Gibson GA50 Clone Noise Issue

[Phil_S]

Looks like there is a issue in the area of V2a.

Sure, let's look at this. Unhooking input to C12 kills the noise. This tells us noise is introduces before this point. Work backwards. What's there? Only one thing at a time and reconnect if no change:
Tube -- try another 6SN7. Don't have another 6SN7? Try reversing wiring for the two halves and see if the problem moves to the other triode. This involves swapping pins 1-2 and 4-5. Pins 3 and 6 are both fed from R7, so those can be left alone.
Unhook the output side of C11, pin 1 of V2a.
Unhook the connection to the wiper of RP2.
Unhook input side of R5/C10.
Continue to work backwards. When the problem stops, you have isolated the problem.
Maybe this will work, maybe it won't. However, I continue to have concerns about poor lead dress -- wires too long and/or passing too close to B+ rail.

[Duck Dodgers]

Hey gang...

Yes, phase inverter abbreviated is "PI".

Thanks. For some reason i read it as PL (lower case).

didit suggested unfastening the pots. No change. However, when I touched RP2 the buss increased. With RP2 pushed against the chassis the buzz went back to "normal". RP1&3 didn't do this. Pic below.

Phil_S: Did try another 6SN7. I have new JJ and Tungsol tubes. No difference audibly but the wave forms are slightly different. I think the Tungsol is better balanced.

Disconnected the output end of C11. Buzz got louder. A lot. Same thing disconnecting RP2 wiper, louder.
Disconnecting C10/R5, no change. (I pulled the wire from RP1)

This afternoon I will poke around in the vicinity of RP2 with the scope.

[Stevem]

Is pin 1 on your 6sj7 tied to the it’s metal case?
I wonder if grounding that might help?

[Phil_S]

Too much stuff flying around to know definitively the result of what you do. Where, exactly, is the wire...makes a difference. Put everything in good order, or the best you can manage. Then, start trouble shooting, one thing at a time.

[didit]

Hello --

Does now seem certain the noise gets in at V2a grid. Keep at it.

Once through again noticing that L1 is not the 159P in my build, but the significantly smaller 159M. Given the symptoms wonder if this "under-engineer" choice is contributing? Selecting inductors -- chokes, transformers, etc -- is less straightforward than some of the other passives. The size of 159P, in mass and physical dimension, with close similarity in DCR were my determining inputs. A few available photos of GA50 construction provided estimated dimensions. I did crudely model the power supply with http://www.duncanamps.com/psud2/. Recall determining that bogey maximum current should be greater than that schematic notated "100mA" based on data pulled from a few different 6L6 equivalent sheets, given the design target plate and bias voltages. See for example http://frank.yueksel.org/sheets/127/5/5881.pdf. It doesn't match "100% class A" but is designed to be deeper than our typical "modern AB". I did also opt for silicon rectification, so there will be a bit of variance there. No easy answers.

At this point hard to advise. An earlier suggestion to eliminate L1 is perhaps a way to confirm, and my recommendation if you want to try is doing that substituting a 10W 200-250 Ohm power resistor. Essentially replicate your power stage cathode resistor but here.


Definitely ground pin 1 on V1.

Best ..

[Duck Dodgers]

Hey gang....

Is pin 1 on your 6sj7 tied to the it’s metal case?
I wonder if grounding that might help?

I knew better but let it slide. Now grounded but no improvement.

Once through again noticing that L1 is not the 159P in my build, but the significantly smaller 159M.

I had both the 159P and 159M. I grabbed the 159M and went that way. (Brain glitch.) I replaced the M with the P. Probably more robust but, no noise improvement. When doing this I noticed that the L1&2 leads were close to PR2&3. Moved them out of the way as much as possible. I know this isn't a good thing but, since there was no improvement when I unfastened all the pots and put them "over the side" I don't think it's the issue.

Put the OT back in place, as-well-as the dangling Pots. No more wires hanging over the edge of the chassis.

