGND at all?

User avatar
Peake
I'm in ur DIY. Filling cases with Buchla
Posts: 6780
Joined: Fri Nov 07, 2008 10:00 pm
Location: Loss Angeles

Re: Bucha 275r - noise culprit found?

Post by Peake » Mon Mar 09, 2020 12:42 pm

I traced out most of the 275's parts PCB...the 2164 IIRC are grounded directly to the tank ground, didn't check if that was connected to either of the regular ground planes...I've had people with un-modded builds I've done find no noise when it's in their system while on my vintage 200 Series power supply, 275 builds are noisy. I found if the return amplifier is removed from the PCB and placed inside the tank on a perf board, remotely powered, it's quiet. This of course requires more wires running from the case to the tanks. Sorry you're having problems.
This is not the place I'd imagined it to be.

levelhead3
Wiggling with Experience
Posts: 263
Joined: Mon Jun 11, 2018 8:53 pm
Location: Olympia

Re: Bucha 275r - noise culprit found?

Post by levelhead3 » Mon Mar 09, 2020 12:59 pm

No more problems than you and I originally ran into - I remember how much fun and games you had with this particular module. Bill and I were just appeasing my OCD and obstinate nature by finally resolving the issue surrounding the unobtainable panel-mounted RCA jacks when we stumbled across this.

I know my green PCB version introduces all sorts of bleed and grunge, a trait I've always chalked up to the very same sentiment - "it's just a noisy module". Certainly not noisy enough to keep me from using it, but it is notable. And I've always wondered why it was so much more moisy than "simple" reverb circuits like my old Doepfer A-199's.

I suspect the reason some folks have more issues than other is entirely dependent on the other modules in their system - and what else is being fed to NGND. You and I had explored the possibility of the remote mounted return amplifier a la original Buchla and Verbos, but what if the problem is as simple as the fact that the existing circuitry is just connecting NGND to QGND?

We're certainly going to test this theory by modding one of the new breakout PCB to re-route the send ground to QGND, but before we proceeded I wanted to check here and see if anyone could think of a reason why the sends might have been connected to NGND to begin with. It makes no sense to me at all.

User avatar
Peake
I'm in ur DIY. Filling cases with Buchla
Posts: 6780
Joined: Fri Nov 07, 2008 10:00 pm
Location: Loss Angeles

Re: Bucha 275r - noise culprit found?

Post by Peake » Mon Mar 09, 2020 1:15 pm

levelhead3 wrote:
Mon Mar 09, 2020 12:59 pm
what if the problem is as simple as the fact that the existing circuitry is just connecting NGND to QGND?

We're certainly going to test this theory by modding one of the new breakout PCB to re-route the send ground to QGND, but before we proceeded I wanted to check here and see if anyone could think of a reason why the sends might have been connected to NGND to begin with. It makes no sense to me at all.
Try it and see! Theory: should be as simple as disconnecting the ground wire from the reverb input tanks and connecting them to the output?

Putting the reverb return circuit on a daughterboard inside the tank to kill the noise maintains the same ground connections...
This is not the place I'd imagined it to be.

levelhead3
Wiggling with Experience
Posts: 263
Joined: Mon Jun 11, 2018 8:53 pm
Location: Olympia

Re: Bucha 275r - noise culprit found?

Post by levelhead3 » Mon Mar 09, 2020 1:41 pm

Our first test is actually going to be simply disconnecting the send jack's ground connection from NGND and re-connecting it to QGND. Nice thing about a two channel module - we can mod one side and them compare the two channels to see what's changed, if anything.

But that still doesn't answer the question - why was it connected to NGND in the first place? That makes so little sense I can't help but to wonder if there was a reason we're just not seeing.

User avatar
dougcl
Number 6
Posts: 4040
Joined: Sat Jul 26, 2008 3:04 am
Location: Portland OR

Re: Bucha 275r - noise culprit found?

Post by dougcl » Mon Mar 09, 2020 2:12 pm

levelhead3 wrote:
Mon Mar 09, 2020 12:59 pm
I suspect the reason some folks have more issues than other is entirely dependent on the other modules in their system - and what else is being fed to NGND.
The 266e produces a lot of noise on Q GND and the 275r certainly amplifies it. Removing the 266e reduces the noise.

I am not familiar with the 275r circuit, but it seems weird that it needs N GND. If there is no CPU in the circuit (why would there be?) perhaps the N GND wire can be clipped at the pigtail and the grounds connected where the pigtail connects to the board.

levelhead3
Wiggling with Experience
Posts: 263
Joined: Mon Jun 11, 2018 8:53 pm
Location: Olympia

Re: Bucha 275r - noise culprit found?

