OC3 Tube Voltage RegulatorTo Drop Voltage To EL37 Screen Grids

[hasserl]

Hey guys, it's been awhile since I've checked in. Hope all is going well with you all. I just ran across something that is a new on one me. I got an old Ampeg 835C amp in for fresh caps and new power cord, and doing a search for a schematic, well I didn't find the exact one I was looking for, but I think this one is close enough to get me going.

Anyway, check out the use of an OC3 tube to regulate the voltage to the screen grids. I haven't seen that before. Interesting, aint it?

Can anyone elaborate on why Ampeg would use this? Benefits or drawbacks vs using a choke or large power resistor? The data sheet shows the OC3 should regulate the voltage to 105 volts. That is quite a drop from the 440 going to the plates. Seems like it would throttle back the tubes quite a bit.

[Jazz P Bass]

I would have to surmise that the screen grids are purposely kept low to ensure that what you get is a 'clean' bass amp.

[Enzo]

You have the unit powered and there is 105 on the screens?

I read it different, though I could be wrong. A glow tube - gas regulator - is essentially a zener. So what I see is 440v plate supply, and they DROP it 105v to the screens, so 335v on the screens. TO make a 105v supply, you put the tube across the supply - like a zener. There would need to be a current limiting resistor from the higher voltage.

[Mick Bailey]

That's how I see it. Those tubes are usually seen as a shunt regulator, but here it's in series. Feeding the screens in the Ampeg configuration will maintain a (roughly) fixed voltage drop from the plate voltage, but that voltage drop assumes a relatively fixed load to remain constant. Also, here there's no current limiting - particularly bad in a fault situation. It may have been better to have screen resistors as well, ideally one per tube to reduce interaction.

Edit; Removed incorrect reference to resistance drop.

[hasserl]

You have the unit powered and there is 105 on the screens?

I read it different, though I could be wrong. A glow tube - gas regulator - is essentially a zener. So what I see is 440v plate supply, and they DROP it 105v to the screens, so 335v on the screens. TO make a 105v supply, you put the tube across the supply - like a zener. There would need to be a current limiting resistor from the higher voltage.

No, I haven't measured the voltage yet Enzo, I wanted to get the filter caps changed out and the power cord replaced before I do, the caps are original and the complaint from the owner indicates failing caps. The cord is really frazzled, the insulation is broken in multiple spots. I was just going by the tube data sheet. I'll post back with the actual voltage measurements after I get the caps replaced and get it fired up.

[R.G.]

I think it's there as a dropping zener too; It's a handy way of keeping the screen voltages lower than the would be with just screen resistors. I agree that it's chancy that the screens have no form of limiting, even resistors.

I wonder whether the designer either (a) liked playing with regulator tubes or (b) liked the sound of the amp with a low impedance on the screens. No good way to tell.

It's easy enough to integrate both the ~100V drop and some screen resistance by connecting the screens to a voltage divider between B+ and ground set up to drop the 100V and be screen resistors at the same time, simple playing with Norton-Thevenin equivalents, although that wastes a fair amount of power. It's also easy enough to convert that to a MOSFET based amplified zener, even including a current limit into the zener dropper, or converting it to a simple source follower regulator with series screen resistors.

Come to think of it, *I* have never had a chance to play with regulator tubes...

[Pedro Vecino]

So is. It works like a zener subtracting voltage. But an 0C3 it is much more fun because its violet light oscillates with the intensity of the sound. Here, the same type of circuit with 0B2:

[Mick Bailey]

Also interesting with some tubes the breakover voltage is different in the dark than it is in daylight.

[Tom Phillips]

... an 0C3 it is much more fun because its violet light ...

[hasserl]

Thanks for the replies guys, I'm still swapping out the filter caps. I got the old cans removed last night and made up some blocking plates to cover the holes in the chassis. I'm going to go with individual caps instead of cans, using some terminal strips to mount them. One thing I did find is the wiring around the OC3 does not match the schematic, I'll correct that and upload it later.

[Mike Sulzer]

I think it's there as a dropping zener too; It's a handy way of keeping the screen voltages lower than the would be with just screen resistors. I agree that it's chancy that the screens have no form of limiting, even resistors.

