EL84 pp blocking distortion

[ValveMan]

I'm trying to figure out why blocking distortion is clipping the top off the signal & clamping the input at only +5v when the bias is 11.8v (B+ is 308v).
The UL-configured EL84's are drawing only 75mA at quiescent.

The attached jpg of scope traces are from both sides of the same 68k stopping resistor. I'd increased that from 4k7 to 22k to 68k and although it helped, it hasn't fixed it.

The PI is a 12au7 cathodyne, drawing ~4mA. The signals are as clean & balanced as I could hope for.
The coupling caps are .047uF and the grid resistors are now 75k (were 150k)

Any suggestions are welcomed - thanks in advance!

[Chuck H]

The graphic you've shown has the entire waveform elevating (including the "0" line, which will actually be +11.8V) instead of the amplitude of the signal increasing. Is this what's happening on your scope? No signal amplitude, just a +V shift of the entire wave form?

Have you actually scoped the previous gain stages to be sure they are "clean and balanced" at the level you are testing?

Can you provide a schematic of your design?

Chuck

[ValveMan]

Thanks Chuck,
I'd moved the y positions on the scope to separate the pre-blocker & grid signals for the pic. In DC mode @ 2v/div the top clamps at +5 and the -ve side goes down to -10V still looking clean.
BTW - bias on the cathode rises to +14V at max volume, and the output from each tube squares off at 150v pk-pk

Previous stages are clean & balanced (input to both ends of the 84's are perfect mirror images).

Doesn't make any sense, so I'm suspecting the OT's gone bad . . . . I'll swap it out & see what happens

[R.G.]

Doesn't make any sense, so I'm suspecting the OT's gone bad . . . . I'll swap it out & see what happens

You could ... test... it to see if it's bad. If the windings are not open where they're not supposed to be, and not shorted together where they are not supposed to be AND it passes the internal shorts test, it's good.

Internal shorts test? Yes. It's easy. See the shorted winding tester at geofex.com.

[stokes]

If by "at quiescent",you mean at idle,75ma's is very high.Even for class A,if that is what you are going for at 308 plate volts you should be in the area of 35ma's.

[ValveMan]

If by "at quiescent",you mean at idle,75ma's is very high.Even for class A,if that is what you are going for at 308 plate volts you should be in the area of 35ma's.

Thanks Stokes.
that's for both tubes so it's ~37mA for each

[madkatb]

If you switch the output tubes does it clamp in the other direction?

[ValveMan]

If you switch the output tubes does it clamp in the other direction?

Nope - always at the +5V level on both sides

[Wombaticus]

What impedance load line are you running?
Cathodes are bypassed?

[ValveMan]

This is driving me nuts . . .

I need to step back . . . have a cold beer . . . & think about anything else for an hour.

This is a scratch build which was working nicely last week. I'd get a gain of ~19 from the EL84's with them maxing out at around 220Vpk-pk per side before saturating.

I tweaked a few things (gain stage, PI, tone stack) and put the PI and EL84's back just as they were before. Everything's running nicely again, except that the output from each 84 clips hard at 150Vpk-pk, causing the grid to clamp @ ~+5v

I overhauled the power supply (caps, choke and PT) in case something was amiss there; tried a known good OPT; swapped tubes out; checked for bad connections etc etc

Also tried dropping the B+ to 270V and the bias to 9V to make sure it stayed in the clean zone, but all it'd give me was a +/-50V on each side, still clamping the grid at ~5V.

I drew what appears to be the observed load line on top of Merlin's EL84 PP example on the attached jpg, to see if it tells me something (anything!!)

After I've had a cold one I'll check to make sure the UL leads are still properly connected to the screens. They're routed behind the board so it's conceivable something could be wrong with them.

If anyone has any ideas - please don't be shy . . .
Cheers

[ValveMan]

Sorry - forgot to point out my load lines are the thick green ones. Yes, they do look weird - I may have drawn them wrong.
The shift in current from idle (~35mA/side) to max voltage swing at +/-75V is an additional 10mA, so it's possible that one side pulls 65 while the other drops to 15mA?
I placed them the way they are based on the grid & plate voltages.

[Bruce / Mission Amps]

Go back to square one.
Is that what the signal at the EL84 grids look like with the scope connected to them?
With or without the actual power tubes installed?
What value are the plate, the two cathode resistors? I've seen old amps with 12AU7 type tubes in cathodyne, using really low values, like 12K to as high as 47k... etc.

I've had some really odd things happen to El84s with driver distortion driving their grids... and it doesn't take much grid signal to turn an EL84 inside out. I suspect anything over 25-35vac PP at the grids will start doing that.

[octal]

Bruce-
I believe that the OP said that one scope trace represented the drive signal from the PI and the other end is the actual output tube grid itself. (he's scoping on either side of the grid stopper resistors.) So it doesn't seem to be a PI problem.

Questions for the OP:

Is it possible that you have a measurement problem- i.e. your scope is broken, probes are on X10 or the variable sensitivity knob on your scope is cranked down making you misinterpret the scale?

Is it possible there's a problem with your dummy load or you've used the wrong impedance taps on the OT?

Have you measured the DC voltage at your screen grids? I think you mentioned that this is a UL amp, right?

Good luck,
Nathan

[ValveMan]

Thanks Bruce,

"Go back to square one."
I'm going to take your advice on that & start a fresh build - get back to having a totally clean layout etc.

"Is that what the signal at the EL84 grids look like with the scope connected to them?
With or without the actual power tubes installed?"
With the EL84's installed, those are the traces taken from both sides of the stopping resistors.
With the tubes out, the signal stays clean way over 10Vpk-pk
I'd actually be happy if I could get the gain of 20 back that I had earlier - I'd like to give it only 12Vpk-pk and get a 240Vpk-pk from each tube

"What value are the plate, the two cathode resistors? I've seen old amps with 12AU7 type tubes in cathodyne, using really low values, like 12K to as high as 47k... etc."
The EL84 has a shared 115R (bypassed with a 470uF cap), giving me a bias of 10.5V now, with 307V on the plates
I'm using a Hammond 1650F which is rated at 25W and has a 7600ohm a-a impedance

BTW - although the cathodyne was working nicely with 22K load resistors, I switched back to the LTP I had before, still using the 12AU7. Balance is pretty good & the signals are clean & symetrical

"I've had some really odd things happen to El84s with driver distortion driving their grids... and it doesn't take much grid signal to turn an EL84 inside out. I suspect anything over 25-35vac PP at the grids will start doing that."
Yeah - I'm with you on that, but I'd be happy just to give it 12vac PP (each tube)

I'll see what happens with the fresh build - Thanks!

[ValveMan]

Thanks Nathan,
You're right on the PI piece.
Measurements are good - yes, I'm using 10x probes, but I've been careful to set it up right. Also checked with a Fluke 75 dvm (non-RMS) to confirm AC and DC levels.

It's the same whether I'm using the dummy load or connected to the speaker. I double checked the connections from the 1650F & also tried using a another in it's place - still the same. It's possible I got a bad batch with the wrong colored leads - but like I said - at one point, I was getting a 220v PP from each side with a 11.5v PP signal to each side.

Just checked the DC levels on the screens - they're actually 2V higher than the plate (both idle & working) - could that be a problem?
Cheers
Gerry