uCU based bias supply

[GainFreak]

Hi,

I was thinking about using a uCU to bias the tubes in a power amp. A uCU will "measure" voltages and currents and upon pressing a button will bias the tubes at preset working points. For example you can enter several types of tubes' anode dissipation in the uCU selectable with a button and have 3 buttons for 50, 60 and 70% of idle current or somethnig similar.
The question is what to use instead of bias trimpots because digital pots are out of question in this application.
I would appreciate any comments and ideas ot if you can point to similar existing solutions.

[loudthud]

A simple circuit could be devised to simply amplify the range of a digital pot. Pot adjusts from +4V to +5V and circuit outputs -40V to -50V. Basically a high voltage op-amp or descrete transistor circuit.

[R.G.]

I designed exactly this into the first design of the Workhorse amplifier line back about four years ago. A microcontroller sensed tube current, then twiddled the bias voltage to get the bias into position. It worked as expected when I solved several practical problems.

One problem was the use of digital pots. I used digital pots and ... surprise, surprise, used an opamp to amplify the digipot range up from 0-5V to 0 to -75V.

It's not easy to get opamps to work there. High volt opamps are expensive and tricky. I used an ordinary opamp, but put a 200V transistor on the output to amplify the opamp output again, and enclosed the transistor in the feedback loop to get it accurate.

That increased the gain of the opamp+transistor high enough that the opamp's internal compensation would no longer keep it stable. I custom-stabilized the opamp/transistor. Along the way, since the transistor also inverted the signal, I flipped the use of the inputs; + became - and vice versa. A little careful gain-phase work and it's stable, gives 0 to -75V and works under uC control.

It was a user's nightmare, and I scuttled the design in favor of the red-green stoplight indicators of bias, which did exactly what a user needed.

The problem with the uC control were ones of when to make the uC sit still and NOT dink with the bias. You see, tube bias drifts with temperature. We want the biaser controller to set the proper bias so that the tubes run correctly when they are just turned on, and when they're hot. So the biaser has to pick when to be active: when the tubes are cold? When they're hot? All the time?

Of these, you can't pick "all the time" because if they're passing signal, the current of a class AB tube changes strongly with signal level. The biaser either has to actively look for times when there is no signal to diddle the tubes because if it tries to bias when the owner is playing, it's fighting the natural operation of the tube. And a player does NOT want to hear what tube biasing sounds like. 8-) I designed an active mute into the amp to make sure that humans didn't listen to that.

If you pick "when it's cold" you can't be sure the sound is right when the tubes heat up.

If you pick "when it's hot" you have to either not let the player use the amp while you're biasing or make the player allow biasing at some particular time. Maybe a "please can I bias myself now?" light on the amp would work. My solution was to put a "go bias yourself and be quiet about it" button the player could press. Otherwise, the uC put the bias where it was last time it woke up and stayed out of mucking with the amp in competition with the player. Kind of like a sober amp tech sitting behind the amp.

That reduced the utility a lot. Having to put in a button that the player pressed to bias was easy enough, but to a player, it was not much easier than the red light/green light system which a player could use to re-bias themselves, in seconds, at any time. And introducing a uC to do something that was simple enough for a human to do themselves but also introducing more things to go wrong was just wrong.

So we ditched a perfectly good, working computer auto/self bias system.

As with all computer systems, making the computer understand what the user *wants* is the tough part.

[GainFreak]

I see that R.G. has been through this already

The problem with the uC control were ones of when to make the uC sit still and NOT dink with the bias.

I was thinking of using the rebiasing function depending on how often I'm playing - once a week, or every couple of weeks but definately not all the time. I thought it would be useful also when changing and using different tubes and when they are not matched.

[Steve Conner]

I play with tube amps as a hobby to take my mind off the uCs I deal with at work, so my viewpoint is, if you pardon the pun, biased.