Post by levelhead3 » Mon Mar 09, 2020 2:25 pm

dougcl wrote:
Mon Mar 09, 2020 2:12 pm
levelhead3 wrote:
Mon Mar 09, 2020 12:59 pm
I suspect the reason some folks have more issues than other is entirely dependent on the other modules in their system - and what else is being fed to NGND.
The 266e produces a lot of noise on Q GND and the 275r certainly amplifies it. Removing the 266e reduces the noise.

I am not familiar with the 275r circuit, but it seems weird that it needs N GND. If there is no CPU in the circuit (why would there be?) perhaps the N GND wire can be clipped at the pigtail and the grounds connected where the pigtail connects to the board.
We were wondering the same thing. But as you can see it uses both 5v and 12v, for what I don't know. Also weird, and hard to see because the cable is partially blocking it, but there is also a jumper that says "Q/N" - for connecting QGND and NGND on purpose??

And even if you've got IC's or something that does need NGND, why on earth would you tie an audio ground to it? Especially one that's part of an in-line audio chain that is also connected downstream to QGND. It really doesn't make sense to me.

Image
You do not have the required permissions to view the files attached to this post.

User avatar
dougcl
Number 6
Posts: 4040
Joined: Sat Jul 26, 2008 3:04 am
Location: Portland OR

Re: Bucha 275r - noise culprit found?

Post by dougcl » Mon Mar 09, 2020 2:34 pm

Maintaining separate grounds in the layout is hard, and many designers either connect them at at least one point or provide jumpers for it. Either way it's a mistake and probably an attempt to solve noise issues caused by the separate grounds.

In any case, eliminating N GND won't solve the problem because the 266e puts noise on Q GND.

levelhead3
Wiggling with Experience
Posts: 263
Joined: Mon Jun 11, 2018 8:53 pm
Location: Olympia

Re: Bucha 275r - noise culprit found?

Post by levelhead3 » Mon Mar 09, 2020 3:26 pm

e is for evil? :hide:

User avatar
Peake
I'm in ur DIY. Filling cases with Buchla
Posts: 6780
Joined: Fri Nov 07, 2008 10:00 pm
Location: Loss Angeles

Re: Bucha 275r - noise culprit found?

Post by Peake » Mon Mar 09, 2020 6:23 pm

There is indeed a header on the 275 to tie together Q and N grounds, not sure if that will have the same result as removing the reverb tanks from N ground and tying them to Q. If anyone doesn't yet know the required devices for the jump are 2.54mm hard drive jumper shunts, easily found on ebay etc. Not specified in the BOMs.

Image stolen from ebay
Screenshot_2020-03-09_16-22-59.png
You do not have the required permissions to view the files attached to this post.
This is not the place I'd imagined it to be.

levelhead3
Wiggling with Experience
Posts: 263
Joined: Mon Jun 11, 2018 8:53 pm
Location: Olympia

Re: Bucha 275r - noise culprit found?

Post by levelhead3 » Mon Mar 09, 2020 10:52 pm

still raises the same question - under what circumstances would you ever want to? understand that eventually they tie back together, but my understanding was the goal was to do this as close to the power supply as possible. why would you want to do it on a module?

User avatar
cygmu
Super Deluxe Wiggler
Posts: 1480
Joined: Mon Feb 15, 2016 3:33 pm
Location: UK

Re: Bucha 275r - noise culprit found?

Post by cygmu » Tue Mar 10, 2020 1:25 am

levelhead3 wrote:
Mon Mar 09, 2020 10:52 pm
still raises the same question - under what circumstances would you ever want to? understand that eventually they tie back together, but my understanding was the goal was to do this as close to the power supply as possible. why would you want to do it on a module?
Here is an excerpt from the schematic of the 208 card 12, the reverb driver. You can see that the signal from the driver is returned to noisy ground, and the output from the tank itself is returned to Q (imagine that the little arrowheads are connected to the corresponding arrow-tails)
208reverb.png
I don't know why the reverb driver would be like that but it is a fair bet that Roman's design has just copied this.

Buchla himself found that it caused noise issues: there is a note on the schematic that reads "shorted Q and N on this board to eliminate pulse feedthrough (especially with gate on pulser) osc noise @ high freq., osc feed thru. Done from 4th on, no deleterious effects yet noted."

So he gave up on separate Q and N for this module, to reduce noise on the reverb. What this does not explain is why the N was used in the first place. Someone who knows something about reverb drivers may know, but naively it does look like a recipe for noise. It also seems very unlikely that the tank itself was connecting the two grounds, as is happening in the case we're discussing, because then there would be no need to short them.
You do not have the required permissions to view the files attached to this post.

User avatar
Peake
I'm in ur DIY. Filling cases with Buchla
Posts: 6780
Joined: Fri Nov 07, 2008 10:00 pm
Location: Loss Angeles

Re: Bucha 275r - noise culprit found?

Post by Peake » Tue Mar 10, 2020 1:03 pm

Dave Brown came up with a mod for the Roman 208 where the tank output leads are wired directly to the return amp input pin and all traces beyond that are cut...doesn't change the grounding of course but in the 275 the cable to the tank could act like an antennae and there's far more of it than the traces on the 208.

