JMP 2203 eats EL34s

[blackbird]

This Marshall JMP 2203 100 Watt head manufactured in 1979 is burning through EL34s. The 1 amp HT fuse doesn’t blow, although the 4 amp fuse blew when the amp ate the first two, V5 and V6 (one push – one pull).

There are no visible signs of arcing or burning anywhere, on anything, and no capacitor leakage. The only previous repair was a new AC cord.

All power supply filter caps and power resistors measure to within 5% of spec.

I replaced the screen resistors (1K / 5W) on (V5 & V6) Old ones were at 980 ohms. Also replaced one of the Voltage drop resistors to lower the preamp voltage supply. (DC voltage on plates of V1 – V3 were 10 to 15% too high).

Also replaced two coupling caps on phase inverter plates (V3) with .033uf 630V caps.

Voltage readings in power amp with tubes installed and Variac at 120Vac:

EL34s (V4 – V7): Pin3 = 453Vdc, pin 4 = 450Vdc, pin 3 = -39.8Vdc, pins 2 & 7 = 3.2Vac each.

Bias supply output is -58Vdc.

Output transformer is 15.5 ohms on one side (V4 & V5) and 14.2 ohms on the other side (V6 & V7). Choke tests at 105.3 ohms.

The last EL34 blew while attempting to read the voltage drop on each side of the O.T. to calculate the plate current. I saw .83V on the V6 & V7 side. So .83V divided by 14.2 ohms = 58 milliamps, or 29 milliamps per tube. A little low. But when I went to read the voltage drop on the V4 / V5 side of the O.T., a chattering sound occurred. I withdrew the meter probe immediately and shut her down. V5 was blown.

All the tube sockets had previously been cleaned and inspected for burns, cracks, etc. All I could see were a couple of small chips out of the plastic next to pins 2 & 7, the heater. These chip-outs do not extend to adjacent pins. And no burns or cracks,

Has anyone encountered this problem before? What the heck is wrong?

Any insights would be greatly appreciated! Thanks in advance.

 

[Pete Farrington]

My experience is that probing output valve anodes is asking for trouble, eg it can easily invoke oscillation. That’s probably what the chattering sound was.

I suggest to fit a 1R cathode current sensing resistor for each EL34. Then you’ll know exactly what HT current each valve is drawing, and measure voltage at the rectifier output / reservoir cap / HT fuse node. The V DC there will be negligibly different to the anodes.

 

[Spanngitter]

How do you determine bad EL34? Just by the blown main fuse?
Your G2 Voltage is on the upper end for EL34 (450Vdc) so I would anticipate that your max Power Dissipation on the Grid (8W) gets exceeded and your poor EL34 suffers internal melt down, producing a Grid/Heater or Grid/Suppressor Shortage (and that's the reason the main fuse blows, if if shorts to the Anode your HT would go)

 

[blackbird]

My experience is that probing output valve anodes is asking for trouble, eg it can easily invoke oscillation. That’s probably what the chattering sound was.

I suggest to fit a 1R cathode current sensing resistor for each EL34. Then you’ll know exactly what HT current each valve is drawing, and measure voltage at the rectifier output / reservoir cap / HT fuse node. The V DC there will be negligibly different to the anodes.

I have always measured voltage drops on output transformers that way (center tap to anode, and this is the first time I've heard that chattering sound.
What is a 1R resistor - 1 ohm? And how does that tell me the plate current?
Thanks again.

 

[blackbird]

How do you determine bad EL34? Just by the blown main fuse?
Your G2 Voltage is on the upper end for EL34 (450Vdc) so I would anticipate that your max Power Dissipation on the Grid (8W) gets exceeded and your poor EL34 suffers internal melt down, producing a Grid/Heater or Grid/Suppressor Shortage (and that's the reason the main fuse blows, if if shorts to the Anode your HT would go)

I determine the EL34 is bad with a B&K 707 tube tester.
I've seen many amps with plate and screen voltages very close, like this one, which shows 453V on the plate and 450V on the screen, and none of them were blowing EL34s because of it.
The main 4amp fuse only blew once - the first time - when two EL34s blew, one on each side of the amp.

 

[PelliX]

I have always measured voltage drops on output transformers that way (center tap to anode, and this is the first time I've heard that chattering sound.

Some circuits are more susceptible to oscillating than others.

What is a 1R resistor - 1 ohm? And how does that tell me the plate current?

Yeah, 1R is typically a 1 Ohm resistor - read on. With the resistor attached as Pete suggests, you can use a determine the current, because the other two factors are defined. Voltage = (current * resistance), so if you know the Voltage (you do) and the resistance (now you do) then you can determine the current.

I determine the EL34 is bad with a B&K 707 tube tester.

So, what, transconductance is way down after the amp 'eats' them? Or do they develop some kind of internal short?

I've seen many amps with plate and screen voltages very close, like this one, which shows 453V on the plate and 450V on the screen, and none of them were blowing EL34s because of it.
The main 4amp fuse only blew once - the first time - when two EL34s blew, one on each side of the amp.

