50w Marshall..uneven bias

[JustMike]

 It's an amp I built a while back and although it's been playing fine for a few months now, when I plug it into an attenuator and crank the volume, I can see the tube in the left socket red plating. And it's the socket, not the tube. When I measure the bias voltage, this socket is always 4-5 mv higher than the other.
 What can be causing this? I was told over on Metroamp that it could be a leaky PI cap, but I changed them with no different results.
 I did find that the 5w grid leak(I think) resistors were mismatched. I had accidentally put in a 1.5k on one tube (the one with the higher reading)and a 1k on the other. They're now both 1k. Still no difference.

[bobmegantz]

some suggestions:

1. install 1-10 ohm resistors from each output tube cathode to ground and measure the bias current through each tube.  The tube that is red should have higher bias current.  If the currents are different, swap the tubes and see if that changes anything (I know you said you did this already, but make sure).
2. Disconnect the coupling caps and remeasure.  Since you already changed the caps, let's assume that nothing changes.
3. Check the bias resistors to make sure they're the correct value.  Assuming they are,
4. Then, you have a wiring problem.  Something else is touching that grid pin and dragging it positive.

hope this helps...

[JustMike]

Hmmm...I've double checked everything. The only thing I can think of is the socket itself. I've heard that there can be some conductance thru the socket. I got them at a surplus shop. They were new and look like black plastic, not the old brown or ceramic.

[plexi50]

Remove the power tubes and power up the amp. Check the socket voltages and grid VDC

[HotBluePlates]

When I measure the bias voltage, this socket is always 4-5 mv higher than the other.

Explain this statement.

Does this mean that you measure the negative voltage at pin 5, and the redplating socket has 4-5mV more negative voltage than the other socket?
   - If this is what you meant, I can't see how a couple millivolts of bias voltage difference will have any impact.

Or does this mean that you have 1 ohm cathode resistors installed, and when measuring millivolts across the 1 ohm resistors that you read 4-5mV more across the redplating socket (implying 4-5mA more idle current through that tube)?
   - If this is what you meant, that could mean the redplating socket is idling a little hotter than the other tube. 4-5mA of idle current difference should not really matter (I bet most "matched tubes" aren't that tightly matched).

You mentioned 1k resistors. I assume those were connected to pin 4 (screen grid), because you mentioned 5w resistors. If the offending socket had a 1.5k 5w screen resistor, it would actually be less likely to redplate because the increased screen resistance could have the effect of limiting plate current under certain circumstances. So that change would have no effect on this issue.

If these 1k resistors were connected to pin 5, they would probably be grid stopper resistors, but you might have used a larger value here (5.6k is the common value quoted for Marshall-style amps, which I assume describes your amp). Exact matching of these is unimportant. Do you have any grid stopper resistors? Don't confuse these with the "grid reference" or "bias resistors" which are commonly 220k or 100k resistors running from pin 5 to the bias supply. A grid stopper resistor would commonly run from the junction of the coupling cap and bias resistor (or a wire connected to this point) and terminate right at pin 5 of the socket.

Too-big grid reference resistors (1M or more instead of 100-220k) could cause redplating due to gas within the tube. Are your grid reference resistors in the 100-220k range?

Lack of grid stopper resistors might lead to redplating indirectly because of oscillation. Grid stoppers reduce the response of the tube to very high frequency, reducing the risk of oscillation. An oscillating tube might oscillate higher than you can hear (or the speaker can reproduce), and the tube dissipates excessive power at this ultrasonic frequency. This is more likely if you have very long grid wires running to pin 5, or if the output tube plate wires running to the OT pass anywhere near the phase inverter.

What idle current and plate voltage do you have in your amp? (This means, "Is the bias wrong to start with?")
Is the redplating socket loose? ("Could the socket be lossing contact with pin 5, and losing bias?")

[darryl]

If the redplating is only occurring when the amplifier is cranked while driving an attenuator, it may not be a DC bias problem, but rather an AC signal problem. If by "cranked" you mean driven well into clipping, then the output valves may be receiving an asymmetrical signal from the phase splitter. This can cause one valve to dissipate more heat than the other. If possible, have the amplifier tested with a signal generator, dummy load and oscilloscope, so the output waveform can be viewed as the amplifier is cranked.

Another potential source of trouble is the attenuator. Some versions of these have a rather bad reputation for presenting a mismatched load to the amplifier. This mismatch can cause excessive heating in the output valves.

[JustMike]

When I measure the bias voltage, this socket is always 4-5 mv higher than the other.

Explain this statement.

Does this mean that you measure the negative voltage at pin 5, and the redplating socket has 4-5mV more negative voltage than the other socket?
   - If this is what you meant, I can't see how a couple millivolts of bias voltage difference will have any impact.

