PI questions

[wizard333]

What affect do the grid resistors in a LTP have on tone? 1M vs 330k vs?

Also, what are the advantages/disadvantages of fixed bias on a PI instead of cathode bias? I've seen this on a few guitar amps but I'm not sure why its being used.

[txstrat]

What affect do the grid resistors in a LTP have on tone? 1M vs 330k vs?
Also, what are the advantages/disadvantages of fixed bias on a PI instead of cathode bias? I've seen this on a few guitar amps but I'm not sure why its being used.

The resistors set the balance of the "autobias" and therefore the gain of that tube. If that's what you mean.
The bias options (fixed vs. cathode) are just different.
Cathode bias tends to sound "warmer", "creamier" or whatever you like to put it.
Fixed is more effective if you go for clean headroom.

The fixed bias circuit needs a separate negative voltage which has to be set properly.
Cathode bias adjusts itself via the cathode resistor.

[wizard333]

The resistors set the balance of the "autobias" and therefore the gain of that tube. If that's what you mean.

Thought that was the anode and tail resistors? What about the grid resistors; can you expand on that?

Cathode bias tends to sound "warmer", "creamier" or whatever you like to put it.
Fixed is more effective if you go for clean headroom.

In general; is that also true in the PI? Is this being used at an attempt for clean head room? If so, why is it not more common?

[pdf64]

As the grid resistors are bootstrapped, their effective impedance is greater than the resistor value. This pushes the low freq break point, formed by the coupling cap and input impedance, lower than would first seem to be the case. So with regular coupling cap values, 330k cf 1M doesn't seem to make much difference.
Traynor used low value grid resistors on some versions of the YB1A, under 100k.
These seemed to work as per normal, until the phase splitter gets overdriven, after which point the loss of bootstrapping effect (and maybe grid conduction) greatly reduce the input impedance, and seems to result in low freq loss as the break point frequency moves up.

Re fixed bias on the phase splitter, sorry, I've not looked into it as yet.
You are thinking of the PaulC mod to 5E3 type cathodyne, or the LTP grid bias arrangement on some dominators WEM Dominator/Custom 15 Schematic ?

[loudthud]

The Wem Dominator is an example of fixed bias of the PI (as noted by pdf64) and another example I can think of are most Hiwatt amps. I can't think of any commercial examples that use a negative supply but KOC shows several examples in his book Principles of Power. I like having the grids at ground potential when there is a Resonance control in the feedback loop because it eliminates one or two capacitors.

[wizard333]

with regular coupling cap values, 330k cf 1M doesn't seem to make much difference.

So tonally, using very low grid resistors in the LTPI can reduce the knee point on the low end if the PI gets overdriven, if I understand that correctly. But in regards to 1M vs 300k vs.......why use one vs. the other?

As regards fixed bias PI, I'm not familiar with a WEM or the Paul C mod. Examples that do come to mind are a Mesa DC3 or a Koch twintone 2, both of which use fixed bias on the PI.

[anson]

Smaller LTP grid resistor values will produce less noise than larger resistors. As far as I know, using lower values should reduce reisitor-induced noise, and should also reduce the sensitivity of the grids to induced niuse from sources such as the transformers. But I've
experimented with using LTP grid resistors ranging from 100k to 1M and personally haven't heard any difference in noise. It is afterall pretty late in chain of gain stages and so whatever noise is produced doesn't get amplified much further.The lower values sounded like they had a little less bass, not surprising since they formed a more significant low cut filter, in conjunction with the preceeding coupling cap(s), as some folks have already mentioned.
Anson

[wizard333]

So the difference in using a large value like 1M vs a smaller value like 330k basically comes down to a low cut filter? Nothing more than that?

[pdf64]

Pretty much. I class both 1M and 330k as large value resistors; in this application, going down to even 100k still achieves reasonable impedance bridging, due to bootstrapping.
The noise considerations may be more appropriate for applications such as a tube op-amp input stage.

[Daver]

Reducing the size of the grid resistor does reduce the the drive to the PI a bit, too. I've been experimenting with this on an amp I'm building now that has a post phase inverter master. Lowering the grid resistors from 1M to 470K reduces the amount of distortion and grind a little when the preamp is cranked up. It also smooths out the overdrive some and reduces the boomy bass that comes with it. When clean, it doesn't seem to change the tone much. I like the effect overall.
Dave

[imaradiostar]

Keep in mind the resistors are bootstrapped in a long tailed phase inverter- so their "value" is in excess of the measured value.

jamie

[wizard333]

reduces the boomy bass that comes with it.

Very interesting. Does anyone know a good formula for calculating the knee frequency?

[imaradiostar]

It's in Merlin's book. Per his book the input impedance appears to be 2x the value of the grid resistor. So a 1 meg resistor would appear to be a 2 meg input.

jamie

[wizard333]

That gives you the virtual value, but what I was looking for is a frequency response calc.

[imaradiostar]

That gives you the virtual value, but what I was looking for is a frequency response calc.

Here's an easy way to calculate the -3dB point:

Guitar Pedals: R-C Filter Calculator

I also use a simple calculator on my android phone.

If I'm not around a phone or the internet I can calculate the corner frequency the old fashioned way:

frequency= 1/(2piRC) where F is Hz, R is Ohms and C is Farads. You can sub L into the equation too if you need to calculate a resonant tank circuit.

jamie