Marshall TSL601 conversion into JCM 800 (of sorts)

[Fiat_cc]

Hi all.
I've built the circuit attached into a Marshall TSL601 chassis. V2 onwards is basically a stock 2204 JCM800 circuit, with a switchable parallel gain stage in front, inspired by a combination of Alessandro mods and the Marshall 1974 18 watter.

On first power up, and after setting bias, voltages measure as follows:
V1
1 - 185.6
2 - 0.008
3 - 1.631
6 - 185.6
7 - 0.008
8 - 1.631

V2
1 - 287.6
2 - 0.005
3 - 2.525
6 - 340.4
7 - 0
8 - 3.87

V3
1 - 217.6
2 - 0
3 - 1.51
6 - 402
7 - 218.4
8 - 220.4

V4
1 - 260.3
2 - 30.95
3 - 49.7
6 - 240.7
7 - 31.95
8 - 49.3

V5
1 - 0
3 - 495
4 - 487
5 - -40
8 - 0


V6
1 - 0
3 - 495
4 - 487
5 - -40
8 - 0

Heaters are at 6.6VAC.

OT resistance to V5 measures 40.2 ohms, to V6 is 41.2 ohms. Voltage drop to V5 is 0.98V making 24.3mA current draw and 12.02W plate dissipation (a little cool) and voltage drop to V6 is 1.31V making 31.7mA and 15.6W (right where I want them).  I guess these tubes just aren't real well matched.

To me, preamp voltages, especially in the cathode follower tone stack driver, look a bit high.  I feel like the whole lot could come down about 40-50V.
I think I need to look at a few lead dress things too.  It's not super noisy, but it isn't super quiet either, and certainly once things start to get cranked up, there's some buzz and hiss.  I haven't had it up loud enough to determine whether I've got the OT primary in or out of phase with the NFB loop yet, but there does appear to be some form of oscillation as I push the volume harder.  Not sure though whether that might be the cathode follower trying not to loose its poor little mind running at 400V.

I've attached my layout drawing and a photo of the chassis completed, except for heater wiring from V3 back to the PT.  I ran heaters overhead Fender style. 18AWG from PT to V6, and V5, then 22AWG from there through the preamp and to the relay PSU. The layout should be accurate with component numbers.  There are some differences like I wired the relay correctly in reality (in the layout it's incorrect), and the big power filter caps are reversed in the chassis (preamp filter closest to faceplate, power tube plate filter closest to rear of chassis along with bridge rectifier).

Does anything stand out to anybody?  It's been a rush to get this done for a friend who lives in Tasmania, which is a 5 hour drive and 10 hour boat trip away, so I've not taken my time as much as I'd have liked to.  Not to say I wasn't careful, but it's been a bit of a process!

Thanks for your help.
Reuben

[tubenit]

Overall, it looks like you've done an excellent job, IMO.

IF it were mine,  I'd want another filter cap for a node E in the B+ rail.  Not a lot of filtering there for a high gain amp?  Node E could connect to node C (instead of node D) and simply use the same type 5k/3w resistor.  You could use a 20uf/450v for node E for V1.

To cut some higher end noise, you might experiment with an "enhance cap" across the plate resistor V4-1. I'd try maybe a 220p? If you still have noise, you could try a smoothing cap from plate to cathode like on the Dumble style amps  (maybe on V2)?

https://el34world.com/Forum/index.php?topic=12723.0

You could also try increasing R37 on the B+ rail to a higher value?  Maybe 18k? instead of 8.2k and see what that does to the noise. With most of my builds, I like around 160v on the plates of an 12AX7 using 100k/1.5k  plate/cathode.

With respect, Tubenit

[Fiat_cc]

Thanks Tubenit.
I'm still finding my way with higher gain amps.  Mostly worked more on blackface Fenders, 18 watt Marshall type designs etc.
If I can find space in the chassis, another filter cap for V1 is a possibility.  I think first I'll try lowering my voltages, as I'm scared that over 400V on that cathode follower is going to pop the poor little guy! I have heard of people splitting the screen cap into two nodes instead of paralleling them, but I'm pretty close to 500V there, so not sure that I'd really want to do that. It would free up a 50uF cap though.
I'm thinking you're right.  Maybe a 15-18k between Node B and C is a good bet (I'll do the measurements and calculations next time I have the chassis out), and the 220pF enhance cap, plus perhaps the Dumble style cap as well.  I noticed after putting this together that several 2204 clones have that cap as stock, so it's possibly a worthwhile mod.

What would you say are expected voltages at the cathode follower for a 2204 JCM800? I can't imagine they'd drive the plate over 300V by much.
Cheers,
Reuben

[tubenit]

I am not sure what the cathode follower voltages should be?  I can't find a Marshall schematic with voltages?

However, the 1st Marshall was a Fender 5F6A Bassman and the voltage listed is 325v.

https://el34world.com/charts/Schematics/files/Fender/Fender_bassman_5f6a.pdf

A Laney schematic that has some similarities has 260v on that cathode follower.

https://el34world.com/charts/Schematics/files/Laney/Laney_aor50.pdf

I "googled" for Marshall Plexi 50w volts and found the attached chart that may (?) have come from the Metro Amp Forum?  I don't know if the chart is accurate or not?

OK, I found something that looks like it could actually be a Marshall amp voltage chart for comparsion.

I think your amp might benefit from one more filtering stage IF you can find room for another cap?

With respect, Tubenit

[Fiat_cc]

Thanks.
Your Googling turned up better results than mine. Looks like I need to drop a fair bit of voltage to get that cathode follower in the ballpark.