Or does this mean that you have 1 ohm cathode resistors installed, and when measuring millivolts across the 1 ohm resistors that you read 4-5mV more across the redplating socket (implying 4-5mA more idle current through that tube)?
   - If this is what you meant, that could mean the redplating socket is idling a little hotter than the other tube. 4-5mA of idle current difference should not really matter (I bet most "matched tubes" aren't that tightly matched).

1 ohm cathode resistor method.

You mentioned 1k resistors. I assume those were connected to pin 4 (screen grid), because you mentioned 5w resistors. If the offending socket had a 1.5k 5w screen resistor, it would actually be less likely to redplate because the increased screen resistance could have the effect of limiting plate current under certain circumstances. So that change would have no effect on this issue.

Yes, 1k 5w  between pins 4&6.

If these 1k resistors were connected to pin 5, they would probably be grid stopper resistors, but you might have used a larger value here (5.6k is the common value quoted for Marshall-style amps, which I assume describes your amp). Exact matching of these is unimportant. Do you have any grid stopper resistors? Don't confuse these with the "grid reference" or "bias resistors" which are commonly 220k or 100k resistors running from pin 5 to the bias supply. A grid stopper resistor would commonly run from the junction of the coupling cap and bias resistor (or a wire connected to this point) and terminate right at pin 5 of the socket.

5.6k grid stoppers

Too-big grid reference resistors (1M or more instead of 100-220k) could cause redplating due to gas within the tube. Are your grid reference resistors in the 100-220k range?

Lack of grid stopper resistors might lead to redplating indirectly because of oscillation. Grid stoppers reduce the response of the tube to very high frequency, reducing the risk of oscillation. An oscillating tube might oscillate higher than you can hear (or the speaker can reproduce), and the tube dissipates excessive power at this ultrasonic frequency. This is more likely if you have very long grid wires running to pin 5, or if the output tube plate wires running to the OT pass anywhere near the phase inverter.

220k reference resistors. But I guess I could redress the grid stopper wires, but they're run under the board directly to the sockets...pretty conventional.

What idle current and plate voltage do you have in your amp? (This means, "Is the bias wrong to start with?")

Don't know, haven't checked it yet.

Is the redplating socket loose? ("Could the socket be lossing contact with pin 5, and losing bias?")

I replaced the sockets.

[JustMike]

If the redplating is only occurring when the amplifier is cranked while driving an attenuator, it may not be a DC bias problem, but rather an AC signal problem. If by "cranked" you mean driven well into clipping, then the output valves may be receiving an asymmetrical signal from the phase splitter. This can cause one valve to dissipate more heat than the other. If possible, have the amplifier tested with a signal generator, dummy load and oscilloscope, so the output waveform can be viewed as the amplifier is cranked.

Another potential source of trouble is the attenuator. Some versions of these have a rather bad reputation for presenting a mismatched load to the amplifier. This mismatch can cause excessive heating in the output valves.

I'm using a home made attenuator with 4 25w 8 ohm resistors.

[sluckey]

When I measure the bias voltage, this socket is always 4-5 mv higher than the other. What can be causing this?

Sounds like you have eliminated everything except an unbalanced PI or unbalanced OT. Unbalanced PI is not causing a difference in idle current, but could be causing the redplating in the hotter tube when driven hard. Here's a simple method to test for each of those conditions without test equipment.

1. Disconnect the NFB wire and recheck voltage/current across your 1Ω cathode resistors (just for reference).

2. Temporarily reverse the OT plate leads. If the higher idle bias current/voltage moved to the other socket, the OT is slightly unbalanced. Either way, put the OT wires back like they were.

3. Temporarily reverse the signal leads from the PI to the output tube grids (pin 5). Now connect your attenuator and drive the amp hard. If the original power tube is OK now and the other power tube redplates now, the PI is slightly unbalanced. Either way, put the signal leads back like they were and reconnect your NFB wire.

4. That's all the testing.

How do you correct this unbalanced idle current and red plating when driven hard? Just tweak the bias pot to cause the amp to idle cooler. Check for redplating while driving the amp hard and readjust bias until redplating condition is OK. Don't worry about a 4-5mV difference. If the redplating is caused by an unbalanced PI, you should be able to correct by using another PI tube, or lowering the value of the PI plate resistor associated with the redplating output tube.

If it were my amp, I'd use a sig gen and scope to check the PI balance, but probably would not try to correct any unbalance, unless it was grossly unbalanced. Then I'd just set the bias cooler until the redplating was minimized. I'd do the same if the unbalance/redplating was caused by a 'not so perfect' OT. And I would not be concerned about a 4-5mV difference during idle.