runaway current draw on new build

[luthierwnc]

Hi All,  I'm trying to put the finishing touches on a NOS build and came across something I haven't seen before.  The amp is along the lines of the T-Lite using a pair of 7591 outputs.  At idle, I have the power tubes biased at 32ma at 420VDC on the plates (using the shunt test).  That's at a bias voltage of -22VDC.  Other voltages at the nodes are about where I expect.  But as soon as I put any signal in the amp, the meter pegs the needle and the plates drop 70 volts in runaway. 

Any ideas what to test would be appreciated.  Thanks, Skip

[sluckey]

I have the power tubes biased at 32ma at 420VDC on the plates (using the shunt test).

Probably a good idea to disconnect this meter when you apply a signal.

[PRR]

> using the shunt test..... as soon as I put any signal in the amp

The meter is likely a dead-short for audio. Any signal, current wants to go to infinity.

If you truly need with-signal DC current (or just like living), put a low-value resistor in the cathode return and meter Volts there.

[luthierwnc]

I'm pretty sure I have a 2w 1R resistor in the bin and will solder it in tomorrow.  Still, a 70 volt drop isn't a measurement error.  Be back soon, sh

[PRR]

> a 70 volt drop isn't a measurement error.

No. It is the expected result when an amplifier is driven into a near-short.

[luthierwnc]

I only have one 1-ohm resistor and got .071 VDC on both tubes at idle.  IIRC, that's the same draw I got with the shunt method adding the two tubes together at around 34 ma each.  With the volume turned up to 3 using a sine wave and an 8-ohm dummy load the voltage increased to 251 ma.  I didn't leave it there long.  I'm not using the overdrive or the reverb tubes right now -- just the first two gain stages and the PI(both 12AX7's).  SS rectifier . 

Now I just need to trace the gremlin.  Thanks for any ideas, Skip
 

[sluckey]

Your tubes are idling just fine. But I suspect your test signal may be unrealistically high. Your sine wave input should only be about 0.1V peak to peak. Just hook up a speaker and plug a guitar in. What does that sound like?

[Dave]

If you can see the negative voltage at the grids and the tubes are biased correctly at idle, it is probably fine. Play it. See what it sounds like. It is very unlikely that a guitar signal is going to somehow nullify the negative voltage on the grids. However, it is possible that your power tubes are bad. If they are, you will probably see them red plating. If that happens, the first thing I would do is double check bias voltage and then try a different pair of tubes.


Dave

[luthierwnc]

Sluckey;  Not too bad.  I don't think it is a sound I can use but it isn't woeful.  Noodling on a strat the plate volts were 408 and all the preamp tubes were where I wanted them.  I haven't changed the output setting on the signal generator in 25 years but I understand 7591's need a lot less drive.  I've got 8k2 resistors on the grids and 1k/5watters on the screens.  I'll order a couple more exact 1R resistors before I sew it up.  I also think I'll reduce the treble cap in the stack since it seems a bit middy.  Thanks all and good luck with your projects.  sh
Dave; brand-new tubes within a ma of each other.  Bias voltage is around -22ma at the 220k junction. 

Thanks gents and good luck with your projects, sh

[luthierwnc]

Morning update:  Yes, I must have been driving things too hard.  I got a Rolls signal generator from Lord Valve 25 years ago and have just left it at 5db.  To get the input down to .1v it has to be dialed a shade under 2.  Waveforms have cleaned-up nicely.  The wave into the PI gets ugly (cut off top, dimple at the bottom) with the master dimed so I will bleed some of that to ground.  There is also a .022uf cap there and that might get a hair smaller too.

As always, thanks for everyone's interest, cheers, Skip