Marshall 1959HW Bias runaway

[Mathias]

Hi all, seeking some advice on my Marshall 1959HW. I've had the amp for a little while and figured I'd service it. Came to the conclusion the bias was running away, one tube was red plating. From what I learned now those 1959's are notoriously to bias. I've had in the back of my mind to make this to super bass spec so I figured it's a good time to do. As it eliminates some possible causes of bias runaway other than power tubes. Conversion is done, all coupling and bias caps are replaced. From the 4 power tube sockets only the outside had control grid resistors. I've added 5.6k's on all 4 sockets. I've placed in a fresh quad Electro Harmonix EL34's but unfortunately idle current is still going up from 27mA to 30-35-40-45... at this point I shut the amp off. Without power tubes the amp seems stable and bias voltages on pin 5 are all good and stable. Under load around 485V plate that drops slowly with the current rising. I wonder what could cause this to happen. One valve seems 5mA out of spec, not sure if this imbalance is causing the current draw.

Any advice is much appreciated! Added some gut shots after conversion.

 

[radardoug]

You need to watch the grid volts, does the grid bias drop? If so, could be cap leakage from the driver stage.

 

[Rusan]

By "under load around 485V plate that drops slowly with the current rising," are you referring to the plate voltage dropping because of Ohm's Law losses across the output trafo primary's DCR, or the negative bias voltage dropping? And, on two tubes or all four? As radardoug said, it's likely a leaky coupling cap (regardless of whether it's new) if it's happening with just two tubes. If the current drift is happening across all four EL34s, I'd say it's certainly rare/odd that you'd have two bad coupling caps, but not impossible.

I once tried some Sequa paper in oil caps, because they were about 1/6 the price of Jensen PIOs. After one was leaky when brand new and another couple went leaky shortly afterward, I junked em all and cut my losses. Just several months ago, I got a bad tube of opamps from a credible, authorized vendor. Bad batches of components do happen sometimes, however infrequent they may be.

Forgive me if you're meaning that the bias voltage is still stable when the current rises, but if not, record the voltages throughout the bias supply without the power tubes in, then when the amp is first turned on and the tubes start conducting, and then when the current rise happens. That should let you know exactly where the bias voltage drop is originating from.

I once ran into something kinda sorta similar to this on a 50 watt JCM 800, where both power tubes would hotplate for a little while, then stop and repeat again intermittently, usually after being on for 30 minutes or so. The owner had already tried different tubes, so that was ruled out. As it turns out, the series resistor after the bias supply rectifier diode was intermittent when the amp got hot. Lightly tapping on it would cause the bias voltage to flicker. I've seen that before with a carbon comp, but never with a film resistor; didn't think it was possible.

 

[Mathias]

I have the current rising on all four. I use a TAD Bias Master with 2 probes, switched them from inside to outside and seems to have current rising on all four.

With no load and no power tubes:

Wall 232.5v / Standby B+ 517v / On B+ 510v
Voltages on all 4 grids with in Standby and in full operation are averaging -49.6v with a 0.1v difference.

With load and power tubes:

Initial startup:
B+ 490v
Grids -49.1V
Idle current +- 27mA

After 10-15 minutes
B+ 484v
Grids -46.5V
Idle current +- 35mA

Seems like the Bias is dropping and yes I assume the B+ is dropping due to ohms law...

It did this on all four tubes prior to changing caps, and the new ones are quality SoZo. It's not impossible but seems very rare to have a bad pair, only to be replaced with another bad pair.

 

[Rusan]

Sorry, I didn't catch that it was already doing this before you replaced the coupling and bias filter caps. And, I wasn't thinking about the fact that the PI coupling cap outputs aren't isolated from one another at DC, due to the connected bias feed resistors. Forget that two tubes or four tubes thing.... Duh, I amaze myself sometimes. I've worked on so many cathode biased amps that it sometimes pollutes my thinking when it comes to grid bias, lol.

With no tubes, tubes in and freshly on, and then 10 - 15 minutes later, check your AC voltage both before and after the 22K resistor, the DC voltage after the rectifier diode, and then at the junction of the bias trimpot, 15K and 220K resistors. This will determine where the C- voltage drop is originating. I know that measuring the AC voltage sounds silly, but it's best to leave no stone unturned.

Also, after it's been on for the 10 - 15 minute period, check to see if the output trafo feels abnormally hot; it should only be slightly warm at most.

 

[CJ]

which tube socket is red plating?

did you try swapping tubes to see if it follows the tube?

monitor the bias voltage on pin 5 with the tubes in. is it dropping?

those amps are not known for conductive circuit boards, but is on the list of things to check for,

you can experiment with 56 K grid resistors instead of 220 K,

there is only about 100 or 200 ohms between the plate supply and the screens, so those tubes are running hard.

you could experiment with adding a resistor to drop[ the screen volts down a bit,

 

[radardoug]

Max screen voltage is 425 volts according to Philips. Most guitar amps are run at stupidly high voltages. Dont be surprised if they dont like it.

