large PI tail resistor

[daz]

On a LTPI, If one wanted to use a large resistor on the tail to achieve better balance, would you want to also change the 82k plate to 100k ?

[Dave H]

On a LTPI, If one wanted to use a large resistor on the tail to achieve better balance, would you want to also change the 82k plate to 100k ?

Yes

[Jazz P Bass]

Of coarse, the very term 'balance' is kind of an oxymoron when it comes to the LTP.

[bob p]

On a LTPI, If one wanted to use a large resistor on the tail to achieve better balance, would you want to also change the 82k plate to 100k ?

I think that what you're asking could be rephrased as this: "If you were to increase the value of the tail resistor, would that result in matching the value of the two different plate load resistors needed for each side?"

The answer is yes, but with a caveat about output.

Another way to say this is that if you want to use the same value of plate resistor on each side of the inverted circuit, that would require you to have a tail resistor that is infinitely large in value to obtain a matched output.

The problem is that increasing the value of the tail resistor is a two-edged sword; as the value of the tail resistor increases it does result in better balance between the two signals, but the price that you pay for that match is decreased voltage swing in the output. That may not be a big deal for power sections that don't need a lot of voltage swing to drive them, but it can be a huge problem if your output stage requires a large swing.

The way that amp makers have typically gotten around the voltage problem is to intentionally use different value plate load resistors to deal with the mismatch in the output. This allows them to use a smaller value tail resistor, which in turn increases voltage swing on the output to the desired level that's needed to drive the output stage.

Here's a handy calculator that helps to illustrate the principles:

Balancing Long-Tailed-Pair Phase Inverter Gains

[daz]

Thanks. the reason i asked is i have tried a 47k b4, but i don't know if the resulting sound is the effect of better balance because the plates are 82 and 100k. So i wante to know because i don't like the tone of a 47k tail, but if the PI is not balanced because of the plate differences i'd ty using 100/100. All because i want to know what a balanced PI does to the tone. But from what i'm getting the 47k does create better balance even with the 82/100 setup. So thats all i wanted to know, which is basically it appears i prefer a unbalanced PI i guess. Or at least one that is not close to perfect balance.

[Enzo]

Or gather a few resistors of the values you discuss, then tack solder them into the circuit in the various combinations you find interesting. Inside of 20 minutes or so you should have empirically decided the question.

[pdf64]

Trimmers would be even better, especially as every tube will be different.
That should enable the actual gain desired from each side of the LTP to be determined and achieved, both as the tube ages and on replacement.
Pete

[Dave H]

My simulation says it’s quite well balanced with either an 82k plate resistor on the left and a tail of 470,10k or two 100k plate resistors and a tail of 1k, 47k. If you have 2 x 100k plate resistors and a 10k tail it’s about as unbalanced as having an 82k plate and 47k tail but it’s mirror image.

[JoeM]

Even with a theoretically balanced PI (maybe with a CC in the tail), wouldn't you need perfectly matched halves in the PI tube (12AT7 or 12AX7)? Also, doesn't the NFB loop negate to some extent the need for 'perfect' balance.

[daz]

Even with a theoretically balanced PI (maybe with a CC in the tail), wouldn't you need perfectly matched halves in the PI tube (12AT7 or 12AX7)? Also, doesn't the NFB loop negate to some extent the need for 'perfect' balance.

Bingo ! Thats exactly what led me to this question. I was trying several AX7's of the exact same brand/model in the PI and noticed some sounded better. So i began to wonder if balance was the reason and if i could maximize the potential by making it perfectly balanced or whether it was a IMBALANCED tube that was what sounded best. So i been thinking about ways to see which is is like best, balanced or imbalance.

[Enzo]

So run a nice sine wave through the amp, connect a scope to both PI plates and watch. The existing circuit will be a given, then number your tubes, and see what the size of the waveform is at each plate and note it down for each tube. Also note which tubes sounded best while playing. Then compare your collected numbers. That should tell you by looking at the tube that sounded best versus the closeness of the balance between plate signals. If tube number 3 sounded best, then either its plates were pretty close or they were not, and there will be your answer.


And whether you wind up liking balanced or unbalanced better, remember that what is important is the circuit, not the parts. For example, what if you really prefer the balanced PI sound. You may find that the circuit around the tube is fairly unbalanced, and so is the tube, but in the opposite direction, so together they balance out.

[bob p]

This might be a decent way to screen your twin triodes to find the ones that are well matched: set the tail resistor value pretty high, which balances the PI output, then you can just operate the PI into a matched resistive load. If you do a differential compare of the outputs, that would give you a pretty good idea if your twin triode is well matched. With differential summing (A-B) the best matched tube will have the flattest output.