Cascoded Long Tail Pair Inferno!

[panos29]

After having much trouble building a long tail pair splitter driving differential as a general use building block for PP amps. I ended up in a cascoded long tail pair around 6N6P double triode, using the included schematic.

The circuit with 10Vpp SE input (one side grounded) gives 150Vpp at each anode or a total of 300Vpp from anode to anode when loaded with 47K parallel with 220pf. Square is excellent to at least 30KHz as well as at 20Hz.

The problem is that when I load this with the output stage (2A3 PP). Even with no HT applied to the anodes of 2A3s (only filaments) the driver is clipping the upper side of the output waveform heavily but only when measured from anode to ground. When I measure this differentially (from anode to anode of 6N6) the distortion is obviously cancelled as is at the output of the amplifier. Noteworthy is the fact than when the driver is loaded with the ouput tubes the clipping is evident even at very low signals (i.e 5Vpp output). This gives me hell! can you propose whats happenning, or is this the way it supposed to be working(very unlikely)???
Any help would be appreciated

[ilimzn]

Originally posted by panos29
with no HT applied to the anodes of 2A3s (only filaments) the driver is clipping the upper side of the output waveform heavily

Sounds like diode action here from G to K, coupled with leaky coupling caps?

[panos29]

I thought that might this happened so I changed the coupling caps with new ones as well as the output tubes, but no change at all.

[EC8010]

Is the grid bias present on the 2A3? If there isn't a significant difference between grid and cathode voltage but the heaters are on, then positive excursions of the differential pair will be driving into a thermionic diode connected to ground.

[panos29]

Yep, the bias is present. Even with the output pairs fully working the symptoms are present! The bias is o.k and its cathode bias through 400ohm resistors bypassed with 1000u.

[EC8010]

I suggest that you DC couple your oscilloscope, set both channels to the same sensitivity, shift both traces to the centre line and put one probe on the grid and the other probe on the cathode of one 2A3. Increase the signal amplitude until the problem occurs and look at what the scope tells you.

[kevinkr]

What is the working current in your driver stage? From your description it sounds to me like you may have insufficient current drive to handle the small amount of grid circuit current present. (See third paragraph - if this is not the case you may have an inappropriately chosen operating point.)

Note that the positive waveform will clip when there is no plate voltage present because in a cathode (auto) bias amplifier there is no bias present either and you are essentially driving a shunt connected diode clipper to ground which is forward biased on positive going peaks.

When you apply plate voltage this should only happen once the positive peaks exceed the effective negative bias which should occur at something like +40Vpk at which point the output tube is effectively saturated unless you intended operation in A2 or AB2. If you can swing this type of voltage at the output of the driver before clipping you really don't have a problem, above this level the output tube grid will start to look like a diode and clip the waveform.

edit: add a couple of extra thoughts...

Kevin

[panos29]

So, I tried what suggested from EC8010 and the cathode remained a straight line moved to about -40V while grid got all the available swing. What I noticed though is that there is present a strange waveform at the output of the driver when no signal is present at the input. There is about 1Vpp trigonal wave like a hum or buz but at the top its becoming very complex like RF. When I ground the input the trigonal wave becomes clean and the "rf" is gone. I will try to reduce the grid leak of the input to 100K from 470k. So Kevin, you are sayng that when the drive signal to 2A3 becomes more than 40Vpp the signal will be rectified as the overdriven 2A3 grid will act as diode?

[EC8010]

Increase your grid-stopper resistors from 100 Ohms to 1k and make sure they are right on the valve pins. In this very specific case, carbon resistors are better than metal film. Also try a 100n plastic capacitor from the top of the anode resistors to 0V and also one across the negative bias supply. The gate-stopper on the FET might be a bit small too, I often use 1k.

And yes, we are saying the 2A3 acts as a diode.

[kevinkr]

Yes, precisely as EC8010 indicated... And this is normal behavior..

Kevin