The light is finally starting to come on, I think.

[JazzGuitarGimp]

Hi All,

Regarding the LTPPI plate load resistors. I know Fender used 82K on the top and 100K on the bottom. I have assumed for years (and now I think incorrectly) that this was done in an effort to balance the PI. The D-style amps employ a trim pot between the two resistors, and I am pretty sure this is intended to balance the PI. In doing some reading, I've learned that some players prefer the sound of an UNBALANCED PI - and this is what leads me to believe what Fender did with the mismatched resistors was to achieve an unbalanced sound.

Questions:

Am I correct in thinking the trim pot in D-style amps is to balance the circuit (rather than unbalance it)?

If I want to balance the PI (as close as possible) without the trim pot, would I use two 100K resistors, or an 82K and a 100K?

I'm trying to learn all this stuff!

Thanks,
Lou

[jazbo8]

Am I correct in thinking the trim pot in D-style amps is to balance the circuit (rather than unbalance it)?

Yes, it is used to balance the outputs.

If I want to balance the PI (as close as possible) without the trim pot, would I use two 100K resistors, or an 82K and a 100K?

82k/100k should be closer that's why Fender used them.

[ToneMerc]

Hi All,

Regarding the LTPPI plate load resistors. I know Fender used 82K on the top and 100K on the bottom. I have assumed for years (and now I think incorrectly) that this was done in an effort to balance the PI. The D-style amps employ a trim pot between the two resistors, and I am pretty sure this is intended to balance the PI. In doing some reading, I've learned that some players prefer the sound of an UNBALANCED PI - and this is what leads me to believe what Fender did with the mismatched resistors was to achieve an unbalanced sound.

Questions:


If I want to balance the PI (as close as possible) without the trim pot, would I use two 100K resistors, or an 82K and a 100K?


Lou

Lou IMHO without having a balanced PI tube, using fixed resistors is a shot in the dark. Nevertheless, even with the PI trimmer I still use a balanced PI tube.

TM

[David Root]

The 82K/100K is intended to improve the PI balance, and does.

The Dumble PI trimpot is to allow you to choose the level of balance or imbalance. Hence he started this mod using 100K/100K plate loads, but changed later to 100K/110K or sometimes even 110K/120K, presumably to further refine it and make it easier to get perfect AC signal balance. Or maybe, he's so smart, to allow balancing/unbalancing the output tubes by ear(!).

In both these situations we are talking about balancing the signal AC outputs of the PI, not the DC plate voltages. The DC plates can be quite a long way apart yet the AC signals can still be balanced very well. So long as the two triodes are reasonably well matched DC wise to start with and you use a big enough tail resistor.

Case in point, I just finished a small 8W amp using 4xEL91 output pentodes. Since this was a 7-pin tube build I used the 6J6 for the PI, which is similar to a 12AT7 so suitable for a PI. I used the 6J6 plate curves sheet to design the LTP PI and settled on 100K/110K plate loads, with a 1k8 Rk and a 33k tail resistor. The DC plate voltages are 16V apart, but the AC balance is within 2.5%, as is, without a trimpot.

A trimpot should allow to get virtually perfect AC balance, but you might actually prefer the harmonic content with a bit of AC imbalance.

[Firestorm]

Who knows why Fender did anything? The first use of the long-tailed PI was in the late '50s models, using a 12AX7 tube and relatively small tail resistors (as small as 10K). These always had the 82K/100K plate resistors and are pretty well AC balanced (ignoring the effect of negative feedback, which Fender did not use much of). This plate resistor formula continued through the 6GX amps, but in the AA763 models, Fender went with the 12AT7 PI tube and a much larger tail resistor. This must have led them to believe that the plate loads could be identical 100Ks, which turn out to be pretty well balanced. But they immediately went back to 82K/100K in the AB763s. Why? This configuration is no better balanced (in fact, with feedback it may be a bit worse). Maybe they made the decision by ear. Maybe the equal loads was just an experiment that didn't accomplish anything audible. It wasn't any easier then to find ATs with balanced triodes than it is now (and I'm sure they didn't check).

[roberto]

The higher the value of the tail, the better thebalance, even with both Rp at 100k. 33k up to 47k are ok.