 

[Mathias]

Sorry, I didn't catch that it was already doing this before you replaced the coupling and bias filter caps. And, I wasn't thinking about the fact that the PI coupling cap outputs aren't isolated from one another at DC, due to the connected bias feed resistors. Forget that two tubes or four tubes thing.... Duh, I amaze myself sometimes. I've worked on so many cathode biased amps that it sometimes pollutes my thinking when it comes to grid bias, lol.

With no tubes, tubes in and freshly on, and then 10 - 15 minutes later, check your AC voltage both before and after the 22K resistor, the DC voltage after the rectifier diode, and then at the junction of the bias trimpot, 15K and 220K resistors. This will determine where the C- voltage drop is originating. I know that measuring the AC voltage sounds silly, but it's best to leave no stone unturned.

Also, after it's been on for the 10 - 15 minute period, check to see if the output trafo feels abnormally hot; it should only be slightly warm at most.

No worries at all, any way of thinking to fix this is much appreciated. I'm very grateful you guys are helping me out. From my measurements it seems that the bias supply is stable. Without and with power tubes the AC voltage has only 0.5V fluctuation over a 10-15 min period. DC voltage about 1 volt of fluctuation after the diode and bias junction. But that is on a 25mA difference on the power tubes!

Would it be safe to say we can rule out the transformer (feels cold as well) and bias circuit?

 

[Mathias]

which tube socket is red plating?

did you try swapping tubes to see if it follows the tube?

monitor the bias voltage on pin 5 with the tubes in. is it dropping?

those amps are not known for conductive circuit boards, but is on the list of things to check for,

you can experiment with 56 K grid resistors instead of 220 K,

there is only about 100 or 200 ohms between the plate supply and the screens, so those tubes are running hard.

you could experiment with adding a resistor to drop[ the screen volts down a bit,

I'm using a new quad of tubes, haven't seen one redplating as I shut off the amp in time. Current rises on all four sockets.
Seems like the bias supply is stable but bias voltages on pin 5 are indeed dropping around 5 volts after 10-15 mins.

The screen voltage is indeed running very high at 475 / 465.

I'm gonna halt the troubleshooting for a second. The original factory set of tubes were Tung-Sol rated at 425V, same as my current set of EHX.
So it's not impossible both set of tubes suffer the beating... I'll see if I can change out EHX's for a cold set of TAD STR's. Those TAD's screens are rated to 500V. Maybe a colder set will help as well in self stabilising... just a thought, I'd love to hear opinions on this.

 

[Tubetec]

Looking closely at the solder joints on the bias filter caps , they look a bit funky , not as clean and shiny as the rest ,
ocassionaly the paper tape components come rolled up in leaves a gluey residue behind on the lower end of the component legs , its always best to clean this off before using the component or it can react with the solder/flux .

 

[MagnetoSound]

Have you checked for good tension on all the tube socket pins?

My MO in these situations is to install the power tubes one pair at a time. Try each pair in both pairs of sockets. Then you can rule out any poor connection between tube pins and receptacles.

Apologies if I’m stating the obvious or something you’ve already covered.

 

[Tubetec]

Look for any signs of arcing across the tube sockets , than can sometimes leave behind a slightly conductive carbon residue .

So this is a Marshall amp that youve rewired point to point ?

Maybe the amp is osscillating or unstable at a very high frequency ,
the purple feedback wire from the transformer secondary can be a source of this trouble if its badly routed in the chassis

 

[Mathias]

Look for any signs of arcing across the tube sockets , than can sometimes leave behind a slightly conductive carbon residue .

So this is a Marshall amp that youve rewired point to point ?

Maybe the amp is osscillating or unstable at a very high frequency ,
the purple feedback wire from the transformer secondary can be a source of this trouble if its badly routed in the chassis

Thanks for the input, It's a factory Marshall 1959HW, PTP by default, I only changed out the capacitors. So all the wiring is factory.
I'll check that out. I've read some quality control issues regarding the wiring.

 

[Mathias]

Have you checked for good tension on all the tube socket pins?

My MO in these situations is to install the power tubes one pair at a time. Try each pair in both pairs of sockets. Then you can rule out any poor connection between tube pins and receptacles.

Apologies if I’m stating the obvious or something you’ve already covered.

I'll give that a go. I'm always assuming the amp is fairly new so I tend to see over those obvious things... thanks for the suggestion.

 

[Rusan]

From my measurements it seems that the bias supply is stable. Without and with power tubes the AC voltage has only 0.5V fluctuation over a 10-15 min period. DC voltage about 1 volt of fluctuation after the diode and bias junction. But that is on a 25mA difference on the power tubes!

1V can't cause that much current increase. You've got something else happening.

Would it be safe to say we can rule out the transformer (feels cold as well) and bias circuit?