[Fiat_cc]

OK, so I've finally got a chance this week to revisit this.  It seems to start with that I have a very microphonic power tube or phase inverter tube.  Both these are post master volume, and tapping on the chassis in the area creates noticeable clanging and ringing in the speaker. I think it is V6 power tube, but its hard to tell because tapping anywhere near those 3 tubes will cause noise.

So, I also want to lower the voltage and add some more filtering. Looking at 2204 schematics, they use 3 x cap cans which are 500V 50uF-50uF.  The first cap is paralleled, then choke, then single cap section, then 10k resistor, single cap section, then 10k, single cap section, 10k, single cap section. I will do the maths and figure out the voltage drops required etc to get into the ball park, but I'm curious how close people feel comfortable running a 500V cap to its 500V rating (or indeed exceeding it by 2 or 3 percent).  My line voltage here is 232V.  Australian line voltage is rated at 230V +/- 10%.  My plate voltage is 466VDC, and screen voltage is 464VDC at 232VAC, so if line voltage was 9% higher than that my screens are at 505.76.  Now I have seen line voltage in Australia at 252V.  Not often, but it happens.  Is this going to upset those cap cans if I wire this thing up like an original 2204 circuit suggests?  There really isn't room in the chassis for another cap anywhere.

[Fiat_cc]

So, I got a chance to do a little work today.
I've attached a revised schematic and layout.
Something isn't right though.  I can't figure out what.  I had limited time today, but I'll need to trace the circuit, and check for bad solder joints.
It's loud, and the guitar tone isn't bad, up to a point, but then it starts kinda falling apart.  Gets noisy, and just doesn't seem right.  At 5 on the preamp gain and 5 on the master, its monstrously loud.  I haven't even tried the boost circuit yet.

[sluckey]

I've attached a revised schematic and layout.

OK, where did you attach them? 

Something isn't right though.

Maybe NFB is wrong phase? Disconnect the NFB wire from the speaker jack and see if it gets better.

[Fiat_cc]

Good call on NFB I'll try that out.

Attachments actually attached this time.

New voltages too.
V1
1 - 163
2 - 0
3 - 1.377
6 - 163.4
7 - 0
8 - 1.377

V2
1 - 249
2 - 0005
3 - 2.083
6 - 292
7 - 0
8 - 3.279

V3
1 - 187.3
2 - 0
3 - 1.25
6 - 338.3
7 - 187.5
8 - 188.9

V4
1 - 241.9
2 - 30.2
3 - 46.2
6 - 244.7
7 - 28.5
8 - 46.1

V5
1 -
2 -
3 - 455
6 - 453
7 - -38
8 -

V6
1 -
2 -
3 - 455
6 - 453
7 - -38
8 -

Plate current is 35mA

[Fiat_cc]

Ok.
Disconnected NFB.
No change. Replaced all preamp tubes (already put a new set of power tubes into get rid of microphonics). No change. Started chopsticking around. Grid stopper resistor and wire from resistor to pin 2, and plate wire from board to pin 6/coupling cap itself for input at V2 are all incredibly microphonic. Nothing else on the board really reacts to chopstick, except the cathode resistor bypass cap on pin 3 of v2, but it’s very marginal by comparison. Could I have over heated cap whilst installing, or does this indicate something else?

[Fiat_cc]

I’m just going to have to wait until I get home. Trying to work on it on my folks kitchen table is almost impossible. Once I have my scope, function generator and dummy load, I’ll get a much better picture of what is going on.

Thinking of redesigning the layout to avoid some of the crossing of wires etc. There’s an oscillation in the tone stack which could still be NFB related, but it’s safe to say there’s more than one issue going on here.

Edit: I think I need to look at my grounding as a starting point. I tried to use a bus/star hybrid, but I think I’ve made some compromised decisions. I don’t think that’s going to solve my microphonic capacitor issues, but certainly some buzz and hum problems. Then to track down the source of oscillations. It’s frustrating as it’s a clients amp, but should be rewarding when I nail it.

[Fiat_cc]

OK, so I reworked the grounding, and it seems to have reduced the hum and noise floor, but there is this oscillation problem.  It disappears when I remove V2 (which is V1 in the original Marshal 2204).  It just so happens the the suspect capacitor is the plate coupling cap from V2B.  I've not yet had the opportunity to replace the cap, but I pulled out the relay board and bypassed V1 so I can concentrate on just getting the standard Marshall circuit working.  Once that works I can add stuff back in.  Moving OT leads around will effect th4 oscillation, and at times I manage to move those leads such that the oscillation stops, however what exactly causes it to stop (proximity to what components etc) isn't entirely clear.  I will try replacing the suspect microphonic cap and see where I go from there.  I did replace the tube in V2, and it altered some of the qualities of noise and oscillation, but it didn't stop it.

It is very definitely linked with gain though.  Master on full and gain on full, with all tone and presence controls at zero will usually set it off with no input signal at all.  Master at full, gain at half, and add just the tiniest bit of tone stack, and that will set it off too, but remove V2, and the tone controls set to full with master at full won't set it off.  I have tried with and without NFB connected, and no change.  It is a big signal too.  Driving into a 16ohm dummy load, it is reading upwards of 50V on the scope. Some times well above if I let it get out of control.  Once it starts, I can usually set the controls to let it oscillate continuously at about 35V.  Both the amp and the dummy load get quite warm after prolonged testing under those conditions, but no signs of red plating etc.