I just don't see the point of doing this. Tubes don't drift with temperature like transistors do, and the bias point isn't at all critical anyway. If they do drift with temperature, it means they have high grid leakage and are probably getting towards end of life. Well, excepting some tubes like the EL34, 6550 and KT88, where the grid is so close to the cathode that thermal expansion actually affects the gain.

If you really wanted to do it, I'd integrate the bias controller with the standby switch and power-up delay. On throwing the power switch, it would wait a minute until the tubes were warm, then turn on the HT, bias the tubes, and unmute the audio.

If there were a standby switch, I'd have it repeat the bias/unmute procedure every time the amp came out of standby.

I agree that a DAC (doesn't need to be a digital pot, DACs can be cheaper) amplified by a high-voltage amp is the way to go.

Bear in mind that unregulated bias makes an amp's bias point more stable, because it tracks the (also unregulated) screen voltage. Once you regulate one rail, you end up having to regulate them all!

I prefer the "meat in the seat" method with a meter attached to the amp's front panel and an easily accessible bias knob. The meter adds extra geekpoints.

[R.G.]

If you really wanted to do it, I'd integrate the bias controller with the standby switch and power-up delay. On throwing the power switch, it would wait a minute until the tubes were warm, then turn on the HT, bias the tubes, and unmute the audio.

If there were a standby switch, I'd have it repeat the bias/unmute procedure every time the amp came out of standby.

That's exactly what I came up with. I designed in a uC controlled mute and also a HV switch so the uC could do some fail-safe work as well; "standby" was a suggestion to the uC.

The fly in the ointment was that you couldn't wait one minute. The tubes would drift. They'd drift for about three minutes, give or take. True, it was small drifts, but a uC wants you to tell it where the line is. Telling it to bias and tinker until drift was below X rate meant that a drifty tube would keep the amp muted for a long time. And worse yet, I couldn't very well put into the instruction manual that "your amp will self bias its output tubes for a while after you turn it on. Some tubes will take longer than others to warm up properly. If your amp takes longer than ten minutes to un-mute, think about changing power tubes, or send it in for service, whichever makes you less angry..." I had to come up with rules for uneducated guitarists; after all, what's the point of a uC controlled amp that you have to get a degree to use??

I prefer the "meat in the seat" method with a meter attached to the amp's front panel and an easily accessible bias knob. The meter adds extra geekpoints.

It does.

If you ever find one, try biasing a Workhorse. I think you'll like it. Amp on, tubes warm up. Peer through one hole and twiddle the screwdriver (which is contained in a tube, can't get to anything else) and get a green light in the hole. Tube is biased. Red light is too much current, black is too little current. Repeat for the other tube. Re-do this 10 second exercise whenever you like, or when you change output tubes from 6L6 to EL34s between sets.

And you'll really like the 9Vdc outlet on the amp to run your pedalboard.

Sigh...

[Steve Conner]

The fly in the ointment was that you couldn't wait one minute. The tubes would drift. They'd drift for about three minutes, give or take. True, it was small drifts, but a uC wants you to tell it where the line is.

And you'll really like the 9Vdc outlet on the amp to run your pedalboard.

Sigh...

I didn't mean that the uC should run some sort of loop that tries to converge, but that it should do a once-off setting of the current at the end of the standby timer period, and ignore any drift. If the user wanted to rebias, he'd have to put the amp in standby and back out again. That seems to me the only sane way to do it.

I'd have loved an amp with a 9v outlet. Almost as cool as a SVT with a built-in car cigarette lighter. Why the sigh?

[GainFreak]

If you ever find one, try biasing a Workhorse. I think you'll like it. Amp on, tubes warm up. Peer through one hole and twiddle the screwdriver (which is contained in a tube, can't get to anything else) and get a green light in the hole. Tube is biased. Red light is too much current, black is too little current. Repeat for the other tube. Re-do this 10 second exercise whenever you like, or when you change output tubes from 6L6 to EL34s between sets

Do you mean a circuit similar to the one found in SVT2 Pro? I was thinking about it but maybe it should be modified to allow individual tube biasing and a pot or something that would go from "cold" (for example 50% dissipation ) to "hot" (70% dissipation)?