The output transformer temp check is a "poor man's test" for ultrasonic oscillations. Sometimes, if an amp is oscillating at high enough frequency and amplitude, the OPT will heat up pretty quickly.

Tubetech mentioned poor lead dress in the NFB connection causing an oscillation (good call!). I'd thought about this next thing earlier but didn't say anything, because from your description the tubes were already hotplating while just playing the amp normally. For future reference, I've had a few Marshalls oscillate when the power tubes are plugged into bias probe sockets. If you disconnect the NFB wire from the OPT to the PI it generally stops, so I just preemptively unsolder it before biasing new tubes.

I did notice that one of your PI coupling cap to grid stopper signal wires is running parallel to the main B+ ground. That can cause issues, as there are heavy currents flowing in that ground wire. It might be good to float the signal wire up in the air, or at least away from the ground wire and chassis in that area.

An example of what improper lead dress can do: I once worked on a hi-fi amp that was rewired with silver signal wire, and it was oscillating afterward. This particular amp was poorly designed in many ways, but the physical layout was especially horrendous. Among other things, there were signal wires routed directly against one of the output transformers, which were inside the chassis. The OPT's magnetic field was inducing a voltage in the silver signal wires, which caused the oscillation. Silver has 200% higher magnetic susceptibility than copper, so the same run in copper didn't cause any major issues. I re-routed the wires as best I could, which stopped the oscillation.

Another example: some blackface and silverface Fenders make an annoying "ticking" sound when using the tremolo circuit, which is easily solved by tidying up the lead dress, running a direct ground, and installing a .01uF cap. AAMOF, I have a friend's silverface Twin on my bench right now, that was ticking before I fixed it 8 or 10 years ago. I bet CJ, Tubetech and some of these other guys have fixed a thousand of 'em.

 

[Tubetec]

I once saw an otherwise working and reliable tube amp go red plate during a gig ,
We had to stop the show for a while , I checked the mains sockets and sure enough there was no ground connection ,
We ran a plugboard from another nearby socket with an intact ground , no problems after that .
Of course I let staff know they had a potenitally lethal situation up on stage ,

I had an EBS bass amp in one time , it had a strange noise going on in the backround , the owner wanted to sell it but was afraid he wouldnt make the money back if the thing was showing a fault ,
After a lot of head scratching it turned out to be a bundle of cables secured down to chassis had come adrift from an adhesive mount , it had shifted no more than an inch from its original position ,but that was enough to allow the RF gremlins in .

I once got a call from an Austrailian band on tour here , they had a homemade AC-30 style amp they had a problem with ,
I got it up on the bench ,plugged in switched on and it opperated more like an oscillator than an amp , according to the other band members it always sounded a bit like this ,but had gotten worse to the point it was unusable .
The guy who made the amp had included a schematic inside ,
I cant remember precisely what it was in the end ,but a component was wired up wrong ,
I solved the problem and the amp didnt sound half bad , but the guitarists sound was completely gone and his head was all done in .
I refused to turn his amp back into an osscillator , wished them the best and sent them on their way . I did include a note inside for the guy who built it describing the changes Id made .

 

[Matador]

Seems odd that the -49.1V slides down to -46.5V after 10 minutes (this is measured at pin 5?)

Do you have a DC voltage difference across any of the 5.6K grid stopper resistors?

 

[delgado]

The drifting grid voltage worries me too.

Did you once continue the measurement (post #4) by switching to Standby again and then looking at the grid voltage again?
Does it stay at -46.5V or recover to the initial -49.6V DC?

 

[CJ]

The pin 5 voltage tells me that the tube sockets have become conductive. Voltage is leaking from 4 to 5 which causes current to flow through the 220 k resistors in a direction which is causing an opposition voltage to the negative bias voltage.

That's why I suggested the 56k resistors in place of the 220 k. You don't have to pull the resistors, you could tack on a couple of 82 k resistors in parallel with each if the 220 k resistors.

What this does is lower the voltage drop across the 220 k resistors caused by the leakage. The conductive path from pin 4 screens is shunted to ground via the bias supply with a reduction of 400 percent leakage voltage, which should be enough to tell you if it is the tube sockets. I just did this to a pair of JCM 2000's that same through the shop.

Don't worry about the screen voltage being over the handbook voltage ratings. This has been going on ever since they started making tubes. There are thousands of guitar amps out there running these voltages.

 

[Mathias]

Seems odd that the -49.1V slides down to -46.5V after 10 minutes (this is measured at pin 5?)

Do you have a DC voltage difference across any of the 5.6K grid stopper resistors?

The drifting grid voltage worries me too.

Did you once continue the measurement (post #4) by switching to Standby again and then looking at the grid voltage again?
Does it stay at -46.5V or recover to the initial -49.6V DC?

No DC differences but let me check if that voltage recovers back. I haven't thought about that.