PI Bias Feed Resistors

Marshall Amp Discussion

Moderators: pompeiisneaks, Colossal

Post Reply
FourT6and2
Posts: 178
Joined: Wed May 31, 2017 11:11 pm

PI Bias Feed Resistors

Post by FourT6and2 »

Not sure if that's the correct term for these. Grid leak/bias feed/bias splitters... the resistors on the output of the phase inverter right before the (usually) 5.6K grids on the power tubes.

Let's take a typical 50-watt 2204/JCM800 circuit. The PI has 0.022uF output coupling caps with 220K bias feed resistors, 5.6K grid stoppers on the power tubes.

What does changing the value of these resistors do to the sound of the amp? I know lowering them was commonly done on Marshalls shipped to the US with 6550s instead of EL34s. And that a lower value can help with tube longevity in the case a runaway with modern tubes. But does the value affect the actual sound of the amp?

Let's say going from 220K down to 180K, 150K, 120K, 110K. EL34s in each case.
User avatar
martin manning
Posts: 13080
Joined: Sun Jul 06, 2008 12:43 am
Location: 39°06' N 84°30' W

Re: PI Bias Feed Resistors

Post by martin manning »

The coupling cap and the grid leak resistor make a high-pass filter, so if you halve the resistor value you will raise the -3dB corner point one octave, reducing the bass response. The corner point frequency is given by fo = 1/(2 Pi R C), where R is the sum of the grid leak and the Zo of the PI, which is typically 40-50k. Doubling the coupling cap value would lower the corner point back to where it was. Reducing the value of the grid leak resistor will also load the PI, rotating the AC load line clockwise, which will change its clipping characteristics.
FourT6and2
Posts: 178
Joined: Wed May 31, 2017 11:11 pm

Re: PI Bias Feed Resistors

Post by FourT6and2 »

martin manning wrote: Mon Jul 24, 2017 9:52 pm The coupling cap and the grid leak resistor make a high-pass filter, so if you halve the resistor value you will raise the -3dB corner point one octave, reducing the bass response. The corner point frequency is given by fo = 1/(2 Pi R C), where R is the sum of the grid leak and the Zo of the PI, which is typically 40-50k. Doubling the coupling cap value would lower the corner point back to where it was. Reducing the value of the grid leak resistor will also load the PI, rotating the AC load line clockwise, which will change its clipping characteristics.
Thanks for the info. I understand some of what you said lol. A lot of it is above my level of knowledge here.

I understand the part about halving the resistor value = raising bass cutoff one octave (thus reducing bass response).

What is "Zo" ? Is that the impedance of the stage?

So .022uF caps + 220K resistors is about the same as 0.047uF caps and 110K resistors? Is there any other difference in terms of feel between the two or is it really all just a numbers thing? What I mean is if the effect difference on paper is basically zero, is there some other effect that isn't represented by the math?

The part I don't understand is: "Reducing the value of the grid leak resistor will also load the PI, rotating the AC load line clockwise, which will change its clipping characteristics."

What's the net sonic effect of this? Reducing the resistor value makes the PI distort less?
User avatar
martin manning
Posts: 13080
Joined: Sun Jul 06, 2008 12:43 am
Location: 39°06' N 84°30' W

Re: PI Bias Feed Resistors

Post by martin manning »

Yes 0.022 and 220k is about like 0.047 and 110k, and Zo is the output impedance of the PI. The grid leak resistors are the load seen by the PI, so just as changing speaker impedance affects the load seen by the power tubes. I don't think there will be much difference in the PI distortion, but you will lower the PI gain and reduce the drive into the power tubes.
FourT6and2
Posts: 178
Joined: Wed May 31, 2017 11:11 pm

Re: PI Bias Feed Resistors

Post by FourT6and2 »

martin manning wrote: Tue Jul 25, 2017 1:14 am Yes 0.022 and 220k is about like 0.047 and 110k, and Zo is the output impedance of the PI. The grid leak resistors are the load seen by the PI, so just as changing speaker impedance affects the load seen by the power tubes. I don't think there will be much difference in the PI distortion, but you will lower the PI gain and reduce the drive into the power tubes.
Great, thank you for the help!
User avatar
V2
Posts: 122
Joined: Sun May 12, 2013 1:45 am
Location: Vancouver Canada

Re: PI Bias Feed Resistors

Post by V2 »

A couple additional points to keep in mind:
- different tube types have different limiting values for the grid leak resistance.
- the limiting value is typically lower for fixed bias than for cathode bias.