Also the combo 82k 100k with 10k tail is ok.
91k+9k1 (split load) 100k with 22k is even better.

[R.G.]

Let's play circuit theory a while.

A LTP PI is a differential amplifier. It amplifies the difference between the two input grids. It does this **in theory** because the diffamp is actually a common cathode amplifier whose cathode drives the cathode input of a common-grid amplifier.

Uh whaa??

The grid of the non-input side is held (mostly, kind of, pretty much) fixed at one voltage. The input side's grid is wiggled around by the input signal, forcing the cathode to follow it (!) - mostly. To the extend that a voltage on the grid can force the loading of the cathode load impedance to follow it. The load is the loading of the cathode to ground current-setting resistors and the load of the second tube section's cathode when viewed as an input.

If the current-setting resistors are removed and subbed for a constant current source, then the cathode of the input tube can force substantially all of the current source current into the cathode of the second tube, and the second tube is driven very much more strongly than otherwise. With only resistors setting the current in the cathodes, the resistors eat some of the current that the first cathode would force into the second cathode.

The upshot of this is that with a really fixed second grid and a current source load, the diffamp is very, very nearly balanced, and you get equal but opposite gains from the input grid to both plates. To the extent that the current source (which implies very, very high resistance) is resistors, not infinite impedance, the gain of the second half of the tube is lower - it gets less input drive.

You can quasi-fix this by modifying the plate resistors. A lower plate resistor on the input side and a higher one on the non-input side corrects the plate voltage swing imbalance because the lower current swing in the non-input side has to go through a bigger resistor, so it makes a higher voltage.

It can't correct the output impedance imbalance that the different resistors give it. But that's a different concern entirely.

[katopan]

In the standard set up around 90K / 100K is more balanced. 82K is (slightly) closer than 100K and still in the E12 set.

[JazzGuitarGimp]

Okay, so it looks like the assumption I've been going on all these years is confirmed. Frankly, I'm a bit surprised by that!

This is such a great place - I really appreciate all of y'alls help! It means a lot to me that you would take time out of your busy day to help out!

Cheers,
Lou

[Stevem]

The bottom line is sound wise that a 100% balanced PI does not pass the even order harmonics from the preamp section when that section is clipping and this is why a amp that is single ended sounds so wonderfully sweet when it's into clipping!
Every player who is capable of it and has a amp with a A/B output stage needs to at least screw around with a unbalanced PI tube, or a balanced one and mess with the circuit so they can hear the harmonic difference and decide which way they perfere there tone and play feel to be.

[R.G.]

The bottom line is sound wise that a 100% balanced PI does not pass the even order harmonics from the preamp section when that section is clipping and this is why a amp that is single ended sounds so wonderfully sweet when it's into clipping!

Not exactly.

A 100% balanced PI passes whatever signal it gets from the preamp just fine. It can't tell the difference between even order harmonics from the preamp and anything else. It's all just signal to it.

What a 100% balanced PI does is to not generate its own even order harmonics from having plus and minus half cycles come out of the PI at different sizes.

Single ended output stages generate even order distortion because they can't amplify positive-going and negative-going signals the same way. It's not a PI issue - although you may not have meant it's a PI issue with reference to single ended, I'm not clear on that from your post.

[Teleguy61]

Great info RG, thank you.

[vibratoking]

Merlin discusses this a bit here with a few equations:
http://www.valvewizard.co.uk/acltp.html

This has also been discussed in greater detail in several threads on the Dumble side. One is a sticky if I remember correctly. I spent quite a while adjusting the PI trimmer on my 102 type clone using a spectrum analyzer. Looking at relative amplitudes of harmonics and listening for the 'magic' spot that many have commented on. I never found a magic spot. I found adjustments that were different. Some dirtier, some cleaner. Some with better sustain than others, but never a magic spot. With the Dstyle amps, it is always a balance between clean and OD for me. 'Perfectly' balanced is the best for chimey cleans, but then the OD isn't quite right. A little 2nd harmonic is what I have settled on, but I don't remember how many dB down that was from the fundamental. By the way, I inject the test signal into the effects loop return so that I apply a clean signal to the output stage. I take my measurement of the output at the speaker jack after the OT. This accounts for any imbalances in the OT.