Balanced PI Tube

[Craig B]

I have a Vibrolux AB763 with 12AT7 PI tube and I want it to be as clean as possible. I have read about using a balanced driver tube. But when I look at Doug's tubes, he will match the triodes by plate current draw (balanced triode) or by gain (balanced gain). Any thoughts on how these terms would affect the sound?
Craig B

[Firestorm]

Unless the circuit is balanced (it's not), it doesn't much matter whether the tube is. Obviously, you don't want wildly dissimilar transconductance or current readings from the triodes (like one side hot and one side nearly dead).

[David Root]

That's right, I would try a few different tubes in the PI and measure the DC voltage to ground from each plate. Pick the best one, then measure the AC voltage to ground. Should be close.

This is a bit of a false goal in the sense that 1) the power tubes can also affect the final voltage balance anyway, and 2) for best harmonics production and finest tone, the PI usually needs to be several volts off of perfect balance in practice. So don't stare yourself blind on this.

The ideal is a trimpot on the PI B+ to adjust the plate voltages into balance, then out balance them for optimum tone, but of course not practical in a vintage amp.

[Craig B]

Thanks guys. DC voltage from the plates to ground is 174 on the 100k side and 187 on the 82k side. I get no reading from plate to ground when I set my meter to AC volts. Am I measuring that wrong? I will try a couple other tubes when they arrive tomorrow.
Craig B

[surfsup]

Are you sending a signal in when you do the AC measurement?

[David Root]

Sorry, yes, I should have made that clear and I didn't. You will need a signal generator to measure the AC plate voltage, or someone else to play your guitar while you measure. I would use a signal generator myself, much easier to do.

[ampdoc1]

Try installing a bias balance pot per some of the older Dumble schematics.

[gingertube]

Many people say that for best "tone" that the PI needs to be a bit off balance. I'm not sure I agree. Everything I've ever done to balance the power amp in general and the PI in particular seems to have improved the general sound.
If using the "standard" stacked schmitt splitter (like Fender, Marshall etc) then I routinely change the anode load resistors (I find 82K and 91K works well or if I don't have a 91K I use 2 x 180K in parallel for 90K) and I choose the 12AX7 or 12AT7 in the PI for best balance of the 2 triodes.
More out of balance generates more 2nd harmonic distortion which is handy to balance (mask) any excess 3rd (or other odd) harmonic distortion BUT people tend to forget INTERMODULATION distortion. In a guitar amp with fairly loose power supplies (not much filtering) the intermodulation distortion becomes important to overall sound. You want to minimize it. Howard Dumble went to some lengths in his amps to reduce it. Tight power supplies (plenty of filtering) and optimised balance.

If you want to experiment with this then here is a brand new idea from my HiFi Amp design.
Look up the data sheet for cathode bias and get the INDIVIDUAL bias resistor value. For an EL34 this is typically 470 Ohms. Calculate a resistor value of 13 to 15% of this ( for EL34 say 68 Ohms). wire the two output tube cathodes together and put this 68 Ohms between them and 0V.

This generates some balancing action in the output tubes but also generates some COMMON MODE local feedback. This was a scheme intended to reduce 3rd harmonic distortion (which it did - marginally) but what stood out like the proverbial canine testicles when looked at with a distortion analyser, was the almost total elimination of the intermodulation sidebands due to the residual power supply ripple on the B+ line.

You need to wind back the fixed bias marginally to account for that resitance in the cathodes.

I run this scheme in my HiFi Amp, Ultralinear PP EL84 with a common 39 Ohms in the cathodes to 0V. (39 Ohms being approx 15% of the usual 270 Ohm Cathode bias value).

Just my 2 cents worth. If you want to use this idea commercially then I expect more than 2 cents return. If you want to put it in your own amp, feel free. If you try it and decide that its BS then let me know.

Cheers,
Ian

[tubeswell]

Increasing the tail resistance on an LTP will ultimately enable you to 'equalise' the plate resistances of both triodes. If you are using 22k with a 12AT7, try going to 33k (or even 39k or 47k), and using 100k plate resistors on each side (or somewhere between 47k and 100k on each side)

[surfsup]

I would use a signal generator myself, much easier to do.

Well it would be more desireable i would think due to the even signal, for comparison. Having someone play guitar will send in varying AC signal which would make measurements a crap shoot. Yes?

[David Root]

Yes.

[ampdoc1]

A lot of good information here! I put a trimmer across the bias on my first Express amp, hooked up my scope and found that I liked the sound better with a little distortion in the PI. It's also an easy way to balance the output tubes.

Different strokes for different folks.

[Craig B]

Wow! Thanks everyone for the great ideas. I put in my new 12AT7 selected for "balanced gain" and replaced the 82k plate resistor with 90k (two 180k paralleled). DC plate voltage is now 174/176 and I love the sound. I don't have a signal generator to provide steady input for measuring the AC plate voltage but if it's working great, I don't want to know something that might make me hear it differently
Craig B

[ampdoc1]

I think you hit the nail on the head! We all hear differently, and we all have our own idea of the sound we want. The basics of my favorite amp builds have component values and topography that I start with on new builds. Then I usually try some new "twist".

[Structo]

I have often wondered if the normal 100K/ 82K plate resistors used on the PI was close enough for a good balance.

I have seen other amps with the 100K/ 91K and thought that might be better.

Also, I have seen schematics with equal values on both, such as 100K/ 100K.

On my 100w ODS I have 110K/ 120K which was recommended to me.