Not all EL34's are produced equal. Typically 'older production' valves will withstand greater voltages, but that's a rule of thumb. Some amps need tweaking a little to accomodate for current production. What are we talking about here?

 

[NickkiC]

1 ohm resistor on cathode to ground (8&1) for each tube. Read current of this and can act as fuse.

 

[Pete Farrington]

The last EL34 blew while attempting to read the voltage drop on each side of the O.T. to calculate the plate current.

Just to confirm, did you have a load connected to the amp, eg a speaker cab, resistive load?

1 ohm? And how does that tell me the plate current?

Current passing via the anode has to pass via the cathode.
Cathode current = anode current + screen grid current.
Screen grid current can easily be assessed, as it flows via the 1k (1,000 ohms) screen grid current limiting resistor.
1V across the resistor = 1mA.
Typical value for idling EL34 is 5mA (one thousandth of an ampere), ie 5V across the 1k screen grid resistor.

35mV across a 1 ohm cathode resistor means that cathode current is 35mA, anode current will be close on 30mA.

The significant advantages of measuring at the cathode is that due to the minuscule voltage, it’s much less hazardous, and the risk of inducing oscillation is practically 0.

 

[blackbird]

Some circuits are more susceptible to oscillating than others.



Yeah, 1R is typically a 1 Ohm resistor - read on. With the resistor attached as Pete suggests, you can use a determine the current, because the other two factors are defined. Voltage = (current * resistance), so if you know the Voltage (you do) and the resistance (now you do) then you can determine the current.



So, what, transconductance is way down after the amp 'eats' them? Or do they develop some kind of internal short?



Not all EL34's are produced equal. Typically 'older production' valves will withstand greater voltages, but that's a rule of thumb. Some amps need tweaking a little to accomodate for current production. What are we talking about here?

On the first occasion, after being eaten, V5 and V6 were shorted internally. The second time, V5 turned white on top and showed ZERO transconductance. The third time, V5 showed ZERO transconductance.
And that's true that EL34s and E34Ls vary significantly in hardiness.
Still, its always (3 times) V5 that blows, and once V6.

 

[blackbird]

Just to confirm, did you have a load connected to the amp, eg a speaker cab, resistive load?

Current passing via the anode has to pass via the cathode.
Cathode current = anode current + screen grid current.
Screen grid current can easily be assessed, as it flows via the 1k (1,000 ohms) screen grid current limiting resistor.
1V across the resistor = 1mA.
Typical value for idling EL34 is 5mA (one thousandth of an ampere), ie 5V across the 1k screen grid resistor.

35mV across a 1 ohm cathode resistor means that cathode current is 35mA, anode current will be close on 30mA.

The significant advantages of measuring at the cathode is that due to the minuscule voltage, it’s much less hazardous, and the risk of inducing oscillation is practically 0.

Yes, always with a speaker cab for a load. 8 ohms in this case. And I will install those 1 ohm 3 watt resistors on the cathodes., and read the voltage there. Thanks for that. Just ordered them today.

 

[playloud]

Perhaps this goes without saying, but I would measure across the cathode resistors for a while (using alligator clips). There could be something intermittent happening.

I had a similar issue and it turned out to be a dodgy resistor in the bias circuit. In that case, I monitored the bias voltage on pin 5 with the meter on the 'max' setting for several mins (noting that the voltage is negative). Eventually a sudden jump revealed the source of the problem.

 

[PelliX]

On the first occasion, after being eaten, V5 and V6 were shorted internally.

That's seriously worrying, it would mean that the internal insulation of the valves has been destroyed. Yikes. Presuming they were working normally before, of course.

The second time, V5 turned white on top and showed ZERO transconductance.

If the getter flash is white, the valve has gone to air. I've never seen that occur due to anything the circuit would do to it, but it can cause a chain reaction.

The third time, V5 showed ZERO transconductance.

This wasn't the valve that went to air, right?

And that's true that EL34s and E34Ls vary significantly in hardiness.
Still, its always (3 times) V5 that blows, and once V6.

There's got to be something at those sockets... I suggest you follow @playloud 's advice here.

 

[yafal]

This Marshall JMP 2203 100 Watt head manufactured in 1979 is burning through EL34s. The 1 amp HT fuse doesn’t blow, although the 4 amp fuse blew when the amp ate the first two, V5 and V6 (one push – one pull).

There are no visible signs of arcing or burning anywhere, on anything, and no capacitor leakage. The only previous repair was a new AC cord.

All power supply filter caps and power resistors measure to within 5% of spec.

I replaced the screen resistors (1K / 5W) on (V5 & V6) Old ones were at 980 ohms. Also replaced one of the Voltage drop resistors to lower the preamp voltage supply. (DC voltage on plates of V1 – V3 were 10 to 15% too high).

