6G3 Bright Channel Parasitic Oscillation

[FastNBulbous]

Hello EL34 amp experts!  I've read many posts on this forum, but this is my first post.  I've got a lingering issue with a 6G3 I recently built using Mojotone parts, schematic, and Hammond transformers.  My initial issue was distortion in the bias vary tremolo, which I sought help for on the TDPRI forums and ultimately (after a lot of looking) solved by upping the 220k resistor prior to the intensity pot to 470k, thanks to an old post on this forum https://el34world.com/Forum/index.php?topic=18541.0!  However, in the course of troubleshooting the tremolo, I became aware that the Bright channel has parasitic oscillation issues.  I've been dragging my feet out of laziness and not creating an account here, but I've researched and tried what feels like everything imaginable to troubleshoot and there seems to be a lot of expertise and love for the 6G3 here, so I thought I'd give it a shot.

Observation: There's a high-frequency oscillation present in the Bright input circuit that is audible with the Tone control above 75% and increasing with the Volume control.  It can be heard, for example, when rotating a guitar around with the tremolo active (the peaks squeak), and if a cable's plugged into the input and you touch the hot end of the cable, in addition to your standard hum there's a high-pitched howl that can be tuned with the Tone and Volume knobs.  More scientifically, with an oscilloscope across the speaker out, you can see the oscillation appear once these controls reach a certain threshold.  On advice of a TDPRI member, I used a .1 cap to bypass plate resistors in search for the offending stage--if this method is reliable, I was able to dampen the oscillation coming out of the V1 plate to the Bright Tone stack, so I believe that's where my issue is.

About the amp:  I've used the Mojotone schematic with a few alterations: hand-made turret board with ground bus, cliff jack inputs (grounded to ground bus), adjustable fixed bias pot, rectifier diodes, power tube screen and grid resistors (currently 10k, started with 1.5k), 3-way switchable negative feedback (at the back of the chassis), a 25-watt power resistor to bring the GZ34 voltages closer into the range of the original layout, and slightly higher dog house filter cap values (22/22/22/16; although I started with the spec values of 16/16/16/8 and still had the howl).

Things I've tried that haven't worked:

• Power tube grid stoppers; started with 1.5k, currently have 10k.  No change.
• Lead dress.  I've tweaked/moved just about everything I can try to no effect. With the amp live, moving wires hasn't produced any reduction of the sound/scope wave form.
• Grounding: I separated the stages to different power nodes as described in this post (it was already quiet hum-wise to begin with) http://nohair.net/6G3.html
• Shielded wires from the inputs to the tube and from the tube to the tone stack and back again. I even tried shielding the power wire from the 4th filter cap.
• Low-value caps jumping plate resistors--this did eventually kill the oscillation but also completely murdered my tone.
• Grid stoppers on 12AX7 tubes - same as above (tone RIP), except it didn't really change the oscillation.
• Subbed in different caps between the V1 plate and the tone stack, and between the Volume and Tone knobs.
• All the usual component value and solder joint/connection tests.  The OT connections are fine (they had to be reversed when initially installing NFB)

I'm beggin' ya, anybody have any ideas or experience with similar issues?  I'm just starting a Deluxe Reverb build with dual channels and am a bit worried that the same thing will happen.  The amp works fine and sounds great, but this oscillation is definitely there and it's cooking transformers when I play for a while.  It's definitely troubling that it's in the audible range, which seems to reduce the effectiveness of using low-value caps to snuff it out.  At this point I'm kind of out of ideas except for replacing the Bright channel pots or input jacks, which I don't have a lot of faith would help. 

If you want to see the original TDPRI post, here it is.  Also attaching some photos. https://www.tdpri.com/threads/6g3-bias-tremolo-distortion-problem.1036803/

[bmccowan]

I have not read the complete thread on TDPRi, but obviously you have put some time into this.
First - thanks for posting what you have tried already - it helps.
A few things I note from your post here. You do not mention swapping tubes, but I have to believe you have done so as its a first step? And what are your voltage readings - that'll help the experts (not me so much) here.
And as to B+ voltage, you can drop a lot with a different rectifier tube, more elegant than the big resister IMO.
I have built this amp with the same mods as you used - except the big dropping resister, and no issues, so the problem is findable.
Do you have a listening amp setup? since the oscillation is audible, that could help locate the issue.
When I first built mine, I tried to outsmart the schematic/layout a bit - turned out that Leo was smarter than me - go figure.

[jim]

Looks like there is a grid wire that is passing under the sensitive preamp part of the board.  I know the mojo layout also has this wire but I would pass it over top of the board closer to the PI part of the board and if it helps at all replace it with shielded coax.  I 'm not sure but it might help if you grounded your shielded runs at the pots instead of on your buss. I would have bussed the pots togetherwith the jacks and used a star ground for the rest.  You have tried just about everything else.....Jim

[FastNBulbous]

A few things I note from your post here. You do not mention swapping tubes, but I have to believe you have done so as its a first step? And what are your voltage readings - that'll help the experts (not me so much) here.
And as to B+ voltage, you can drop a lot with a different rectifier tube, more elegant than the big resister IMO.