[Fiat_cc]

I replaced the suspect cap with a brand new one, and nothing changed.  It is still as microphonic as ever, and the oscillation is just as nasty as ever. I twisted the OT primaries together, and removed the speaker jacks from the chassis to allow me to move them and the OT secondaries around.  The secondaries seem to make no difference at all.  Moving the twisted primaries around will slightly vary the amplitude and frequency of the oscillation, and strangely, moving them closer to V1 seems to minimise it most. I really am starting to tear my hair out.  Everything looks like it should work, but something that I'm missing isn't right, and I just acn't figure it out!

[shooter]

Since you have the test equipment you'll find it.  1/2 split til the gremlin can't hide
I've had a couple microphonic oddities fwiw
PA tubes biased way hot
crossed up ground/signal wires
bad tube socket

can you "change" frequency by "waving" your hand closer/ further
 

[Fiat_cc]

Thanks Shooter.
I have been mucking around with it for the last hour or so.
I'm stumped. No, it doesn't respond to hand movement. Tapping the board in different places can trigger it it seems.  Mostly the coupling caps on V2.
I put in a pair of old power tubes so I don't trash the new Svetlanas I bought for it.  That has changed the nature of the oscillation somewhat.  It's not such a pure clean sine wave anymore, and I can't get it to settle down at a constant level like I could before, but it still takes off under the same circumstances. I'm sure the coupling caps are a red herring, but I can't see what else it can be.  It's really bugging me out!

[sluckey]

Pull V1. Does it stop oscillating? If not, leave V1 out and pull V2. Does it stop oscillating? If not, leave V1 and V2 out and pull V3. Does it quit oscillating? If not, leave V1, V2, and V3 out and pull V4. Does it quit oscillating?

[Fiat_cc]

Ignore V1 for now.  It is my switchable boost stage, and I've removed it from the circuit for the time being to try and get the standard 2204 circuit working.

The oscillation disappears if I remove V2 though.
When it is oscillating, I can use the treble and mid controls, and the gain control to somewhat control the frequency and volume of the signal.

One thing that has me completely dumbfounded though, is that scoping across my dummy load, I get a perfectly clean sign wave (once I find the gain and tone stack settings that let it settle and stablise), but when I probe the phase inverter input, it is slightly traingulated, and the phase inverter output/power tube grids is almost a perfect triangular wave.  I don't understand how the input to the power tubes can be triangular, but the resulting output is sinusoidal. It's a super high frequency too. peak to peak of the sine wave is roughly 7uS, which if I am calulating correctly is somewhere around 140KHz.  It varies based on control settings anywhere between 110KHz and 300KHz.  Under certain settings, you can clearly hear the output transformer whining, buzzing, pulsing etc. I really need to solve this one, but it's driving me bonkers.  I may need to go through and check every component against both my layout and schematic and the Marshall schematic (again...), and then check each solder joint for a 4th time, in case there's a mistake in there I haven't yet seen.

Edit: OK, so the oscillation disappears with V2 removed. It is still gone with the plate of V2B desoldered.  It is there however with the plate connected and the grid disconnected.  So, as far as I can tell, it has to be originating somewhere in that gain stage, but I've no idea what's causing it.  I'm about to try disconnecting the cathode, to see what happens (this is the 10k cathode resistor cold clipper stage).  So, any thoughts on what could be causing it? I've replaced the coupling cap, and I've tried different tubes in this position.

Edit again:  Disconnecting the cathode also stops the oscillation.  So by my reckoning, its got to be something to do with either the cathode resistor, or the plate load resistor.  I guess coupling cap, and the voltage divider with treble bypass cap after the coupling cap are still in the equation. I really cant see what else it could be. Short of just replacing all those components, what else should I test or look at?  I'll post a close up pic of this part of the board in a sec to see if there's a layout issue that screams out to anyone.

[PRR]

> and the grid disconnected.

An open grid in a multi-stage amp will almost always oscillate.

Ground it. (_IF_ that is the normal DC bias for this grid.)

[Fiat_cc]

Good call.  Hadn't even thought of that.

> and the grid disconnected.

An open grid in a multi-stage amp will almost always oscillate.

Ground it. (_IF_ that is the normal DC bias for this grid.)

Edit: I just thought about that a little harder. That grid is fed through the gain pot, so with the pot at zero, the grid is effectively grounded, and yes, this stops the oscillation.  The oscillation is very sensitive to volume actually.  So, does that then mean that the oscillation could actually be prior to that gain stage, given that grounding that grid prevents it?

[Fiat_cc]

So I thought I had it.  Grounding the output of the coupling cap (C4) on V2a stops it, but lifting the plate wire going to V2a doesn't.  Perfect.  It's gotta be the coupling cap.  Replaced it, and no better. I'm stumped now.

So basically, grounding the grid in V2b stops it, but I don't know why, or what it is in that area of the circuit that could be causing it.  I've played around a little with lead dress.  I've replaced C4 and C6 (coupling caps for V2a and b) and none of it has made any difference.  I've replaced the preamp tube, I've played around with placement of the speaker jacks and OT primary and secondary leads.  I'm running out of ideas.  Is it worth trying new earth positions?  Everything currently is running to a single star ground.  I could try splitting it up into two or 3 different 'sub' star grounds?  I could try removing the output transformer from the chassis altogether and seeing how far away fromV2 I can get it and its associated wiring?  I just don't know where to go from here.