Got pokey with the scope.
Working back from V2 P1; 120Hz on junction of R5, R6, C11 also R5, C8 junction. Nothing on other side of C8 or R6.
120Hz on RP2 wiper, cw side and ccw side. Since the ccw side is ground I started checking the ground buss.
On the components what I'm seeing is pretty much represented by the scope trace shots I've posted.
Similar trace on the ground buss but with out the spikes. Not a classic sine wave but, maybe, 120Hz superimposed on 60Hz? A little less than 5mv.

I have no idea what a ground buss signal should look like.

I thought about building the circuit around RP2 on a piece of perf board and splicing it in place of the existing parts. What do you all think?

Thanks again for hanging in there.

[Phil_S]

Junction R5-R6-C11 and RP2...these are connected to the white wires that fly over the B+ rail. I don't know much, but I'm doubtful creating a parts substitution board is worth the effort. I'm sticking to my notion that you have a lead dress problem. I would relocate those white wires to run under the tag board. Cut new wire. In this case, you'll need the wires to be longer. Run them straight across, not at 45 degrees. Tuck them into the chassis fold and right angle bend them to connect to the pots. This will keep them away from the + terminal of the two filter caps near the pots.

[Stevem]

Here's another thing to try if you the two caps needed.

Place a .22uf and a .022uf cap in parallel then solder that assembly across / in parallel with the power supply filter that feeds the section of the amp where the issue first starts.

Be sure that the two added caps have a high enough voltage rating.

[Duck Dodgers]

Junction R5-R6-C11 and RP2...these are connected to the white wires that fly over the B+ rail... I would relocate those white wires to run under the tag board.

Tried going under the board but was making a big, uncontrollable rats nest. Decided to go over the top. I think the leads are actually further from the B+ and caps this way. Didn't seem to help. If I can find my copper tape I may try shielding them.
Pic below.

Here's another thing to try if you the two caps needed.
Place a .22uf and a .022uf cap in parallel then solder that assembly across / in parallel with the power supply filter that feeds the section of the amp where the issue first starts.
Be sure that the two added caps have a high enough voltage rating.

Don't have a .22uf and a .022uf with high enough voltage ratings. Curious here, a .22 and a .022 across the 10uf filter cap would make 10.242uf total. Not much difference. Is there some frequency attenuation going on?

Since I only had a 22uf/450v. I tried it with clip leads across C4 = 32uf. Wowsers! Knocked the 120Hz way down. I can still find it with the scope but can't hear it at the speaker. Treble control doesn't exacerbate it anymore either. With the 120Hz knocked down the 60Hz is more noticeable but still pretty quiet.

Don't know about other possible negative effects on tone, sag etc. but, the change was amazing.

Where should I (I should say "we") go from here?

Thanks to all for your patience and help.

Cheers.

[Stevem]

It is possible for filter caps to not form up to there rated uf value because there being run to far below there rated voltage.

Also that 22/450 cap may have a 20% tolerance and only really be 17 uf.

[bepone]

then was needed only to do this on the pic, separate g2 suppy from the anodes.. even better also if C1 increase to 47uF

th.png

[Phil_S]

@Bepone. I'm surprised this wasn't noticed sooner.
I checked the original GA-50 schematic from the master service manual. It shows the CT connected to 1st filter cap and the screen supply.

[didit]

It shows the CT connected to 1st filter cap and the screen supply. This clear in all provided schematics.

Look again? There is a full PI filter before connecting primary center. Increases in filter capacitance, particularly if the parts in service are out of spec, is a good measure.

..

[Phil_S]

Yeah, I shouldn't be posting things too early in the day.

[bepone]

OP fixed the problem with increasing the cap to the CLC filter. Looking at original schematic, yes filtering is too low. All is easy to understand, and there is no problem in the amp, only power transformer need to be rotated 90 degrees.

Noise can be put to min also with separating g2 from anodes, then low CLC filtering will be enough because we will unload heavyer currents from the choke and g2 will have good filtering.

But still also with low filtering, 120Hz cannot be present.
This is just a indication of disbalanced tubes, not matched by current, which now i think is main problem. So orig scheme with boosted caps in CLC ike 10uF -L , 47uF , matched output tubes, and point closed.