While biasing a Fender Tremolux (AA763) the biasing is current is increasing?

[Jazz P Bass]

I would suggest that you get another set of output tubes.
The one that is glowing orange is bad.

[MWJB]

Both sockets drop to -13vdc on pin 5, or just always the socket with the bad tube? If the symtoms follow just the tube, then start again with "known good" tubes.

AB763 is not cathode biased by a common cathode resistor at the power tubes.

[Dandrix]

thread #6 gave you the that info, started with tung-sol then moved to the TAD's
once again, what next that i'm missing here?
is this a mystery and this never comes out on top?

[MWJB]

Take out the tube that redplates, play the amp with the tube that does not redplate, first in one socket, then the other...what happens?

If you redplaste a tube, it will be prone to further redplating.

A tube redplates because it is underbiased (not in this instance), or you have a failure in the bias supply that causes the negative voltage to decrease (this is what we are trying to find out), or because you have a bad tube...or two (try some good tubes).

You have in this thread all in information you need, but you are jumping from one thing to another without fully eliminating the last. You need to be more thorough & methodical.

[Dandrix]

dear mwjb,
please read the threads, i have said all along that i have swapped out the tubes. thread #6, #22. we can rule out the bad tube because, i'm using the the tung-sol tubes in my super reverb right now. it's loud, great bass bla bla bla. it red plated with the tung-sol tubes first, then the TAD's and that were i'm at now. BOTH SET I BOUGHT IN THE BEGIN OF THIS YEAR 2011. i don't have to buy an other set... i will try the one tube at a time idea next. thank dan

[riz]

Check if the fiberboard has gone conductive.

[MWJB]

So now you know it is the bias supply losing voltage, look for dropping resistors that could be drifting upwards in voltage, or a dodgy bias pot, even if the problem affects both tubes, it's likely that one will redplate before the other.

[trebazond]

Hi all,
I wonder if you have sorted this Dandrix?
I have the selfsame problem I believe, with a 6G9B Tremolux that has had the power transformer replaced. I don't have to play it for the bias current to run away, any valves other than the original Mesas show the problem.
I am of the opinion that this is a grid leakage current problem made worse by the B+ voltage being higher than originally designed for. The schematic shows a B+ of 365V on load, but new valves don't draw sufficient current to drop it that low with the given fixed bias voltage and the new transformer.
With the Tremolo Intensity at maximum, we have 220K from each grid and then the 250K intensity pot before the low-impedance bias supply voltage. The intensity pot carries the leakage current from both grids so effectively drops double voltage.
Any grid leakage current acts to make the grid voltage lower in magnitude (less negative) which produces more anode current in the 6L6s, which in turn causes more grid leakage current and so leads to runaway. The grid resistors are too high I think (spec for 6L6s is 100K in one datasheet I have seen) especially with that intensity pot in there, but I guess it worked with the valves and transformer it was designed with originally.
Thoughts? On reflection Dandrix's problem sounds more like RF oscillation caused by the lack of grid stopper resistors in the design, but it could be related.

[MWJB]

220K grid loads for the 6L6 are fine. Plate voltage is fine if under about 530vdc.

Replace the 33K fixed resistor with a 50KL cermet pot wired asa variable resistor, in series with the 33K and dial in the the negative voltage to -65vdc at the junction of the 220K grid loads (you might get away with less, but you don't state your plate voltage), no power tubes installed. Power down, fit the power tubes and powr up, fine tune the pot to give 30-35mA at each 6L6 plate.

The 6G9B runs the tubes fairly hot for a fixed bias Fender, as you increase plate current then this will get worse, you need to alter the bias voltage divider to increase negative dc at pin 5/220K grid loads. You can, of course reduce the value of the 100K resistor that taps the rectifier winding too if necessary (also raises negative bias voltage).

[trebazond]

Hi MWJB,
modification looks inevitable, though I hate to do it to a customer's amp. Your suggested mod makes setting the initial bias point easier (though the bias supply cap is only rated at 50V - I've already upped this one to 63V) but I'm not sure it will stop the runaway. I'm having a 50% success rate with valves so far, the initial condition is 40-45mA at 370-385V but half of them eventually run away, even if it takes half an hour, with the intensity pot at max.
If I'm going to do mods I may as well go the whole hog - drop the source impedance of the bias supply from its current 25k-ish (the new transformer has a bias winding so should be straightforward), make it variable and maybe find I way to get the intensity pot out of the dc bias path altogether. But I might also change those 220ks for 100ks.
I have to get the amp in a state where the next person to replace the valves doesn't get deceived into thinking that the biasing is ok when in fact they're going to have a fire after an hour's use.

