Microprocessor Bias Control For Tube Amps

[NateS]

If you can monitor current, theres no reason not to monitor at least B+ voltage, so you can get power. Even if you have only minimal instrumentation and your screen current is lumped in with plate current under "cathode current", it's still a 1st order approximation in the right direction. I think error integration is probably a good step towards ideal behavior. Ie, we can tolerate a 10% overcurrent for 1 minute, but a 20% overcurrent would shut down in half the time.

As for the lunar phase tracking.. I dunno. That might be handy. Track the lunar cycle and refuse to power on during days it's not a good idea to be making racket in the house.

[Gregg]

There is a real requirement for matching of current through the output transformer. This is what cancels ripple hum in a push-pull amplifier with a less-than-perfect power supply.

That's what I meant by matching. AC matching is more for Hi-FI guys.
The other thing is most manufacturers' and matching companies' tubes in fact are not that matched because they are doing it under some average working voltages which most of the time are different from those in your amp. For example I spoke with an European tube retailer and he told me the are matching the tubes at 400V biased at -48V. That's far from most amps I know but in real life the mismatch is within 5%.
The real advantage of this procedure is the burn in which according to some is at least 24hours. That way at least you can eliminate great percentage of faulty tubes.

Most of the time OT primaries are not even in DC resistance but that's not necessarily a bad thing unless you want 2nd harmonic free sound.

I agree with Enzo. Most players don't have a clue what's going on in the amp and they don't need to know as long as the bias module does what it is supposed to do.

[R.G.]

And are we assuming this is a feature designed into a particular amp? As opposed to a universal add-on that should work in any amp?

Good question. Right now it's a technology demonstration vehicle, as we used to say back at Three Initial Corporation.

SO if say bias supply is -50v five or take 10v, the system should recognized as a problem if it finds bias voltage down at maybe -25v. (As results from bias filter cap failure) That seems to me safer than waiting to find out the tube is drawing excessively.

MMM. Good point. I'll have to think about that, but a simple, cheap "loss of bias" disaster preventer would be a good idea.

RG - picking 6L6 as the example is fine. Once we get a working system for 6L6s it would just be a few number details to set it up for other tube types.

Thanks! That's the kind of reasoning I was looking for. I have a lot of theory, a fair amount of practice, but nothing like you have.

But I thought we were creating an automatic system.

Not at the moment. I may have hidden that issue in my posts by overblathering.
(1) I think that automatic biasing systems are not worth the extra design effort over a made-easier manual biasing system. Maybe someday. I've built and used both.
(2) I hijacked the thread by diluting the comments on automating it into making manual biasing so easy that it is practical to pop in new output tubes *and bias them* in a smoky bar after having two beers.
So no, I'm not suggesting or discussing automated biasing systems. I'm off on the tack of making it so easy for a human to do that the fully automated thing can wait.

The user doesn;t need to know how it is done, he only has to trust the results. Kinda like when I start my car, the computer in it does a self-check. I have no idea what it is doing, but if it starts the car, I am happy with it. If we make our system calculate dissipation, all the user needs in a dark bar is LEDs bright enough to see.

That's true, and I think the "green is good, red is bad" system is OK for a much more advanced setup which does the calculation for you. However, that's not what I'm doing right now. Version 1, Release 1 is to make it so easy to bias tubes that you don't need autobiasing. It would be different if biasing drifted so badly that every power-on needed a bias, but most tube amps go years between biasing.

Hell, if we wanted, the system could maintain a real time clock and a look-up table for moon phase. We could calculate the tidal influence on.... eh, maybe not...

If I had that kind of system setup, I would make it *say* it was allowing for moon phase, tides, and sidereal epoch correction, whether it was or not. Well, I wouldn't, don't have the stomach for it, but advertising types would.

I really, really like the idea about sensing a loss-of-bias-accident (LOBA) much like nuclear reactors sense and try to predict and prevent loss-of-coolant-accidents (LOCAs). Loss of bias voltage kicks all the tubes into overcurrent and needs something done Right Now. But it doesn't cover one tube going mad and, for instance, having a piece of an overheated screen dropping down to short the grid to the cathode.

As a bit of systems architecture, perhaps that goes into some hypothetical disaster-preventer chip that I've left room for. I had the bias indicator setup provide one fault output signal per tube, with the idea that the the disaster-preventer would look at all the tube states, maybe temps, history (there's onboard EEPROM for saving that history, as well as real time clocks inside some of the chips, for free with the chip) automating shutting down B+ or the incoming AC in a prolonged disaster, stuff like that. It doesn't all have to be in there, and simple versions could be non-uC latches and such and give a good account of itself.

But watching the bias voltage for hints about failures - that's a good idea.

[Enzo]

Oh good, I had my good idea for the day. I can now go home for the night. I am getting hungry anyway. Hunger is not good for cogent thinking.

[Gregg]

I did some digging to find out what patents are available for digitally controlled tube biasing. Below are some of the containing concepts, routines etc:

US4924191.pdf

Fender patent:

US7944302B2.pdf

Using opto elements (triacs) to control bias voltage:

US20120242412A1.pdf

[teemuk]

KBO Dynamics "Tube Sync" patent applications:
GB 2462368, GB 2462445

Parsek (DV Mark, MarkBass, MarkAcoustic, etc.) Continuous Power Control patent application:
US 2011-0311076

DAR autobias for PP, SE and composite modes:
US 8,026,765

[Gregg]

GB2462368:
Intellectual Property Office - Patent document and information service (Ipsum)

GB2462445:
Intellectual Property Office - Patent document and information service (Ipsum)

US 2011-0311076:
Method for adjusting the maximum output power of a valve power amplifier stage for an audio amplifier, and corresponding valve power amplifier stage - PARSEK S.r.l.

US 8,026,765 :

Audio frequency amplifier - Giovannotto, Roberto Michele

According to a comment in another forum Marshall EPA circuit is based on an expired patent from the 80s.