So, for example, a JJ 6V6S in fixed bias has a limiting value of 100k, whereas the limiting value for an EL34 (fixed bias) is usually ~250k (if I recall correctly)...
FourT6and2
Posts: 178
Joined: Wed May 31, 2017 11:11 pm

Re: PI Bias Feed Resistors

Post by FourT6and2 »

V2 wrote: Tue Jul 25, 2017 4:01 pm A couple additional points to keep in mind:
- different tube types have different limiting values for the grid leak resistance.
- the limiting value is typically lower for fixed bias than for cathode bias.

So, for example, a JJ 6V6S in fixed bias has a limiting value of 100k, whereas the limiting value for an EL34 (fixed bias) is usually ~250k (if I recall correctly)...
Yup, that I do know thanks :) No plans on going higher than 220K on the board. Typically 5.6K grids stoppers on the EL34s. I don't think too many people go higher than the ubiquitous 220K value. But I've certainly seen lower. Mostly 150K in the case of 6550-equipped amps.

The one I just built, I used 150K. This next one, I'll try 120K. I've been playing around with the ampbooks calculator, which is very handy. So I can tune the bass cutoff response with the caps. And I can use the PI gain calculator to see how much gain is reduced in the PI. So really, you could theoretically find a different resistor/cap combo that gives the same bass cutoff, but buy lowering the resistors, you've reduced the PI's gain. So it's a neat balancing act it seems.
pdf64
Posts: 2688
Joined: Sat Mar 12, 2011 9:23 pm
Location: Staffordshire, UK

Re: PI Bias Feed Resistors

Post by pdf64 »

V2 wrote: Tue Jul 25, 2017 4:01 pm...the limiting value for an EL34 (fixed bias) is usually ~250k (if I recall correctly)...
Surprisingly it's 500k for EL34, or at least it should be http://www.mif.pg.gda.pl/homepages/fran ... e/EL34.pdf
User avatar
romberg
Posts: 501
Joined: Wed Nov 28, 2007 6:10 am
Location: Lafayette, CO
Contact:

Re: PI Bias Feed Resistors

Post by romberg »

On fixed bias amps, the bias feed resistors are only part of the total grid leak resistance. The resistance in the bias power supply is added to the value of these bias feed resistors:

http://valvewizard.co.uk/bias.html

Which is (I think) why things look like they are different for fixed bias versus cathode bias. Really they are not. The tube always has the same max grid leak resistance (from the data sheet). But the path to ground on a fixed bias setup passes through more than one resistor.

Mike
pdf64
Posts: 2688
Joined: Sat Mar 12, 2011 9:23 pm
Location: Staffordshire, UK

Re: PI Bias Feed Resistors

Post by pdf64 »

The limiting value of resistance g1 - k surely includes the bias supply, in the case of fixed bias; usually the bias supply resistive impedance is fairly low and may only add 10 to 20% on to the grid leak values.
And it must also include resistance added by bias vary trem arrangements.
And grid stoppers.
It's the total effective resistance between g1 and k.
Any resistance that is common to more than one tube will have its effective value multiplied by the number of tubes sharing it.

The limit is probably higher for cathode bias because of its self regulating nature, which will act to counteract positive grid current reducing the effective bias voltage; in fixed bias, that doesnt happen, hence any voltage created by positive grid current has to be limited by the g1-k resistance value..
Post Reply