[Enzo]

Please don't redesign the amp. If it needed that, ALL of them would have this problem.


SO one of your tubes red plates, or both? Is it always the same socket? Both sockets?

SO what voltage is on the plates at first, and when they go nuts what voltage is there? More importantly, what is the starting voltage on the control grid? And once it goes red, what is the voltrage on each control grid?


I don't buy your theory of grid resistors being too large. If that were the case, then the tubes don;t red plate with the intensity at minimum but do at max? And if the intensity control does affect the redplating, my first suspect would be a leaky 0.1uf cap from the trem driver, rather than grid resistors.

But you can test for that even. The right end of the intensity control is at the bias voltage. Looks like -36, but whatever yours is, note it exactly. Now what exact voltage is on the left end of that control? Do both readings with the control set at each end, four readings. We'd expect the voltage across that pot to be steady, since no current should be flowing.


And if the bias supply is steady, but the bias voltage at the grids is falling, then either the tubes are doing it - and you can pull them and find ot - or the 0.1uf coupling caps from the phase inverter could be leaky.


And do this. Ground your meter to chassis as usual. Now find those 0.1uf caps to the PI and the 220k grid return resistors that all conect to pin 5 of each tube. Find them on the eyelet board, wherever the wires that go to pin 5 connect. Now touch your red meter probe to the eyelet board itself, a couple millimeters away from either of those eyelets. Get a DC reading? Yeah, I know the eyelet board is insulating material, but search up conductive eyelet board or conductive part board in Fender amps and read the discussions. If you find any voltage on the eyelet board material, there is your problem.

[MWJB]

Trebazond, you mention 40-45mA at 370-385vdc. Note the idle current should be set with the tremolo switched off, not just turned down on the intensity pot.

If your plate voltages are still in this region, with the bias set as above, you shouldn't need as much negative voltage as I initially thought, maybe -45 to -50vdc (before fine tuning by mA).

The plate voltages still seem a little low, can you identify the new power transformer, a typical Tremolux/Vibrolux drop in would normally have more voltage.

These bias vary trems tend to have more impact biased a tad cooler....so bias by ear, if the owner is intending to use the tremolo.

As Enzo says, I don't believe grid resistor value is relevant to this issue.

[trebazond]

Hi guys,
I deliberately went for a power transformer with a lower B+, since a so-called 'drop in replacement' produced voltages off-load of over 460V, threatening to blow the main smoothing caps (and the bias voltage was over 50V as well). So I won't name names.
I have of course disabled the tremolo. Those 220k grid resistors were in fact 245k and 257k, and the 250k intensity pot is actually 290k. But Enzo has hit the mark with the conductive fibreboard idea - I was getting a couple of volts sticking the probe firmly in. That also explains why the problem was different from day to day as when I was soldering and desoldering components, I was drying out the area round the eyelets. It also probably explains why (bizarrely) pulling out the tremolo oscillator valve alters the bias condition of the phase splitter - they are unconnected in the circuit, but next to each other on the fibreboard.
Now to bake the amp for a few hours.
You two will know this, but for the benefit of others I discovered that even a x10 scope probe can't be left attached to the output valve control grid without significantly changing the current. Measurement effect.

[MWJB]

Good luck with the baking, be aware that boards can go conductive in the sense that the lower board can also affect the circuit, I'd take the multi pronged approach and either try and add some non conductive material between the fibre boards, or sometimes spacing out the top & bottom boards with grommets can solve isolated issues....the only real fix, if this & baking don't work, is replacement of the boards & circuitry.

"I deliberately went for a power transformer with a lower B+, since a so-called 'drop in replacement' produced voltages off-load of over 460V, threatening to blow the main smoothing caps (and the bias voltage was over 50V as well). So I won't name names." ~460vdc off the rectifier is pretty much what I'd expect from a Tremolux/Vibrolux PT, some are more than this (I have seen 355-0-355VAC). Even with the lower voltage Brown/Blonde Fenders I use 500v rated power supply caps as default, 100v rated bias supply caps. I tend to rewire the main B+ nodes as series totem pole caps using 2x100uf 350v caps in series, each cap bypassed with a 220K 2W resistor (see AB763 Super Reverb for example) though for B+ voltages of less than 450vdc this is just a matter of preference. Old Fenders typically run more than the voltages quoted on the schems, 500vdc on an amp quoted as 445vdc is perfectly normal (in which case, often I rewire the screen node cap as atotem pole too with 2x47uf 450v caps in series, again each one bypassed with a 220K 2W).