LTPI balance

[nickb]

Interesting, I only did a quick test with a mosfet to compare these two circuits.
At 1Khz the tail currents are basically the same (in ua range). The CCS holds that level while the upper circuit slowly degrades at higher and lower freqs.

Comparing the upper circuit to the "normal" LTPI (without the added tail amplitude) shows a much bigger difference.
The signal traces below show how balanced the upper circuit outputs are at 500Hz with both plate load resistors at 100K.

[ATTACH=CONFIG]52649[/ATTACH]

[ATTACH=CONFIG]52648[/ATTACH]

As the subject of "real" current sources got raised again, here are my results comparing bipolar, NMOS and JFET current sources. All have great performance.

As far as operation when clipped, I'd expect the version uneuman disclosed will be still fairly well balanced as the lpti is pretty good on its own, the 'traditional' type will be unbalanced and the current source version will remain balanced. The more balanced the less even harmonic distortion which IMHO will be preferable since there will be less intermodulation products.

[uneumann]

uneumann, have you run any simulations for operation under overdriven conditions? Im curious what happens to the balance when the signal is clipping. But I like what you’ve done here. It’s simplicity is elegant, and seems quite effective. Hallmarks of good design

Thanks - appreciate your comments, but really, I didn't design much of anything.
If I can claim credit for anything, it's that I questioned why the Fender circuit didn't bootstrap the tail properly. After that, the rest follows. I have to say it is interesting to see how much changes once you do get the bootstrap idea. Also note that the bootstrap signal to the tail bottom doesn't even have to come from the spkr. It could be a buffered copy of the cathode signal, for example. That may even work better, but I used the spkr signal because it's already there and the NFB signal comes from the same place and it doesn't take any more parts to use it.

As for descriptions and explanations, keep it simple. The tail bootstrap connection signal doesn't feed current into anything or change gain or do anything complex. The whole point of that signal is to make tail current constant. The cathode voltage will move with an input signal, but the tail/bias resistor currents should not change. I refer you to my site where I have a pretty detailed description of the idea. I was figuring all this out pretty much as I was writing (or the other way around) so I tried to be really clear in what I wrote.

As for overdrive examples/comparisons, here are two circuits - the left one is the original Deluxe Reverb with 820/47 feedback to the tail and NFB input. It also has the 82k and 100k load resistors to "balance" the output. The right circuit uses two 100K load resistors and the "enhanced tail" (ETPI?) bootstrap signal and the 820/47 feedback to the NFB input. Both circuits get the same 4v peak input. The first set of curves show the plate signals and spkr signal for the left (Fender) circuit. Both the plates and the spkr signals are clipped. The bottom traces show the same signals for the ETPI (right circuit). Those plate signals do look more balanced, although I don't think that means much once the spkr signal is clipping. I didn't show it, but when I overlay traces of the two spkr signals, they are identical (visually). I suspect the output tubes are clipping before the PI outputs clip so the PI output waveforms probably don't matter much after that.




Edit: Thanks Nick for the added details/comments.

[uneumann]

As the subject of "real" current sources got raised again, here are my results comparing bipolar, NMOS and JFET current sources. All have great performance.

As far as operation when clipped, I'd expect the version uneuman disclosed will be still fairly well balanced as the lpti is pretty good on its own, the 'traditional' type will be unbalanced and the current source version will remain balanced. The more balanced the less even harmonic distortion which IMHO will be preferable since there will be less intermodulation products.

[ATTACH=CONFIG]52684[/ATTACH]

Well - seems transistors are still important for such things. Thanks for that comparison.

[loudthud]

I tried ground referenced inputs and transistor current source from the -80V bias supply. When the amp is in Standby mode, all the current goes through the grids. I ended up using a 30K resistor instead of the current source.

Dumble amps have a pot to balance the plate resistors. Most people who will tell you how to adjust the pot, set it for a slight imbalance.

[nickb]

Look at what I just spotted on a Fender Supersonic. They finally parted with tradition.

[SoulFetish]

As the subject of "real" current sources got raised again, here are my results comparing bipolar, NMOS and JFET current sources. All have great performance.

Nick, sorry if this is a stupid question, but what what measurement is represented by the Y axis on the graph(expressed in dB)?

[Enzo]

Better Y Axis than a Y Pestis.

[SoulFetish]

Better Y Axis than a Y Pestis.

Is that a Black Plague joke?

[nickb]

It's the signal current thru the tail. Ideally it should be zero. Here it is again but done as a ratio to the signal current in the left half of the amplifier which is a much more meaningful number.

[Helmholtz]

The Mosfet circuit - while good enough - shows the behaviour of some shunt capacitance.

[Helmholtz]

As for descriptions and explanations, keep it simple. The tail bootstrap connection signal doesn't feed current into anything or change gain or do anything complex. The whole point of that signal is to make tail current constant.

Of course you are feeding current into the tail. Adding a signal voltage to a finite impedance network means injecting AC current. If you look at the AC equivalent circuit, you will see that changing the voltage at the bottom of the tail will also change the voltage at the cathodes to a lesser extent. While this change may be small, it gets amplified by each triode by a factor of around 25 (this being the voltage gain of the 12AT7 in common/grounded-grid mode).

I am not saying the bootstrapping explanation is wrong. The 2 explanations do not mutually exclude each other.

Did you try to invert the speaker signal for a test?

[nickb]

The Mosfet circuit - while good enough - shows the behaviour of some shunt capacitance.

Indeed yes. It's the reverse transfer capacitance that does it. Adding another bipolar fixes it (see below) with an astounding 120dB or so. But now it's just getting a LOT more complicated than the original idea. Take into account that the output tubes will never be anywhere nearly as well matched and it seems rather pointless for this application.