[Fiat_cc]

So, the attached picture is of the scope with one probe (the sine wave trace, at 9.8V per division) on my dummy load, and the other probe (saw tooth wave trace, at around 7V per division).
How on earth am I getting such a perfectly clean sine wave at the output, with such a perfect saw tooth wave on the power tube grids.  I'm beginning to suspect something in the power supply.  Maybe a filter cap can?  I'm thinking maybe whatever the oscillation is, is getting directly into the B+ and oscillating everywhere in the amp.  That doesn't quite explain why grounding V2B grid shuts it off though.  Still, I'm clutching at straws.

[sluckey]

The sawtooth is not symmetrical. Neither is the sine wave. I suspect the inductance of the OT is rounding the peaks of the sawtooth so that it resembles a sine wave.

What is the frequency?

[Fiat_cc]

Frequency varies depending on how the tone stack and gain controls are set, but in this picture, about 140,000Hz.

It’s been anywhere from 110KHz, up to 300KHz or so.

[shooter]

couple easy tests
put a ~ .001 to .01 cap across the plate R on V2a.  try V2b also
put a really big cap on the cathode R of V2a, try v2b also  something > 100uf ~ =>25vdc

measure ripple(AC) at all your rail taps
seriously considering giving V1 it's own cathode/cap for each segment, also it's own rail tap, (E)

[PRR]

> about 140,000Hz.

  Even your dog won't notice if you cut that off.

Get a fistful of 150pFd-330pFd capacitors (Silver Mica sounds spiffy; ceramic in this size will be equally perfect) and start slagging plate nodes until the whining stops. Be generous; you don't want to just-barely stop the squeal, you want to kill it so it can never come back. A couple/few 16KHz hi-cuts is not obvious on guitar.

Triangle to sine: the core oscillator (whatever it is) is slew-limited, running constant current two ways into circuit stray capacitance. That makes a triangle. When current is not dead-equal both ways and at all points, the triangle is asymmetric and unsharp. Now put a 160KHz triangle into an output transformer that barely makes 20KHz. It gets rounded about 8:1, or more, and the ultimate low-pass rounding of any garbage is a Sine.

And fundamentally you have too much gain in too little space. I would wonder where the OT plate leads run relative to small-signal stages; can't work it out from the picture. Unless there is some obvious sneakage route, with the amp already built, and the whine way supersonic, I would just slag everything above the audio band, like this.

Ah, plate-cathode may work just as good as to-ground, and less wiring.

[Fiat_cc]

couple easy tests
put a ~ .001 to .01 cap across the plate R on V2a.  try V2b also
put a really big cap on the cathode R of V2a, try v2b also  something > 100uf ~ =>25vdc

measure ripple(AC) at all your rail taps
seriously considering giving V1 it's own cathode/cap for each segment, also it's own rail tap, (E)

Thanks Shooter.
I've tried a 0.001 across the plate resistor on V2a with a 0.022 (it was easily at hand) across V2b. It stopped the oscillation at all but the most extreme settings (it's stable with all controls on full, except presence.  It starts to oscillate once presence gets to about 8...). At those settings it starts to show an oscillation at 100Hz.  Can you tell me what exactly those capacitors are doing, so I understand the theory a bit better?  I understand PRR's suggestion of shunting those high frequencies straight to ground, but I'm unsure what bypassing the plate load resistor does.

I will see if I have a large value cap to try on the cathodes.  Will that not make the amp more unstable at potentially lower frequencies?

As far as power supply, I have split things up so I now have an E node.  It runs: Plates from A (a 2 x 50uF cap can running in parallel), then through a choke to a single 50uF cap can section at node B feeding screens, then an 8.2K 3W resistor to another single 50uF cap can section at node C feeding the V4 phase inverter, then through a 6.8K 3W resistor to a single 50uF cap can section at node D feeding V3 cathode follower, then through a 3.3K 3W resistor to the last 50uF cap can section at node E feeding V1 and V2.  Would you split this differently?
Also, note that I have V1 out of circuit currently.  I am only trouble shooting the original Marshall 2204 circuit currently.  I figure I want that stable before I even start to worry about the boost stage.

Edit:  The biggest 25V cap I had in stock was 50uF, but putting that across the V2b cathode sets it off into 100Hz oscillation at comparatively low gain settings (master volume at full, preamp gain at 8, and all other controls at 0). Certainly that is close to the most unstable it's ever been, and the oscillation is now at 100Hz, not ultrasonic frequencies.

[Fiat_cc]

> about 140,000Hz.

  Even your dog won't notice if you cut that off.

Get a fistful of 150pFd-330pFd capacitors (Silver Mica sounds spiffy; ceramic in this size will be equally perfect) and start slagging plate nodes until the whining stops. Be generous; you don't want to just-barely stop the squeal, you want to kill it so it can never come back. A couple/few 16KHz hi-cuts is not obvious on guitar.

Triangle to sine: the core oscillator (whatever it is) is slew-limited, running constant current two ways into circuit stray capacitance. That makes a triangle. When current is not dead-equal both ways and at all points, the triangle is asymmetric and unsharp. Now put a 160KHz triangle into an output transformer that barely makes 20KHz. It gets rounded about 8:1, or more, and the ultimate low-pass rounding of any garbage is a Sine.

And fundamentally you have too much gain in too little space. I would wonder where the OT plate leads run relative to small-signal stages; can't work it out from the picture. Unless there is some obvious sneakage route, with the amp already built, and the whine way supersonic, I would just slag everything above the audio band, like this.

Ah, plate-cathode may work just as good as to-ground, and less wiring.