[R.G.]

Why the sigh?

We brought out the Workhorse line, did a LOT of work getting them to production; advertised and sold them, got a lot of positive feedback on them, introduced several new features to the amp biz (like the easy bias, 9V output, treble dispersion speaker grilles, and design-for-maintenance) which did seem to work fine.

Then we decided not to continue sales of them. The financial rate of return was less than we could make with the same investment in making and selling pedals, which is our core business. A small company in a chancy economy has to watch this very carefully or they become an ex-company.

But I liked them; what a surprise, I liked my own work.

I didn't mean that the uC should run some sort of loop that tries to converge, but that it should do a once-off setting of the current at the end of the standby timer period, and ignore any drift. If the user wanted to rebias, he'd have to put the amp in standby and back out again. That seems to me the only sane way to do it.

Yep. I got to there. At that point we found that rebiasing with the traffic-light method was just about as fast for a user after they'd done it even once. It's very easy to look through a visual port at an LED and diddle a screwdriver to get a green light. Takes under ten seconds on the third and following times. And it's probably faster to do if you take into account flipping into standby, having the uC set the bias, then flipping out of standby, since the uC takes some time to let the tubes settle. We decided that there was very little advantage to the uC.

I busted my hump trying to come up with a way to have the uC sense bias current while audio was going through. It's just too inaccurate to try to do that through the massive change in average current from no signal to part-signal on the amp.

Do you mean a circuit similar to the one found in SVT2 Pro? I was thinking about it but maybe it should be modified to allow individual tube biasing and a pot or something that would go from "cold" (for example 50% dissipation ) to "hot" (70% dissipation)?

I don't know about the circuit in the SVT2 Pro. Never looked in there.

I did think about a setting that would let a user set a hotness/coldness of bias and then hit it repeatedly, but decided against it because only the more intellectual of guitar players ever get this deep (lots of them talk about it, but few understand it) and after all, it was a "Workhorse" amp - designed primarily to be there and work all the time for a working musician. The temptation to make it a gold-plated, ebony-handled Swiss-Army Hammer (!?), adaptable to anything, was big, but I fought it.

So the setup was to present a green light for a specific bias current, red light for over that and black/dark for under that. No "78% hotness" settings. There were switch settings for two bias points (normal and hot) for 6L6, and the same for EL34. These points were set based on some detailed reviews of where working techs actually set amps they biased.

It is possible to make this work for a variable hotness, but then you start getting complicated, and that wasn't the point. The point was to fix the issue of a guy playing in a bar where he's had an output tube go out on him during a set. He tells the audience that he's going to get a drink, then pops in a new pair of tubes. The tubes warm up, he does the five second green-light thing and he's now good to go back to playing. If he's trying to be fancy, he can change between 6L6 and EL34 between sets, or from "normal" to "hot" bias points.

[Steve Conner]

I busted my hump trying to come up with a way to have the uC sense bias current while audio was going through. It's just too inaccurate to try to do that through the massive change in average current from no signal to part-signal on the amp.

I'm not surprised you came to a dead end on that. It's the same problem as setting the bias on a Class-AB solid-state amp, and engineers have been struggling with that since the 60s. There's still no way to do it with a feedback control loop, you have to twiddle a pot for the specified idle current at the factory, and then rely on the sensing transistor thermally tracking the output devices, which it hardly does.

Too bad the Workhorse line of amps weren't profitable. I took one look at the music electronics business and decided to stay out of it altogether, and do industrial electronics instead.

[R.G.]

Too bad the Workhorse line of amps weren't profitable. I took one look at the music electronics business and decided to stay out of it altogether, and do industrial electronics instead.

That's the rub: they were profitable. Just not as much as we'd make doing pedals. If you only have a limited amount to spend, you have to spend it where it makes the best return.

I know it's the responsible thing to do, it just tastes bad.