Strange tube blowing issue

[catalin gramada]

Hello
I had a problem with a 6550 power tube. The tube in discussion was suspected from couple of days having grid leak issues because supplied from same negative source draws 10mA more current than the rest. I checked everything in supply chain and even I changed the coupling cap despite the fact the first one shows me no leaks. It was continued to draw more current than the rest. I changed its position in other socket and somehow, I don't know why it shows me draw the normal current I expected.
...till today when starting the amp the fuse in PT primary blown. Checked the circuit shows the 150K grid leak resistor was blown in two half. Ok but why my 6A in primary winding is over ? So checking further found the 100 ohm resistors in heater balancing circuit are blown the same with a big spot of carbonized rests allover around.
May someone enlighten me why the heaters circuit was affected please ? All my power tubes are individual RC coupled biased from a common negative rail with dual pots split.. thanks.

Late: screen grid resistor is carbonized at one end, and even the plate wire is burned at the socket...

[Mick Bailey]

I've had situations where a tube has shorted internally to the heater. With some Marshalls that have a DC heater supply the bridge rectifier is destroyed. Mostly though with regular amps with AC supplies the 100R resistors burn out or if a humdinger pot is fitted the track vapourises. This type of short can draw a lot of instantaneous current and blow the mains fuse due to high voltage appearing across the heater winding.

[Helmholtz]

Are you aware that the max. grid circuit resistance of a 6550 is 50k?
If the grid leak is much larger than that and a tube has unsufficient vacuum it can go into thermal runaway.
This means that the grid voltage continues getting more positive until something blows.
If the grid leak is even shared by paralleled tubes, also these will increasingly draw more current.
A tube with a poor vacuum is likely to show an increased idle current.

This said, it is suspicious that the idle current changed with the socket.
So there might be something else involved.

[catalin gramada]

Thanks. Just saw the tube heater is kaput. Have no continuity.

[catalin gramada]

Does a quick measurement over grid stoppers to be a good indicator about grid leakage please ? I measured couple of el34 I have at roughly biased 70 percent. I get everything in between 0.6 and 10 mV steady DC. over 1.5K stopper and bit of bias drift as consequence. Grid leak 150K. Individual bias may solve with dc settings but I wonder how much dc grid current is to much to be safe for circuit please ? I didn't establish a relation between grid current and warming but is clear for certain tubes grid current rise with tube temperature then stop fixed to certain value. I will try tomorrow to cool the tubes for test as I did all measurement on chassis with the tubes upside down, if this somehow count...Thanks.

[Pedro Vecino]

Are you aware that the max. grid circuit resistance of a 6550 is 50k?

That's right. I think some of the confusion with the "typical" bias feed resistors on the 6550s comes from the values Marshall used on the amps exported to the US (150K). I guess this really depends on the condition of the circuit as I've had a few of these amps in mint condition after forty years or so with their original tubes. Another thing to note is that these were GE 6550s. There is no better one.

[Helmholtz]

I guess this really depends on the condition of the circuit as I've had a few of these amps in mint condition after forty years or so with their original tubes.

I think it mainly depends on individual tube quality (vacuum, getter, purity of materials) as well as bulb temperature.

Positive grid leak current will show as a positive voltage drop across the grid leak resistor (i.e. grid side more positive).

[Pedro Vecino]

I think I explained myself wrong. I'm referring to the conditions that the circuit establishes for the operation of the 6550 in that case.
I've always wondered about the small difference in the adaptation of the Marshalls with EL34 to work with 6550 and it was just a couple of months ago when, following the data sheet, I checked the recommendation of not exceeding 50K in those resistors.
I think they didn't go to that extreme because of the need to avoid further changes that would turn it into something problematic. I want to think that in those circuits it was a compromise value for an adaptation as simple as possible.

[Helmholtz]

Well, lowering grid leak resistance per tube to 50k would significantly reduce available PI drive and gain.

[catalin gramada]

Thanks
My statement regard switching tube position seems wrong.( Maybe the tube was not warmed enough at the time I picked value). Just I shifted a tube with same problem of derating grid current in the same chassis. There are 6 tubes and the actual position not allow a very a good ventilation. Still is a temperature difference from the middle to the corner of chassis - 10*C difference measured at the middle of the glass bulb (115*C vs 105*C). This temperature is pretty stable after 1 hour running , but still not did any diference in grid current of the tube I followed and moved around, or if it did is minimal ( 10.6 - 11.2 mV dc)
I hoped putting the tube in a cooler place of chassis may help to reduce the leakage, but keeping coller with minimal 10*C on the bulb do not did differences.

[g1]

I've always wondered about the small difference in the adaptation of the Marshalls with EL34 to work with 6550 and it was just a couple of months ago when, following the data sheet, I checked the recommendation of not exceeding 50K in those resistors.

It seems to me that they exceed the maximum for those resistors in some of the EL34 versions as well, can't recall right now if it was for 50W or 100W amps.

[catalin gramada]

I have just one more question please: May a bias sense resistor with a certain power to be an effective protection to limit a failure just to the tube responsible for that ? .... heaters circuit damages is already to much...

[Helmholtz]

EL34 grid circuit limit is 500K. Gives some freedom.
But 6L6/5881 limit is only 100k, so look at Fender amps.
In principle grid circuit resistance should be halved when 2 tubes are paralled like with 100W amps.

[Helmholtz]

I have just one more question please: May a bias sense resistor with a certain power to be an effective protection to limit a failure just to the tube responsible for that ? .... heaters circuit damages is already to much...

Maybe consider individual cathode fuses?
I don't trust resistors being used as fuses.

[g1]

In principle grid circuit resistance should be halved when 2 tubes are paralled like with 100W amps.

This is what I was thinking of. It seems to me there was a Marshall model that used 470K or 500K for both the 50W and 100W versions.