I've attached a chassis pic so you can see my OT leads.  The primaries are the blue, green and purple twisted bundle I have running outside the top of the chassis.  The secondaries go to the speaker jacks which are in very close proximity to V2, but I have also removed those from the chassis and wired them in outside, and this didn't help either.

I want to again clarify, forget V1 for now.  This is now currently a standard Marshall 2204 circuit (with a slightly different bias circuit to make use of the bias tap on the PT I was using).  V1 is currently disconnect from the signal path, although it is still receiving plate power and heater power. I have just removed it completely from the socket, just to verify that it is not involved in some way.  And I can verify now, there is no change with it completely removed.

I have just clipped in 3 x 270pF silver mica caps I had on hand, from the plates of V2a, V2b, and V3a to ground.  It brings the frequency of the oscillation down to between 1KHz and somewhere up around 10KHz depending on settings.  The higher frequency was running away to fast for me to scope it properly before I had to shut the amp down.  The OT itself was audibly screaming and resonating.  One thing that doing this has shown though, is that as the frequency of the oscillation becomes lower, the wave form on the scope becomes more triangular.  Stil nothing like the signal we see at the phase inverter, but definitely closer to that asymmetric triangle shape.

[Fiat_cc]

I've gotta go out for a while, so experiments will need to stop for now.

I'm thinking I definitely want to remove the OT from the chassis completely, and place it as far away from the preamp section as the current wiring will allow (should be able to get at least 30+cm away), and see if that helps anything.

I'm beginning to think that the actual problem is at a much lower frequency, but somehow keeps doubling on itself as it feeds back and takes off.  When I've had it plugged into a speaker and it's started, before I've very quickly shut it down, the frequency is low, like 100Hz, but the scope was showing frequencies that are way into the ultrasonic range, so I just assumed that the speaker couldn't handle the high frequencies and was just doing whatever it could with the huge amount of high frequency signal it was receiving.

So... Do the results of either set of experiments so far shed any light on the subject?  I'll post results of my OT move experiments as soon as I have time to do it.

I've attached the latest schematic too (which should be 100% correct with what is currently in the chassis), just to help clarify things, but remember that V1 is currently completely out of the circuit, as is the relay/switching etc.  The input jack runs straight to the grid stopper of V2a.  Also note that removing all that circuitry etc had no effect on the oscillation.

Cheers for the help so far.
Reuben

[Willabe]

Do you have the preamp tube and PI tube shields on? If not try keeping them on while you test.

You have the OT and speaker jacks next to the preamp tubes, you might have to put a shield between them, like you have around the OT.

The OT secondary wires should be twisted together tightly and tucked up into the chassis corner.

I saw this; "I've attached a chassis pic so you can see my OT leads.  The primaries are the blue, green and purple twisted bundle I have running outside the top of the chassis.  The secondaries go to the speaker jacks which are in very close proximity to V2, but I have also removed those from the chassis and wired them in outside, and this didn't help either."

I'm not exactly sure where you put all of this but, I'd try this test, unbolt the speaker jacks/impedance selector/line out jack and move them and their wires out and away from the chassis (to the left of the chassis with the controls facing you) and onto the bench  -away-  from the input jack/guitar cable and with the preamp tubes shields on. Then try the amp and see if the oscillation stops.

And you should try using shielded wire from the input jack to the 1st tubes grid and on the other preamp tube grid wires and the PI input grid.

If it was V2(?) that seemed to be the problem area, then start there 1st.   

And I'd move the input jack 68K R's to the preamp socket grid pins, with as little lead as possible between the 68K and the tube grid pin. Like 1/16". Grid stoppers work best this way.

And I'd try grid stop R's, at least 10k, might have to go up as high as 100K(?) on V2B and V3A, again with as little lead as possible between the 68K and the tube grid pin. Like 1/16". I'd probably start with 100K and see if that knocks it out, then try lower and lower values until it starts to come back.

Read this on grid stoppers;

http://www.valvewizard.co.uk/gridstopper.html

(I saw that you had this, shielded wire/grid stop tight to the grid pin, on the 1st preamp tube before you disconnected it.)

And I'd try twisting the 2 power tube grid wires from the PI together tightly too. Better yet, use shielded wire. But it seems your problem area is before the PI. So start on the far input end of the chassis.

And I'd probably move the power tube grid stopper that runs over the screen grid wire and run it tight to that power tubes grid pin.

And I'd move that red B+ wire that runs the length of the board from under the components and twist it around the ground buss. And there's a 2nd red B+ wire coming off that same B+ filter cap, I can't see where it goes. Is it under the board?

Do you have the power section ground separate from the preamp ground? If not, because your amp has a -FB loop with a presence control, the PI, including the presence control should be grounded with the power tube grounds. The MV feeding the PI should be grounded with everything before it by the input jack. So B+ nodes A/B/C get grounded together and nodes D/E get grounded together.

Here's a link on grounding;

http://www.valvewizard.co.uk/Grounding.html

If it were me, I'd get rid of that 3rd 50uF/50uF can, too much B+ filtering, and put in a 20uF axel cap for B+ node D and a 20uF axel cap for B+ node E. The current draw is very small on those little preamp tubes. (Too much filtering can/will make the amp too stiff for guitar. The B+ node C doesn't need to be that high a value either, 20uF would be plenty there.)

Or you can get small radial caps that will fit in a tight space. Someone here just built an amp using just small radial caps for the B+ filtering and I like to use them, on some things. They worked fine for me. I'd put them very close to the circuitry they feed. I'd run a tightly twisted pair of wires underneath, along the back side of the board, control side, from the B+ node C to the new axel caps. (But that's a different discussion.)