Also replaced two coupling caps on phase inverter plates (V3) with .033uf 630V caps.

Voltage readings in power amp with tubes installed and Variac at 120Vac:

EL34s (V4 – V7): Pin3 = 453Vdc, pin 4 = 450Vdc, pin 3 = -39.8Vdc, pins 2 & 7 = 3.2Vac each.

Bias supply output is -58Vdc.

Output transformer is 15.5 ohms on one side (V4 & V5) and 14.2 ohms on the other side (V6 & V7). Choke tests at 105.3 ohms.

The last EL34 blew while attempting to read the voltage drop on each side of the O.T. to calculate the plate current. I saw .83V on the V6 & V7 side. So .83V divided by 14.2 ohms = 58 milliamps, or 29 milliamps per tube. A little low. But when I went to read the voltage drop on the V4 / V5 side of the O.T., a chattering sound occurred. I withdrew the meter probe immediately and shut her down. V5 was blown.

All the tube sockets had previously been cleaned and inspected for burns, cracks, etc. All I could see were a couple of small chips out of the plastic next to pins 2 & 7, the heater. These chip-outs do not extend to adjacent pins. And no burns or cracks,

Has anyone encountered this problem before? What the heck is wrong?

Any insights would be greatly appreciated! Thanks in advance.

"...I replaced the screen resistors (1K / 5W) on (V5 & V6)..."
Hi, this could be a long shot, but what if those resistors were wrongly labeled...? It happens! I would go with the original ones, nothing wrong with them at 980 Ohm reading. At least, check the new ones and recheck if you by chance welded them wrongly (sorry!)

 

[Dirty-D]

Replace them with KT88's 😀

 

[Pete Farrington]

I had a similar issue and it turned out to be a dodgy resistor in the bias circuit. In that case, I monitored the bias voltage on pin 5

Just to note that the loading of the meter, combined with the high resistance in that circuit, will act to reduce the magnitude of the bias voltage applied to the valve, causing the valve to idle hotter.
eg assuming about 10k bias supply output resistance, 220k grid leak, 5k6 grid stopper and 10M meter loading
10M / (10M + 235k6) = 0.977
So -40V bias at the control grid will be pulled down to, and measure, -39V. So the valve will idle slightly hotter.

But if the meter input resistance is 1M (check the manual), the loading reduces the voltage by a factor of 0.81, with -40V bias now being pulled down to, and measuring, -32.4V. so the valve will idle much hotter.

Thankfully most meters are 10M rather than 1M loading.

Whatever, to measure bias voltage, it’s better to probe the ‘V’ node of the 220k grid leaks. To assess any voltage drop issues between that point and the control grid, leave one probe on the V node and put ghe other probe to the control grid terminal. A typical output valve will drop a few tens of mV across the grid leaks, due to grid current.

Another hazard with probing a control grid is inducing oscillation, which will tend to affect DC voltages (because the circuit is oscillating rather than idling).
That can be mitigated by probing at the grid leak-grid stopper node, rather than directly at the control grid terminal.

 

[playloud]

Just to note that the loading of the meter, combined with the high resistance in that circuit, will act to reduce the magnitude of the bias voltage applied to the valve, causing the valve to idle hotter.
eg assuming about 10k bias supply output resistance, 220k grid leak, 5k6 grid stopper and 10M meter loading
10M / (10M + 235k6) = 0.977
So -40V bias at the control grid will be pulled down to, and measure, -39V. So the valve will idle slightly hotter.

But if the meter input resistance is 1M (check the manual), the loading reduces the voltage by a factor of 0.81, with -40V bias now being pulled down to, and measuring, -32.4V. so the valve will idle much hotter.

Thankfully most meters are 10M rather than 1M loading.

Whatever, to measure bias voltage, it’s better to probe the ‘V’ node of the 220k grid leaks. To assess any voltage drop issues between that point and the control grid, leave one probe on the V node and put ghe other probe to the control grid terminal. A typical output valve will drop a few tens of mV across the grid leaks, due to grid current.

Another hazard with probing a control grid is inducing oscillation, which will tend to affect DC voltages (because the circuit is oscillating rather than idling).
That can be mitigated by probing at the grid leak-grid stopper node, rather than directly at the control grid terminal.

Interesting and useful info. Thanks

 

[ginmoon]

This Marshall JMP 2203 100 Watt head manufactured in 1979 is burning through EL34s. The 1 amp HT fuse doesn’t blow, although the 4 amp fuse blew when the amp ate the first two, V5 and V6 (one push – one pull).

Has it worked properly since you’ve owned it? Does it have grid stoppers on the power tubes? I added 10ks to my 71 SB that had similar problem and that solved it (along with running with step down transformer at 115v).

 

[myersbw]

How about tubes? Did you install a new set or are you buying "NOS" that might actually be old "OS" that are just reading good on your meter? Are you using an attenuator to push the EL34's? Or, just relying on a master volume in the amp? I'll assume the latter with that amp...?