Suppose we have a design using the usual screen resistors that is a good compromise between power output and screen life and safety. Suppose we measure the minimum screen voltage at high power with a sine wave drive. This corresponds to minimum plate voltage, maximum plate current, and maximum screen current during the cycle. Suppose we replace the screen resistor with a screen supply that provides a solid voltage at the minimum that we measured. Both the screen voltage and current during the rest of the cycle are lower in this new design, and so the average screen dissipation is lower than with the design using the resistors. The peak screen current is the same, since we set it up that way. So is this not a better design that will have longer tube life, and probably better linearity (if that is what you want)?

[R.G.]

Good suggestion, worthy of some thought.

Suppose we have a design using the usual screen resistors that is a good compromise between power output and screen life and safety. Suppose we measure the minimum screen voltage at high power with a sine wave drive. This corresponds to minimum plate voltage, maximum plate current, and maximum screen current during the cycle.

I need to think about this for a while, but it seems to me like the screen current (and hence voltage) varies during the cycle, since screen current is caused by the plate getting close enough to or below the screen voltage so that some of the electrons forming the cathode current are attracted to the screen grid instead of the plate. If my mental picture is right, this is an instantaneously varying quantity that's a distorted form of the assumedly sine wave current we want through the plate. In fact, a distorted half-wave-rectified version.

If my mental model is right in this, then the introduction of a screen resistor between the low impedance screen supply and the screen electrode introduces some distortion just by its gain modulation effect and by stealing some plate current on the peak-ish parts. That's in the right place and sense to introduce a form of compression on the output tubes, alternately. I'm not sure how big this effect is.

Suppose we replace the screen resistor with a screen supply that provides a solid voltage at the minimum that we measured.

Mental model in place and running.

Both the screen voltage and current during the rest of the cycle are lower in this new design, and so the average screen dissipation is lower than with the design using the resistors.

Yes. The now-fixed screen voltage is constantly lower, the tube transfer function is lower, and does not vary during the cycle as it does with a screen resistor to a higher fixed voltage. However, this does remove the instantaneous variation in tube gain that the resistor model does. I don't know whether the effect of the removal of the resistor/screen modulation is a good or a bad thing. My guess is it's good for hifi, bad for rock-n-roll.

The peak screen current is the same, since we set it up that way.

I'm not sure it is. That's the point that is making me insert weasel-comments. It depends on whether the internal impedance of the screen grid is resistive and linear or not. I'll have to ponder that and do some research in my arcane-tube-literature stuff to figure that out. I've been bitten by nonlinear electrode impedances before. Twice shy and all that.

So is this not a better design that will have longer tube life, and probably better linearity (if that is what you want)?

It's a good point to start some research. I do remember a lot of hifi amps with the screens of the output n-todes tied to a much lower than B+ point, without screen resistors. I remember most guitar amps as tying the screens to B+. I remember ultralinear amps modulating the screens at a fixed percentage of the plate voltage for better linearity's sake.

It's a how-you-say rich field for some investigation.

[mozz]

You think a 5651A might have any use? I have a bunch. Maybe for 6v6 which only draws a few ma's.?

[Mike Sulzer]

... worthy of some thought.

And possibly a simulation.

[nickb]

And possibly a simulation.

Using an EL34 with good screen current model, I get 3.7W for the screen dissipation fed from a constant 320V. The peak screen current at 30W out was 46mA so a 2.6K resistor off the 440 rail was substituted. The power out was about the same but the screen dissipation went up to 4.5W. I did not adjust the bias current between sims so the EL34 was baking at 30W idle with the resistor. If I reduced the idle current to be the same as the voltage drop case, the screen dissipation went down to 3.6W.

The most notable thing was the crossover distortion in the constant voltage case was significantly higher:

Constant V: 1kHz:22.6dB, 2kHz 3dB, 3kHz -20.9 db
2.6k + 175 cathode: 1KHz: 23.5dB, 2kHz -10.7dB, 3kHz -25dB
2.6k + 240 cathode: 1KHz: 20.1dB, 2kHz -11.9dB, 3kHz -25dB