Any 1 of these things might be the fix or it might take a few of them together to fix it. To tame the oscillation in this amp it might need to be addressed at both the input and the output, grid/plate, of the preamp tube(s). 

I'd try all these things with those plate bypass caps in place 1st, get rid of the large K (cathode) cap, then if the oscillation stops, maybe I'd try pulling those caps 1 at a time and see if it comes back. Just to see/hear what happens. 

[Willabe]

I think first I'll try lowering my voltages, as I'm scared that over 400V on that cathode follower is going to pop the poor little guy!

New production tubes don't seem to take the higher dcv's that the older tubes did. There have been guys here that have killed new production tubes used as CF in seconds/minutes/days.

One way to protect them is to use a dc stand off voltage on the heaters. (It can also quite an amp from tube heater to K leakage.) 

Here's a link that explains it. Scroll down to 'heater elevation.'

http://www.valvewizard.co.uk/heater.html

I like to use 70dcv to 80dcv.

[Willabe]

I don't see that anyone asked this;

Did this amp work fine before you added the boost gain stage?

Or did you build the whole amp together with the boost gain stage and it never worked from the start?

[Fiat_cc]

My responses and thoughts in green.

Do you have the preamp tube and PI tube shields on? If not try keeping them on while you test.
Yes, I have the tube shields on for all experiments.

You have the OT and speaker jacks next to the preamp tubes, you might have to put a shield between them, like you have around the OT.

I had considered this possibility, but when I moved the speaker jacks outside the chassis in an earlier experiment, it didn't help much.

The OT secondary wires should be twisted together tightly and tucked up into the chassis corner.

They were originally tucked into the chassis corner, but not twisted.  I've just twisted them now.

I saw this; "I've attached a chassis pic so you can see my OT leads.  The primaries are the blue, green and purple twisted bundle I have running outside the top of the chassis.  The secondaries go to the speaker jacks which are in very close proximity to V2, but I have also removed those from the chassis and wired them in outside, and this didn't help either."

I'm not exactly sure where you put all of this but, I'd try this test, unbolt the speaker jacks/impedance selector/line out jack and move them and their wires out and away from the chassis....

That is exactly what I did.  It made little, to know difference at all.  I've just done an experiment with unbolting the OT from the chassis, which I'll get to describing shortly.

And you should try using shielded wire from the input jack to the 1st tubes grid and on the other preamp tube grid wires and the PI input grid.

Originally I used much more shielded wire, but as I've stripped different parts of the circuit out, I replaced some parts with standard wire for ease.  FWIW the original Marshall 2204's didn't come with shielded wire in any of the gut shots I've seen. However, when it gets rebuilt for completion after solving this issue, yes, shielded wire will be used.

If it was V2(?) that seemed to be the problem area, then start there 1st.   

That's my plan

And I'd move the input jack 68K R's to the preamp socket grid pins, with as little lead as possible between the 68K and the tube grid pin. Like 1/16". Grid stoppers work best this way.


Yes, I plan to do that. I will replicate what I had done originally on V1 with the grid stopper.

And I'd try grid stop R's, at least 10k, might have to go up as high as 100K(?) on V2B and V3A, again with as little lead as possible between the 68K and the tube grid pin. Like 1/16". I'd probably start with 100K and see if that knocks it out, then try lower and lower values until it starts to come back.

Interesting.  I'll try that.

And I'd try twisting the 2 power tube grid wires from the PI together tightly too. Better yet, use shielded wire. But it seems your problem area is before the PI. So start on the far input end of the chassis.

Interesting.  Yes, I will try twisting these.

And I'd probably move the power tube grid stopper that runs over the screen grid wire and run it tight to that power tubes grid pin.

More like I've run the other one?  Can do.

[Fiat_cc]

And I'd move that red B+ wire that runs the length of the board from under the components and twist it around the ground buss. And there's a 2nd red B+ wire coming off that same B+ filter cap, I can't see where it goes. Is it under the board?

In the first version it was a single run under the board.  When I split the power filtering up for one extra node, it was impossible to lift the board and run that extra lead under there, so I ran it on top, trying to cross all signal points at 90 degrees.  I could wrap it around the ground bus, if you think that'll help.  Yes, the other red B+ is still under the board.

Do you have the power section ground separate from the preamp ground? If not, because your amp has a -FB loop with a presence control, the PI, including the presence control should be grounded with the power tube grounds. The MV feeding the PI should be grounded with everything before it by the input jack. So B+ nodes A/B/C get grounded together and nodes D/E get grounded together.

EDIT: MV should be grounded with the PI tube grounds.

Here's a link on grounding;

http://www.valvewizard.co.uk/Grounding.html

Currently, preamp and power amp have separate busses as it were, but everything in the amp all runs back to a single star ground.  There are seemingly many different opinions on grounding, and the star ground is popular in some circles.  They say basically, not to use the chassis as a return path or ground plane, merely as a reference, so that is what I attempted to do.  I have been considering grounding the preamp bus at a separate point near the input jack, and possible moving a couple of other things too (OT ground reference for example) to another 3rd point.  I have read that Valve Wizard article, but it was a while ago, and the 18 watt Marshall I was building was much less finicky.

