CCS and tube aging

[dgta]

In my other CCS thread, DF96 brought up the issue of what happens as a CCS-loaded tube ages. Wondering if others have considered this issue and/or dealt with it.

In a regular voltage amplifier, tube aging causes a decrease of gm. The tube wants to pass less current for a given Vg and Va increases. But Vk decreases, and that biases the tube "up", compensating to some degree and bringing Ia and Va back in line. This self-regulation appears to keep most amps working well for the life of the tube.

When you have a CCS loaded tube, this mechanism no longer works. The decrease in gm cannot change Ia. But Va changes. Vk cannot change, because you have a constant current forced through Rk. The tube DOESN'T rebias itself. If you have LEDs in the cathode, this is more so.

Would be curious how designers of CCS-based circuits deal with this. Do you just let it be? I haven't seen any designs that attempt to deal with it. What if the stage is DC-coupled to the next? Do you just not use CCS on DC-coupled stages?

[SY]

In most cases, I just let it be. As long as the CCS has enough room to swing, the aging process may shift the plate voltage, but the stage performance won't change much, certainly no more (usually less) than it would with resistor loading. If plate voltage is critical, it can be held constant by using a servo or a gyrator circuit.

As a practical matter, I'm using a six channel line amp and a stereo phono amp, chock full of CCS; two of the line amp channels have five years of use, the phono stage has about 3 years of use. No issues.

[Magz]

In my case, I see it as an advantage - kind of a built-in tube checker. I run 2C22 tubes in my linestage with a CCS plate load and Schottky diode bias. Since these tubes have anode and grid top caps its a simple matter to check the anode voltage occasionally with a multimeter; when it starts going up I know the tube is aging.

3 years so far with the same set of tubes and the voltage has been rock-steady.

[dgta]

Yes, I'm concerned with maintaining Va constant for DC-coupled stages.

If you fix Ia, Va (by whatever means) and Vk, something's got to give. Not enough degrees of freedom. You would have to servo the B+ to the stage. If you just servo Ia, I think your grid bias will get out of whack, no?

[DF96]

At least with a CCS you have the option of adjusting it. OK for DIY, less useful for a commercial product? You could add a servo, but then the complexity is shooting up and you have to start weighing the issues of that versus the advantage of the CCS.

[SY]

Check out Wavebourn's gyrator variation of CCS. I haven't tried it myself, but the design appears solid.

[Wavebourn]

Adding servo to CSS we make gyrator that stabilizes DC voltage, but adds roll-off on low frequencies, like choke load. It can be referenced to B+, or to ground. Here is the thread with some examples:

www.wavebourn.com • View topic - Triode load with high DR on AF and low DR on DC&VLF

Here is another version, referenced to ground (attached): you can take output from source of the FET decreasing output resistance (kind of SRPP). It has much better PSRR and stability of voltage across tube is more convenient than across load.

[dgta]

I'm not sure I understand this gyrator very well (the 2-pnp transistor one) without circuit values, perhaps I need a word description. While I understand how it has large AC Z, not sure how this one acts as both constant current source and voltage regulator at the same time. Appears to me it's a voltage regulator only.

Also don't understand this statement:"stability of voltage across tube is more convenient than across load." If you're DC-coupled to the next stage, stability of voltage across the load is not only convenient, it's the law

[richwalters]

10 yrs with CCS in a Williamson driver and the only sound difference in that period is my own hearing, that I've cut down wind. One of my favourite retros, Herb Albert Magic Man.

richy

[dgta]

When you say driver, do you mean the differential pair driving the output tubes? That's a pretty short tail to put a CCS unless you have a negative supply available.