Bias Drift?

[Firestorm]

If I'm reading the incredibly ugly (even for Marshall) schematic right, everything in the bias circuit is shared by both sides, except the 220K grid load/bias feed resistors and 22K trim pots. If you already replaced the 220Ks, the trimmers would be suspect wouldn't they? BTW, the pots do not seem to have "safety" resistors to guard against the wiper opening up; you might consider adding those.

Are you measuring current using the 1R 1W cathode resistors? I think I'd replace those too. If Marshall could use crap resistors in one place, they could certainly do it there as well. That is, the imbalance could be a current sense resistor imbalance rather than an actual current imbalance. If you have steady hands and paid-up life insurance, you might shunt the OT to confirm the current readings.

[Buddha's Guitar Tech]

Brand new JJs.

Swapped the power valves around and the phenomenon exists on two sockets. V5 & V6

Even if the 1R resistors are dodgy, they wouldn't cause the current to rise appreciably would they? After all, it's only 1 ohm. Suppose i could shunt them to negate any effect, but i don't think it's anything to do with them.
If they were to increase in value, current would decrease and plate voltage would rise. If they were to decrease (by 1ohm to a short) then current would settle. Unless time and space don't exist in my house, which, sometimes, i think happens.

The facts i've measured are: during this phenomenon, plate and screen voltages voltages go down - consistently across all four EL34s - a sign of increased current - AND bias voltage 'rises' towards zero. (on V5 and V6 appreciably, and less so on V7 &
Forgetting bias voltage is negative, let's just say it's falling too... towards badness.

If grid voltage is not the cause, what else would cause increased current?
The valves themselves? Possibly, but as stated above, the phenomenon seems to be local to two sockets.

One leg of OT primary resistance dropping, thus increasing plate voltage? Hardly likely, considering the OT is far from the main sources of heat.

If, for some reason the B+ is rising just enough to cause more current and sucking the B+ down again (from perhaps a dodgy reservoir cap) then i'd expect an even current increase over all four valves, but that's not happening. (sounds far fetched, i know)

I did look at the trimmers (and how close they are to the valve bases) and said a wee prayer. But it is plausible they would be culprits.

Why why why why why did i ever think that taking stuff apart was cool? I could've been an actor.

[Firestorm]

Even if the 1R resistors are dodgy, they wouldn't cause the current to rise appreciably would they? After all, it's only 1 ohm. Suppose i could shunt them to negate any effect, but i don't think it's anything to do with them.

What I meant was IF you are measuring current by checking voltage across those resistors and they drift, it will make you think the current on each side is different even if it isn't. If you are measuring plate current by shunting the OT, never mind.

You did replace the 220K bias feed resistors didn't you?

[Buddha's Guitar Tech]

220K feeders were going to be next.

Hadn't shunted the OT. The dropped voltages at the plates were proof enough for me that current was on the rise.

I've been reading more about this problem. This is a pretty nice-looking page with an interesting solution....
http://www.hullerum.de/Marshall/TSL122repair.html

Some say the material of the board itself causes problems.

You'll notice the amp in question is a TSL and not a DSL, but the board is the same as the one on the bench - JCM2-60-00

[Firestorm]

220K feeders were going to be next.

Hadn't shunted the OT. The dropped voltages at the plates were proof enough for me that current was on the rise.

Didn't you say that plate voltages were dropping consistently across the four tubes?

[Buddha's Guitar Tech]

Now that i think about it, i measured the voltage at the B+ node, sorry. At the time, i was only interested in knowing whether or not current was rising in general.

Although, I do remember for certain, the individual screen voltages dropping consistently.

The most striking thing so far is the dropping bias voltages. And that others have seen this same thing.

Since the drop across the 1R cathode resistor read as an increase in current, AND the associated bias voltages were fluctuating, i saw no reason to mistrust the cathode resistors.

[Firestorm]

I get you. Yeah, if the negative voltage is different on V5-V6 from V7-V8 that would be "real" (not a measurement error from drifting current sense resistors). Sort of implies pots, bias feed resistors or some of the other nonsense in the article you linked. Really a deeply flawed design.

[paddy]

I have repaired this problem in four of these amps over the last year. Three of them had dry joints on the bias adjustment pots. The other one was due to a fault in the main PCB. More specifically, it was related to the PCB material itself. I read an article online (can't remember where-sorry) which suggested that the PCB material became conductive over time. When I replaced the board, the bias was rock solid. From memory, the board cost $250 Australian dollars from Marshall.
Try resoldering the bias pot board first.

Good luck,
Paddy

[bfrederi]

FWIW I have had about six of these Marshals pass through my shop over the last few years with very similar problems. In two or three instances customers brought me their amps with tubes they had already purchased and they happened to be JJ EL34's in these instances I had to replace the 220K grid stoppers (replace with 5.6K)and had to use a different EL34. In both cases I remember I had a set of Svetlana pulls that were in pretty decent shape and installed them and had no problems. I would re-install the JJ's and notice one or two of them drifting badly.... they were actually starting to draw grid current. Those customers returned their JJ's and replaced with Svetlana's and their amplifiers have been working fine.

I have wondered if there was a faulty run of the JJ EL34's as for the most part they have been a great tube. I just have not been recommending them in the TSL Marshall's.


I have also run into other issues with this bias circuit, but wanted to see if you have been able to install a different make of EL34 and had the same results?

Good Luck
-Brad

[Jana]

Scroll way to the bottom of that link that I posted. It discusses what they say is the cure for the bias drift--modifying the board so that the conduction through the board doesn't change the bias.

[Buddha's Guitar Tech]

Done. I've employed a Fender-type solution to a Marshall problem.

Much better performance now, although there is still a little bit of drift; a few milliamps in half an hour. Beats the old data by a mile.
The slight rise might well be normal, but when I blew on the circuit board, the current normalised in seconds. So I'm telling the owner to use a fan to try circulating what little air he might.

Many thanks to Jana for suggesting the link, its author JC Maillet, and ultimately, Fred Skenderbeg for the winning solution. And of course, everyone who participated here, thanks a bunch.

[img:640:480]https://lh3.googleusercontent.com/-NgpQ ... G_3079.JPG[/img]

[img:640:480]https://lh6.googleusercontent.com/-kUxc ... G_3077.JPG[/img]


[img:640:480]https://lh4.googleusercontent.com/-EAzs ... G_3075.JPG[/img]


[img:640:480]https://lh3.googleusercontent.com/-dvLK ... G_3066.JPG[/img]