If it were me, I'd get rid of that 3rd 50uF/50uF can, too much B+ filtering, and put in a 20uF axel cap for B+ node D and a 20uF axel cap for B+ node E. The current draw is very small on those little preamp tubes. (Too much filtering can/will make the amp too stiff for guitar. The B+ node C doesn't need to be that high a value either, 20uF would be plenty there.)

The filtering as it is, is as per 2204 schematics I found, i.e. it is an exact copy of the original Marshall circuit.  The only area where I've deviated (besides the switchable boost) is in the bias section, where I copied the bias circuit from the donor amp, to make use of the bias tap on the donor PT.

Or you can get small radial caps that will fit in a tight space. Someone here just built an amp using just small radial caps for the B+ filtering and I like to use them, on some things. They worked fine for me. I'd put them very close to the circuitry they feed. I'd run a tightly twisted pair of wires underneath, along the back side of the board, control side, from the B+ node C to the new axel caps. (But that's a different discussion.)

Any 1 of these things might be the fix or it might take a few of them together to fix it. To tame the oscillation in this amp it might need to be addressed at both the input and the output, grid/plate, of the preamp tube(s). 

I'd try all these things with those plate bypass caps in place 1st, get rid of the large K (cathode) cap, then if the oscillation stops, maybe I'd try pulling those caps 1 at a time and see if it comes back. Just to see/hear what happens. 

About to post another couple of posts with some photos, but to cut a long story short, it is more stable without caps from plate to ground.  I will try clipping them in as plate resistor bypass caps tomorrow, but so far it is most stable with the OT moved well away from the preamp section, and now extra caps clipped in.

[Fiat_cc]

I think first I'll try lowering my voltages, as I'm scared that over 400V on that cathode follower is going to pop the poor little guy!

New production tubes don't seem to take the higher dcv's that the older tubes did. There have been guys here that have killed new production tubes used as CF in seconds/minutes/days.

One way to protect them is to use a dc stand off voltage on the heaters. (It can also quite an amp from tube heater to K leakage.) 

Here's a link that explains it. Scroll down to 'heater elevation.'

http://www.valvewizard.co.uk/heater.html

I like to use 70dcv to 80dcv.

I did lower my voltages significantly, but yes, I could look into elevated heater voltage.  I'm just not sure how I'd create a 70VDC supply easily in this amp.

I don't see that anyone asked this;

Did this amp work fine before you added the boost gain stage?

Or did you build the whole amp together with the boost gain stage and it never worked from the start?

I built it from scratch with the boost stage, and it has never worked right.  It does play, and sound amazingly good at lower (but still ear splitting) volume, but it has had hum and oscillation issues from the outset.  I've always thought it had excessive gain on tap, so I may need to triple check all resistor values and grounds through the circuit are correct, but I've done that already, and I didn't find a mistake.

[Fiat_cc]

Chassis shot with the OT moved as far as I can from preamp.
And a trace with no plate caps added.

[Fiat_cc]

V2b bypassed (larger trace) and v2a and b bypassed (smaller trace.

[Fiat_cc]

V2 a and b, and V3a bypassed.

[shooter]

Shunting, snubber caps, work the same weather you send it to B+ or ground

are those "injected" signals with junk, or just amp idling no input?

[Fiat_cc]

Those traces are idle, no input with master volume at full and preamp gain at full and all tone controls st zero, except the first one, which had all knobs at full. That first one, with just hash, is about 3.5V per division. The rest are 6.5V per division from memory.

[shooter]

I have been considering grounding the preamp bus at a separate point near the input jack

that WILL help

can't find it in this thread,
did you measure each rail node with your meter VAC?  put the neg lead at the main PS ground bolt, then repeat with the neg lead somewhere on the chassis

[Willabe]

I did lower my voltages significantly, but yes, I could look into elevated heater voltage.  I'm just not sure how I'd create a 70VDC supply easily in this amp.

You just tap the B+. It's in the link.

If your heater wind has a CT you inject the dcv there. If no CT you disconnect the ground end of the 100ohm heater R's and inject the dcv there.

Get the amp stable 1st.   

[Willabe]

Originally I used much more shielded wire, but as I've stripped different parts of the circuit out, I replaced some parts with standard wire for ease.  FWIW the original Marshall 2204's didn't come with shielded wire in any of the gut shots I've seen. However, when it gets rebuilt for completion after solving this issue, yes, shielded wire will be used.

The grid wire is very sensitive to stray signals. The signal from guitar PUP's is very small and can easily be overridden with stray signals and magnetic fields.

After a plate wire goes through the coupling cap, it is now a grid wire.

The amp is unstable, I'd put shielded wire on every grid AND grid stoppers on every grid pin, tight to the pin.

And, like I wrote before, you still might need to use the plate caps along with the shielded wire and grid stops.

And I'd probably move the power tube grid stopper that runs over the screen grid wire and run it tight to that power tubes grid pin.

More like I've run the other one?  Can do.

Yes.

Read the link on grid stoppers, it explains why they need to be tight to the tube pin.

[Willabe]

And I'd try twisting the 2 power tube grid wires from the PI together tightly too. Better yet, use shielded wire. But it seems your problem area is before the PI. So start on the far input end of the chassis.

Interesting.  Yes, I will try twisting these.

The signal going out of the PI is larger than at earlier stages, sometimes just tightly twisting the 2 PI output wires going to the power tube grids is enough. If not then you have to use shielded wires for this wire run too. 

But, it still seems that the problem area is before the PI.

I think your getting stray magnetic fields injected into the early gain stages because after taking the OT and it's wiring out of the amp, it got better.

And I think it's being injected into a few spots, not just 1. 