Thanks for the response!  I did do some tube swapping but it was a few weeks ago at the beginning of the tremolo troubleshooting, so I could definitely go a little more in-depth on that.  On the B+, I had folks over at TDPI hypothesizing that the voltages were too high and that high negative bias voltage was causing me issues.  Both the tremolo issue and the oscillation were the same when my rectifier was feeding 460v to the board as it is now at 375, so I'm not really married either way (at one point I had a 5Y3 in there and actually picked up a NOS 5V4 that I haven't tried yet since the tremolo was sorted).  Because of all of the swapping and re-biasing I'll have to decide how I want stuff set up and take new voltages and share.

[FastNBulbous]

Looks like there is a grid wire that is passing under the sensitive preamp part of the board.  I know the mojo layout also has this wire but I would pass it over top of the board closer to the PI part of the board and if it helps at all replace it with shielded coax.  I 'm not sure but it might help if you grounded your shielded runs at the pots instead of on your buss. I would have bussed the pots togetherwith the jacks and used a star ground for the rest.  You have tried just about everything else.....Jim

Thanks.  To which grid wire are you referring?  The one on V1 Pin 7?  I could try flying that one but it's already RG-174 shielded so I won't be able to mold it too well...and the PI board section is pretty far away, I think the closest I could probably get is the tremolo oscillator.  Since you mention it, I see there are a couple grid wires that go under the board.  Curious to hear if you were referring to any of the others.

My most recent efforts have mostly focused on grounding but all of the different arrangements I've had seem basically the same from a performance standpoint.  If it matters, the bus is grounded under the top right turret board standoff, by the normal inputs.  I used to have everything on the board grounded there prior to swapping to the power node-based scheme I tried today.

One other thing I forgot to mention is that I've got a 3-way impedance switch for the multitap output transformer.

[Latole]

Thanks.  To which grid wire are you referring?   

V2 - 7 ?

Use a shielded cable
________________

 "  I've used the Mojotone schematic with a few alteration ; 3 way switchable negative feedback (at the back of the chassis) "

------Return to original circuit

[bmccowan]

375V is appropriate for that amp - it'll take higher V, but no reason to go higher.
Did you try swapping the power tube leads to the OT?
Also, It's hard to tell from the photos, but you are relying on the backs of some pots for a ground connection? Those joints don't look so great - worth checking?

[FastNBulbous]

V2 - 7 ?
------Return to original circuit

I tried disconnecting NFB entirely--no change to the feedback.  Haven't had a chance to rearrange/shield the V2-7 grid wire but will report back when I do.

375V is appropriate for that amp - it'll take higher V, but no reason to go higher.
Did you try swapping the power tube leads to the OT?
Also, It's hard to tell from the photos, but you are relying on the backs of some pots for a ground connection? Those joints don't look so great - worth checking?

I think I may have mentioned before, I had positive feedback when I first fired the amp up, so the OT leads have been swapped already, and that the grounding scheme has seen many iterations with no increase/reduction in hum.  The "back of the pot" grounding for a few items in the Bright stack is the most recent iteration and the joints are definitely solid!

I also just rolled a known-good 12AX7 through each slot to re-verify it's not a tube issue.

Voltages at the current setup (GZ34 rectifier with 510Ω dropping resistor):

V1:
1: 173
2: 0
3/8: 1.6
4/5: 2.8vac
6: 174
7: 0
9: 2.9vac

V2:
1: 182
2: 0
3: 1.4
4/5: 2.8vac
6: 242 (tremolo disengaged)
7: 0
8: 1.45
9: 2.9vac

V3:
1: 253
2: 13.5
3/8: 19.8
4/5: 2.8vac
6: 249
7: 13.7
9:2.9vac

V4 (6v6 1) - Note grid/screen resistors in photos
1/5:-38.4
2: 2.8vac
3: 395
4/6: 390
7: 2.9vac
8: 0

V5 (6v6 2) - Note grid/screen resistors in photos
1/5:-38.4
2: 2.8vac
3: 395
4/6: 390
7: 2.9vac
8: 0

V6 (5AR4 rectifier)
1: 16 (not connected to anything...)
2: 417
3: 0
4: 354vac
5: 0
6: 354vac
7: 0
8: 19

Pre-dropping resistor: 420
Post-dropping resistor: 400

134 between bias power resistor and diode
-39.5 at bias cap (it's biased cool)

Doghouse L to R (22uf/1kr/22uf/10kr/22uf/27kr/16uf:
400, 393, 350, 293