PI biasing - resistors

[EFK]

I'm trying to get some consensus here on the PI resistors; assuming the changes are being made in the same amp, what will it do to the PI (12AX7) if the resistors in the LTPI illustrated below are changed from 820 ohm/6.8K to the 'common' Marshall values of 470 ohm/10K? Would this change cause it to clip or distort sooner?

[tubeswell]

The bigger the tail resistor, the smaller the output signal swing from the plates will be, but the better the balance will be between the plate's outputs. (However it is this imbalance which produces the growliness in LTP inverters, so if you alter the tail without altering the Plate resistors, you will theoretically change this growliness. Whether this increases or decreases the growliness depends on whether you have already compensated for the imbalance by having different value plate resistors)

Also whether changing the tail resistor will result in signal clipping in turn, depends on how high the HT is. The higher the HT, the bigger the tail resistor can be whilst still getting enough output signal to overdrive the output stage.

Changing the tail will also affect the GFB that is required

Whereas if you increase the LTP bias resistor (colder biasing), you will reduce the PI output signal swing, but you will encourage cut-off clipping in the LTP (i.e. more PI overdrive)

Well that is LTPs as I understand it. Somebody else is most likely to have a much better explanation.

[EFK]

So if I understand you correctly (1) changing the tail resistor [and I'm talking about the 6.8K here, not the 4.7] to a higher value *may* [depending on other things you mention] actually decrease 'growliness' as you put it or otherwise clean up the signal, and (2) decreasing the 820 ohm to 470 ohm "Marshall" value will actually clean up the sound? That's odd, because typically with the preamp tubes, lowering the cathode resistors often creates a dirtier sound. Hmmmm.

[Chuck H]

So if I understand you correctly (1) changing the tail resistor [and I'm talking about the 6.8K here, not the 4.7] to a higher value *may* [depending on other things you mention] actually decrease 'growliness' as you put it or otherwise clean up the signal

Yes, well, it depends. The resistor your talking about is the "tail" in the "long tailed pair" or LTP style PI. Increasing it's value will improve balance and reduce output. BUT... If you cannot adequately drive the power tubes with the lower output created by the larger tail resistor you will actually DECREASE the volume of your clean tone because the PI will clip before driving the power tubes to full output.

(2) decreasing the 820 ohm to 470 ohm "Marshall" value will actually clean up the sound?

It all depends. Everything is connected. If you have a large-ish tail resistor and a small-ish bias resistor then you may not have enough swing voltage to OD the power tubes prior to OD in the PI. So you may hear more clipping from the PI at full output power. If you have a small-ish tail resistor and a small-ish bias resistor you may have enough PI output to OD the power tubes prior to PI OD. In this case you actually can get a louder clean tone even though the amps capacity for OD in both the PI and power tubes is increased. A larger biasing resisror may decrease PI output a little (again, it depends on other things) and this could clean up the tone by keeping the PI from saturation that might otherwise occur. IIRC there is little cutoff from the PI that will be heard in the power tubes because they 'should' be in cutoff at the same time in a push pull amp. So, in the first case we got a louder clean tone with a small bias resistor and in the second case we got a cleaner tone with a large bias resistor.

There are many variables to consider. How much voltage will it take on the power tube grids to drive them to full power and beyond? How much plate voltage is on the PI? Is the amp class A push pull, a hot AB1 or a cool AB1 bias on the power tubes? Does the amp use NFB? These are some things that effect PI design for a particular amp and keep PI design from being like, for example, tonestack design. Where the tweaking process is limited somewhat to the function of only that part of the circuit. The PI has a job to do that is relative to power tubes state AND usually within the FB loop so you can't effect the tone of the PI without effecting the state of the power tubes without effecting the FB loop which now effects the PI, etc. etc. So the PI design is PART of a bigger circuit and can't really be considered by itself for tweaking.

Randall Aiken has a great tutorial on the LTP PI values and how to build a PI for proper "function" that may shen a little light:

Long-Tail Pair Design


Chuck

[Alf]

Long-Tailed-Pair Phase Inverter Gain Calculator


Here you can calculate the differences

Very handy site !


Alf

[EFK]

Thanks very much guys - both of those links are extremely helpful!