[Willabe]

I have been considering grounding the preamp bus at a separate point near the input jack

that WILL help.

For peace of mind, I'd do this, like in the link, 1st. Just to take it off the table of adding to the problem, it is a hi gain amp, grounding is more important.

With hi gain amps, placement/layout of ALL components, PT/OT/choke/tube sockets, lead dress, shielded wire, grid stops, grounding, input/speaker jacks, switches, etc. are more crucial/critical. 

[shooter]

not just 1

agree, 3.5vac and 6vac is huge.

I'd pull every tube right of V2, stabilize, verifying with a good known signal, then add a tube, repeat

[Willabe]

I'd pull every tube right of V2, stabilize, verifying with a good known signal, then add a tube, repeat

Agree.

If it were my amp I'd try this;

1. Put the OT and all of it's wiring/jacks/impedance selector/line out jack back in, all OT leads tightly twisted and tucked up into the chassis corner.

2. Change/separate the power tube/PI ground from the preamp tubes ground.
(.....because your amp has a -FB loop with a presence control, the PI, including the presence control should be grounded with the power tube grounds. The MV feeding the PI should be grounded with the PI tube grounds. Everything before that MV goes to the input jack ground buss. So B+ nodes A/B/C get grounded together and nodes D/E get grounded together.)

3. Add grid stop R, and shielded cable. Because the amps unstable, I'd start hi, ~100K and if that knocks it out, then keep lowering the value until it starts up again, then go back to the last higher value stopper R.

4. I'd leave the plate caps to ground in when I did this. (But take out that large K bypass cap.)

EDIT: I'd try the amp between step 2 and step 3. Just to hear if there's any difference. Then I'd go on to steps 3 and 4.     

[Fiat_cc]

I have been considering grounding the preamp bus at a separate point near the input jack

that WILL help

can't find it in this thread,
did you measure each rail node with your meter VAC?  put the neg lead at the main PS ground bolt, then repeat with the neg lead somewhere on the chassis

No, I haven’t done that. Would you do this at idle whilst stable, or once the oscillation has been induced?

[Fiat_cc]

Can anybody see a benefit in putting a tightly wound piece of wire around the OT leads, and grounding at one end, similar to what Fender did in some amps. A friend suggested this might help at least a little.

[Willabe]

Can anybody see a benefit in putting a tightly wound piece of wire around the OT leads, and grounding at one end, similar to what Fender did in some amps. A friend suggested this might help at least a little.

I don't know if that would work? I guess you could try it?

It's about how much coverage you can get. That's why good braided shielding is better than a spiral wrap shielding, more/tighter coverage. 

At this point anything might be worth a try, but I think those grid stoppers and shielded grid wires will get the bulk of it, maybe all of it. Those are a must try in this amp.

Grid stop R's have been used very effectively, in many circuits/amps, as a main go to for decades, because they work so well. That is, if installed correctly, tight to the socket grid pin and a large enough value.       

I have bought braided tube shielding (from Mouser?). I used a flat blade screw driver to open it up and slid it over the wires. Then I put heat shrink tubing over it and grounded it at 1 end. It worked great.

The trick is getting the right diameter size so the braid diamond pattern stays tight.

Mouser and/or Digikey should have the right size of both the braided shielding and shrink wrap tubing. If your careful you might not even need the shrink tubing.

You might be able to buy it on ebay?

I think you might need to add a shielding plate in between the preamp tube sockets and the OT wires/jacks/line out/impedance selector.

The other thing is, you say that just removing the OT wires/speaker jacks/etc. didn't do much, but when you also moved the OT out of the chassis, that helped.

So, maybe the interference is coming more from the OT and not as much from the wires and jacks?

You might try laying a metal plate on top of the OT inside the chassis as a test. I would.

Fender and others used shielded OT's that were completely on the out side of the chassis. Your OT has no shield cover and that open end is inside the chassis. Not good for being right next to the preamp tubes. 

[Fiat_cc]

Ok, so I put the OT and associated switch and sockets back inside the chassis, then read the Merlin article on grounding. I’ve dnded up with a continuous bus, that is grounded only at the input jack, which then has a kind of multiple star arrangement. That article says not to use the chassis at all for circuit return paths, so only the heater center tap and the silverface style wrapped wire shield I put around all the OT leads are grounded to the chassis, plus the ground bus reference at the input jack. Filter caps connect to the ground bus immediately upstream of the last item in the circuit that is fed from that node.

I also did some rerouting of lead dress, and ran shielded sire from input to V2a, and from the gain knob to V2b. I will add a run from V2a coupling cap to the gain knob today. I put 68k grid stoppers on both V2a and b, right at the socket pins. It is much more stable than it was. With MV and gain wide open, it is stable, with around 450mV p to p of hash on the scope at the speaker output (with all tone controls at zero. It becomes unstable again once I start winding up the time controls. One interesting thing that happened was, after all the rewriting, I thought it prudent to fire it up on my light bulb limiter, just to be safe. Fired up fine. I then did a few things, and forgot it was on the limiter. Fired it up and started to do stability tests on the scope. It was oscillating in a minor (comparatively) way, before it started to flash the light globe. The bulb was blink to full brightness, then almost off, around twice a second, and I could hear a corresponding ‘click’ from the OT. Thoughts?

Anyway, later I’m going to add more shielded cable runs where appropriate, add a 100k grid stopper to V3a, and experiment with adding a shield on top of the OT, and one between the speaker jacks and preamp sockets. Will aluminium work ok, or does it need to be a ferrous metal?