He's apparently removed his 208r V2.0 mod from his page, and says to swap card 12 for the Boops version. Unfortunately this means the mod schemo is now gone and I can't confirm if the ground for the tank return went to N or Q.
This is not the place I'd imagined it to be.

jimfowler
Veteran Wiggler
Posts: 547
Joined: Mon Feb 22, 2016 1:13 pm
Location: New York
Contact:

Re: Bucha 275r - noise culprit found?

Post by jimfowler » Tue Mar 10, 2020 8:50 pm

Is this an issue with the green pcbs only? Not sure I have any noticeable extraneous noise on my black pcb 275.

levelhead3
Wiggling with Experience
Posts: 263
Joined: Mon Jun 11, 2018 8:53 pm
Location: Olympia

Re: Bucha 275r - noise culprit found?

Post by levelhead3 » Wed Mar 11, 2020 2:38 am

Nope, the issue is present on both the the green and black PCB.

Interesting on the 208 schematic (and good call - I checked for the 275, but didn't think about the 208). I still can't for the life of me see why you would want to connect an audio signal ground to NGND at all - seems to defeat the whole purpose of having the two grounds. As does connecting them out at the module level - either through a reverb tank or jumper on the module itself. Seems like that would end up contaminating QGND - and not just for this module.

I'm about to pull all my modules to replace the distro PCB, I'll do some testing to see how the change to QGND works and sounds, both by itself and in conjunction with other modules using NGND.

tarandfeathers
Wiggling with Experience
Posts: 320
Joined: Wed May 07, 2014 3:02 pm
Location: Leeds, UK
Contact:

Re: Bucha 275r - noise culprit found?

Post by tarandfeathers » Wed Mar 11, 2020 7:53 am

levelhead3 wrote:
Wed Mar 11, 2020 2:38 am
I still can't for the life of me see why you would want to connect an audio signal ground to NGND at all - seems to defeat the whole purpose of having the two grounds.
The reverb drive is not an audio signal, it's a (comparatively) high current motor drive signal that might potentially be dumping a large and varying amount of current into 0V. That's not really something you want in your quiet 0V, because it will force significant variations in the voltage drop across the 0V traces, which will be audible in other places as bleed/CV shift/other unwanted artefacts.

I suspect the reason that Roman put Q/N jumpers on modules is to allow connection to a distribution system that does not have separate grounds. Several of his modules only have a single 0V connection where the originals had both Q and N so keeping things separate obviously wasn't a big priority for him. This is potentially less of a problem with modern low current LEDs but I don't think he often adjusted how much current LEDs were driven with either... The connection of Q and N on card 12 of the Music Easel was a bodge to work around poor layout elsewhere in the synth, as far as I can see. Roman copied the layout exactly so the same bodge is required.
Dunnington Audio - Various Buchla accoutrements available

levelhead3
Wiggling with Experience
Posts: 263
Joined: Mon Jun 11, 2018 8:53 pm
Location: Olympia

Re: Bucha 275r - noise culprit found?

Post by levelhead3 » Wed Mar 11, 2020 11:05 am

tarandfeathers wrote:
Wed Mar 11, 2020 7:53 am
levelhead3 wrote:
Wed Mar 11, 2020 2:38 am
I still can't for the life of me see why you would want to connect an audio signal ground to NGND at all - seems to defeat the whole purpose of having the two grounds.
The reverb drive is not an audio signal, it's a (comparatively) high current motor drive signal that might potentially be dumping a large and varying amount of current into 0V. That's not really something you want in your quiet 0V, because it will force significant variations in the voltage drop across the 0V traces, which will be audible in other places as bleed/CV shift/other unwanted artefacts.
If that's the case then maybe our interpretation of the problem and our proposed solution are backwards then. Maybe we should try tying the tank return signal ground to NGND instead of QGND?

Man I so wish schematics for the 275r were available...

tarandfeathers
Wiggling with Experience
Posts: 320
Joined: Wed May 07, 2014 3:02 pm
Location: Leeds, UK
Contact:

Re: Bucha 275r - noise culprit found?

Post by tarandfeathers » Wed Mar 11, 2020 11:56 am

For clarity, the above applies to the 208 reverb drive. I don't have any knowledge of whether the 275r is implemented in a similar way, but I would guess it probably is given the presence of LM380s on the BOM.

I'd probably buy a tank with both input and output isolated, use N as the return path for the driver and Q as the return path for the pickup, and experiment with connecting the chassis of the tank to either Q or N and see which provides the best noise performance (typically it would be connected to Q I think). The BOM doesn't specify which tanks the circuit is designed around as far as I can see, which is slightly poor form...
Dunnington Audio - Various Buchla accoutrements available

Post Reply

Return to “Buchla, EMS & Serge”