Bias pot 'shunt' or tail resistor: minimum value?

[hcorneli]

I've built a couple amps (PR, 6G2) with Doug's simple, venerable 'PR' bias scheme with a 10KL pot, which has worked well. You all know it, but I'll try to attach a pic.

In passing, Doug says about the 27K 'tail' or shunt resistor: "You should not make the 27K resistor less than about 10K or you will have too large of a current drain to ground."

In my simple understanding, cutting the size of the tail resistor increases the bias voltage span by upping the relative contribution of the 10K swing on the pot. I've always followed that 10K suggestion, but got to wondering if it was a hard and fast 10K, or depended on the size of the 'range' resistor, or could be tweaked downward a bit. Thoughts?

[sluckey]

Decreasing the 27K will cause the negative voltage fed to the grids to decrease. This will cause the tubes to run hotter.

What is your goal?

[HotBluePlates]

... I've always followed that 10K suggestion, but got to wondering if it was a hard and fast 10K, or depended on the size of the 'range' resistor, or could be tweaked downward a bit. ...

In a Voltage Divider, nothing is absolute about only 1 of the resistances.

The allowable value is about proportion to the other resistance(s) in the divider, which themselves may be sized in proportion to the Source Voltage and desired result.

[hcorneli]

Decreasing the 27K will cause the negative voltage fed to the grids to decrease. This will cause the tubes to run hotter.

What is your goal?

Good question. A first order goal would be as you say to heat up the bias. Second order, more interesting, would be to widen the span of the pot, say in a different amp. If you had a 50V bias tap instead of the HT tap, and a small range resistor to match, you narrow your bias range if you run say a 27k shunt resistor compared to say a 12k, 10k, 8k.... And some folks use a different pot topology, say a variable resistor to ground. At that point, even a small 10k starts to cut into the range you can achieve around a certain mid-value set by the range resistor.... let me know if that doesn't make sense.

[hcorneli]

... I've always followed that 10K suggestion, but got to wondering if it was a hard and fast 10K, or depended on the size of the 'range' resistor, or could be tweaked downward a bit. ...

In a Voltage Divider, nothing is absolute about only 1 of the resistances.

The allowable value is about proportion to the other resistance(s) in the divider, which themselves may be sized in proportion to the Source Voltage and desired result.

Good point. So if I understand, you're saying the size of the shunt/tail is proportional, depending on the bias source voltage and the size of the 'range' resistor? IOW, the limit isn't 10k, it's (in Doug's HT tap example, anyway) roughly 10% the value of the range resistor?


Merlin on bias entertainingly says this resistor acts as an "end-stop" for the pot, and "is also called an 'idiot' resistor." As to its value, he merely says it will be roughly the same as the pot, "but experimentation may be necessary."

[sluckey]

If you want a wider adjustment range, use a larger pot. I like 50K pot and 47K resistor, like this...

     http://sluckeyamps.com/phoenix/phoenix.pdf

[hcorneli]

If you want a wider adjustment range, use a larger pot. I like 50K pot and 47K resistor, like this...

     http://sluckeyamps.com/phoenix/phoenix.pdf

For sure. You won't remember, but a few years back you helped me set up a 50K pot to bias my 5G9. It works perfectly. My interest in the 10K pot is partly theoretical, but also

1) I like the external-adjust slotted-actuator 10KL CTS pot that Doug sells -- handy, safe, discrete for bias adjust with TPs, especially when we get into tray chassis where you can't just drop a tweed back panel to bias.

2) Various kits and plans use the 10K pot in ways similar to Doug's, and in some of those (eg, Mojotone 5F11) people struggle to get the bias range right.

3) Most of all, Merlin provoked my interest by saying, "Since we will usually want to make the grid leak resistors as large as possible to avoid loading down the driver stage, the resistances used in the bias supply need to be small. A 10k bias pot is ideal." In calculated and simulated bias circuits, I've been trying (for fun) to see how to use his 10k pot and still achieve decent bias range.

[hcorneli]

Oh, and you're both helping me see something that (if I'm right) shoulda been obvious. Though Merlin gives us an easy way to figure out the (approximate) max negative voltage we might need, I'm just learning to figure out the min negative voltage we might want. But I'm just now realizing the governing limit on the size of that shunt/tail resistor is one that works in a given voltage divider to provide that 'least negative' and stops short of "too hot." I'm gonna run a sim of Doug's 10k limit on that PR circuit -- just for fun.


EDIT: Duh. It didn't take long to see that in a PR setting, you'd never get that tail down near 10K with 100K on the HT tap. I have seen folks need to drop it in a 6G2, but never lower than 12K and usually more like 17-22K. OK, on a 50V tap, it gets interesting, but now of course I'll recall what you guys showed me -- it's not the arbitrary size of that resistor, it's the min negative bias voltage the whole voltage divider is producing when the pot is set 'max hot.' If I'm right, anyway.... Thanks.

[HotBluePlates]

... It didn't take long to see that in a PR setting, you'd never get that tail down near 10K with 100K on the HT tap. I have seen folks need to drop it in a 6G2 ... it's not the arbitrary size of that resistor, it's the min negative bias voltage the whole voltage divider is producing when the pot is set 'max hot.' ...

Exactly.   

And in addition to what I mentioned before, it matters the resistance of the entire bias circuit back to the Voltage Source (here to a 100kΩ resistor that's not even on the "DC side" of the bias supply).

... Second order, more interesting, would be to widen the span of the pot, say in a different amp. ...

To get there, you should know "widen how much?"  And you should know "how much" based on observing the Pin 5 voltage needed to achieve a target idle current in real tubes (probably because you needed to modify a bias circuit to achieve that idle current).

Why?

I have 50-60 output tubes of a single type, and a cumbersome but precise Russian L3-3 tube tester.  After measuring a bunch of a single tube type, I discovered a basic truism:  tubes that conform to a particular type do not vary widely from their specified characteristics.

See the L3-3 test card below for a 6L6-style tube (open in a new tab, and ignore the yellow circle).  The information along the bottom edge tells us the tube should idle at 72mA +/-14mA (about 19% tolerance), and should have a transconductance of 6mA/volt +/-1.8mA/volt (30% tolerance).  All this with prescribed electrode voltages (including grid bias).



For tubes with strong transconductance, a small change of G1 bias voltage yields a large change of plate current.  For 6L6, that's literally "6mA plate current change per volt of grid-voltage change" (for 6V6 types, closer to 4-5mA/volt, depending on idle current).

    A Princeton Reverb idles with ~410vdc on the 6V6 plates, so 21mA of idle current is a normal bias.

    If the 6V6 has a transconductance as low as 4mA/volt, it still only takes a +/-5v change at G1 to swing the plate current from 1mA up to 40mA. *

    More-sensitive tubes (with higher transconductance, or "Gm") don't require as much bias voltage change.


During my testing of tubes with varying degrees of prior use, they surprisingly measured not terribly far apart.  Only one pair was very-different from the crowd, having a plate current double the rest of the tubes.  It was an obviously-defective set of tubes, and could not easily have its idle current managed by a normal adjustable bias supply.


* The tube will exhibit more Gm at higher plate current & less at lower plate current.  So the adjustment isn't linear, but the example still generally holds.

[tubeswell]

A handy rule of thumb is to aim for a bias voltage output of about 10% of the B+,  +/- a couple of percent.


So work out your voltage divider based on that. YMMV