Marshall JCM900 bias issue

[Randall]

Marshall 50W JCM900 model 4500 Dual Rev comes in for a cleaning, and because "it just doesn't have what it used to have". It has a mismatched set of a big bottle GE and a skinny bottle Telsa (I think), so I sell him a new set of JJ's. Problem is, I can't get them to draw close to the proper current. With the bias pot full CCW I can only get 12.9mA on one and 13.6mA on the other, this is with no meter connected to pin 5. With a meter connected to pin 5 I get 2mA more on each tube.

I don't get it. I know it's not the new tubes, because I found the same thing with the originals. I must admit, I don't totally get how this bias circuit is working, but I do know it is delivering a few volts more negative than the drawing calls for at the set point. And I don't know if this is related, but it makes a "fwooOOP" sound when the power switch is shut off, regardless of the standby.

Here are my readings, both tubes are within a volt of each other:

pin 1: 0v (tied to pin 8)

pin 2 - 7: 6.7vac

pin 3: 465v

Pin 4: 462v

pin 5: -45

pin 8: 0v

It is interesting that the tubes pull more current with a DVM on pin 5, could this be the meter pulling the bias voltage a bit less negative?

http://www.thetubestore.com/lib/thet...-Schematic.pdf

[The Dude]

There's a series cap from the bias winding to the bias rectifier. Check that bad boy and check it well. IMO, one of the dumbest things you could design. When the cap fails, and they do, you lose bias all together. Results are "less than satisfactory".

[nsubulysses]

It is probably pulling more current with DVM probe on pin 5 because it's making the amp hum or oscillate so when it puts out a signal it draws current I would guess. Do you have the volume turned up when you touch the probe to pin 5?

Schematic says to set the bias voltage to between -40 and -42V so it would have to go more positive to be biased to what it wants on the schematic.

Is the problem that the bias circuit doesn't reach a good bias point? You may have to adjust R7, R28, or R29 values to get to a more usable range. Paralleling R7 will bring the bias voltage more positive. Put the bias pot in the middle of its rotation first and then adjust a resistor until the bias is in a good range for these tubes. Then you will have wiggle room up and down withe the trimpot.

[Enzo]

Imagine your meter is a 1 meg resistor. When connected to the circuit, that 1 meg is parallel whatever grid return there may be and so changes voltage division, so I imagine it could knock bias don a volt.

Tubes don't have to be defective to draw more or less current than some other tubes. You may just have a cold set.

Bias voltage on a schematic is just a ball park, they have no way of knowing what voltage any given tube will need. That is why it is adjustable.

DO you have any more tubes? New used whatever? Try them and see.

[Randall]

Enzo you are right, different used tubes yield a better result, in mid 20 mA range, but still not right. Before I go adjusting the bias circuit, I wonder was it designed this way, and has it always been this way, or did something change? To answer a question, yes it seems the bias circuit doesn't give the proper range adjustment. It's as far as the pot will adjust and I can't get to where I need to be. SO, maybe I have two issues, a cold set of new tubes, and an out of range bias circuit.

I tried paralleling R7, but it actually makes the bias voltage go more negative. Example, clipping in another 56K makes it go almost 2v more negative. What gives?

[nsubulysses]

Sorry to mislead you on the paralleling.

So if you parallel it and the bias goes in the wrong direction you need, lift one end of the bias resistor from the board and put yoru ressistor in series with it. This should get the bias going in the other direction. It can be confusing sometimes with negative voltages and capacitors backwards and series/parallel arrangements, so sorry.

Does the amp put out 100W with new tubes? Do all resistors in the bias supply measure the correct resistance? youi could parallel the bias caps C13 and C14 to see if anything changes but I bet it won't . You have plenty of bias, not not enough. So it seems the bias circuit is working good, just a little out of alignment for the current power tubes is what I would guess.

People who have been at it for a few more decades than me, if this bias circuit is calibrated on current draw of EL34s from the early or mid 90s, would this possibly make the bias circuit out of whack with variance of tubes of today?

[g1]

How are you measuring the bias current?
Is it set up for the EL34's? There were 2 versions (6L6/5881 & EL34) and sometimes people swap tube types without making the proper adjustments to the circuit.

[Randall]

Well, as it turns out nsubulysses, you were correct about paralleling R7. I don't know what I was doing wrong, but after clipping in a 50K pot I found the magic number, 17K. I made one out of a 15k and a 2.2K, and voila! Now these cold new tubes are purring along at 33mA/460v, with plenty of room to tame a set on the other side of the spectrum.

The amp is all cleaned up and quiet, it sounds good, so I am going to call this a success! Thanks for the help as always fellas.

[g1]

R7 was 56K and you paralleled a 17K ?

[Randall]

yep. At first I was trying to guess with single resistors, and got nowhere. It gets the bias voltage down to -40v and it seems to be happy, and it still goes up to -49v.

[g1]

That is way too much change to account for tube variations. You should look for a fault. That's a fair bit of extra load on the raw bias and I'd expect it to affect the charge pump cap (C15). Has C15 maybe been changed to a different value?

(edit: like The Dude mentioned in post #2, if C15 is on the way out, there will be a major meltdown)

edit: corrected typo, bias feed cap is C15, not C12.

[Randall]

Do you mean C15? If so, what is a class X?

To clarify, the amp uses EL34s, and I measure current with a Hoffman tube socket jig, whatever he calls it, I think it puts the ammeter in the cathode path. I don't believe the amp has been changed in any way I can notice, and so far all resistors, the electrolytics and the diode measure unsuspicious in circuit. I haven't looked at the series cap that has been mentioned, but I'd like to know what to replace it with. All I see is that it is a 47pF class X.

I really thought I was done with this, another one with seriously way too much time into for the 1.5 hrs labor I can charge him to replace tubes, bias, clean, and replace reverb pan.

[J M Fahey]

Wht do you call "right" ?
What do you want to achieve?
What tubes areyou using today or plan to use?

[Randall]

J M, by "right" I mean biasing the tubes at a reasonable level, above the approx. 16mA max I could get as it was. Some suggested modding the bias circuit, and this I tried. But now some suggest this was not the right way. I want to achieve making this Marshall operate like a good sounding Marshall by putting the outputs at somewhere in the average current draw zone. I am today using a new matched Pair of JJ EL34s. The future I cannot speak for.

The customer brings this to me saying it doesn't sound right, should I not be concerned with such a low current draw on a 50 w Marshall?

[Mick Bailey]

I've found current draw on tubes to be pretty inconsistent and have the same issue right now with an amp where the bias pot has nowhere near the range to get a sensible draw on a set of Shuguang KT88.

As an illustration, I have a pile of Shuguang 6L6 tubes that in a given circuit pull between 19mA and 54mA, though if I disregard the extremes the average is closer to 28mA. That's some spread and most bias circuits would maybe not accommodate that range. You could perhaps infer from this that if I had enough tubes to match into quads I could have some sets that pull 19mA and some that pull 54mA. Both are viable and I guess that's exactly what tube vendors do.

As a bulk manufacturer, Marshall could spec an upper and lower limit from the factory so that everything they buy is within the bias adjustment of their amps.