Generic design question for Pentode CCS

[geezertron]

I'm designing and prototyping a pentode CCS for the "tail" on a LTP. What I've got hacked out is working fine, but there's one thing I'd like to understand better.

It's possible to get the desired current with the pentode using various combinations of screen voltage and cathode bias. In some cases the screen voltage needs to be higher than the plate voltage. This leads to a couple questions:

1. Is there some rule of thumb for how to bias the screen and set the cathode resistor the "best way" (lowest distortion).

2. Is there any problem running the screen at a higher voltage than the plate? I've already verified that the screen current is low enough to not be a problem.

Thanks.

 

[dsavitsk]

1. You are not looking for "distortion", but rather for a high impedance. Look at the plate curves for the most horizontal spot.

2. It is fine.

 

[DF96]

A CCS needs to present a high and linear resistance. To a certain extent, you can balance off weaknesses in one of them against excellence in the other.

My advice would be to start by biasing as though the pentode was being used as an amplifier. If it would give low distortion as an amp, then it will probably give reasonable results as a CCS.

 

[BinaryMike]

High anode resistance is the dominant goal, and that's maximized by increasing cathode resistance. Higher screen voltage increases cathode current, forcing you to reduce cathode resistance.

 

[DF96]

Higher screen voltage also requires higher negative grid bias for a given current, forcing you to increase the cathode resistance. Given that this is a CCS (i.e. constant current - the clue is in the name) raising the screen voltage is the best thing to do as this also raises the required cathode resistance. Raised cathode resistance increases anode resistance.

My guess is that you want to keep screen voltage near or below anode voltage. This is so you keep clear of the region where the current division ratio between anode and screen depends critically on voltage, thus lowering anode impedance.

 

[geezertron]

Thanks to all. After several days of weather-inspired garden work I'm going to get back to this project this week.

I'm going to keep the screen voltage just under the plate voltage, mainly because that removes a separate power supply I was using for this. See where it goes....

 

[BinaryMike]

I ran across an interesting comment in Cherry & Hooper when I was developing my Bonseki 20/20 power amp design. See highlighted text in attachment. I ended up grounding G2 in the CCS because I didn't discover any difference in performance relative to resistor feed (from the V+150 rail). G2 ends up at the same voltage either way, and simplicity is often a good thing, but I would like to understand this issue better.

 

[Hearinspace]

I believe I remember seeing comments from Wavebourn and Michael Koster indicating tightly regulated G2 voltage is preferable. Also VG2 relative to Vcathode, (leading to regulated cathode V). Is this what you're asking about?

P.S. Were you happy with the way the Bonseki turned out?

 

[Merlinb]

I ran across an interesting comment in Cherry & Hooper when I was developing my Bonseki 20/20 power amp design. I ended up grounding G2 in the CCS because I didn't discover any difference in performance relative to resistor feed. G2 ends up at the same voltage either way, and simplicity is often a good thing, but I would like to understand this issue better.

A capacitor was deprecated because it was impractical to decouple the screen down to suitably low frequencies. These days caps are smaller and more reliable, but still, you can't maintain CCS action down to DC with just a cap. A zener diode (for example) solves these issues in an instant. The only thing you don't get with a fixed voltage is natural adjustment with age.

 

[BinaryMike]

The desired understanding came to me shortly after posting the quote from Cherry & Hooper. It's amazing, how this happens so frequently after stating the problem in writing! The CCS implementation in Bonseki has a minor design flaw. There is a small amount of subsonic common-mode current noise. That noise is cancelled in the output stage, of course, but it still interferes slightly with bias current measurement. When I was working with the prototype, I increased measurement averaging in my DMM to be more certain of the setting, but failed to make the causal connection.

This issue makes it more difficult than ever to justify the use of pentode CCS circuits. Bonseki is a fine power amplifier in any case, especially when its very low cost of construction is considered. More info in this thread: Bonseki 20/20 | Audiokarma Home Audio Stereo Discussion Forums

 

[geezertron]

Here's as far as I've got things (grid stoppers not shown). About a 10X voltage gain, and seems perfectly balanced. Haven't got good distortion measurement yet because I need to clean up some strong heater hum first. What I don't understand is the 25V difference between the plates on the triodes. I've checked things a couple of times and haven't found any component mistakes, and it looks fine on a scope. Even sounds OK-ish (ignoring the hum).

Is this just the variation between the two "identical" triode sections, or should I be looking for a mistake somewhere?

Once I get it cleaned up I'll try some other BH11 tubes and see if the difference persists.

 

[BinaryMike]

Triode matching was a problem for me as well. The typical DC imbalance amounted to an input offset voltage of a substantial fraction of one volt in my project.