Power Scaling in Morgan amps?

[NateS]

If you use a fixed cathode bias, you can take your bias programming voltage off of a divider from B+ and voila - instant bias tracking.

[Merlinb]

I remember one Marshall amp that used a pot and a few diodes to simulate grid current clipping on the power tubes.

Which model was that? Or was it an after-market mod?

[Steve Conner]

I can't remember which model it was, but I'm sure I remember seeing the circuitry on a factory schematic, so it wasn't a mod. I'll look in the industrial espionage folder when I get back to work.

[Alan0354]

If you use a fixed cathode bias, you can take your bias programming voltage off of a divider from B+ and voila - instant bias tracking.

To invert a +ve voltage, you need an opamp......a high voltage opamp. Then you need power supply to run the opamp unless you can find a rail to rail input high voltage opamp. Or you have to play tricks to bias the opamp to a slightly negative reference to avoid a ground reference summing junction. All of a sudden, the simple tracking idea become not so simple anymore.

[jazbo8]

If you use a fixed cathode bias, you can take your bias programming voltage off of a divider from B+ and voila - instant bias tracking.

Whoa?! Tapping off the B+? Also when you say "fixed cathode bias", does it mean both fixed AND cathode bias? Please provide a simple drawing so we can see what you are suggesting...

Jaz

[trobbins]

This may be controversial, but I also think a properly implemented master volume can give you much of the benefits of power scaling with none of the cost.

Certainly the fundamental mode of operation is very different with respect to the output stage - power scaling can overdrive that stage, but wouldn't overdrive the OT or speaker - so as you imply, the relative similarities are exactly the same all the way to the PI stage, and from the OT onwards (when comparing just an output stage variable B+ scaling technique to Post PI MV).

[NateS]

Whoa?! Tapping off the B+? Also when you say "fixed cathode bias", does it mean both fixed AND cathode bias? Please provide a simple drawing so we can see what you are suggesting...

Jaz

Fixed cathode bias holds the cathode positive to the grid. You need to be able to sink the maximum cathode current, and max dissipate cathode current at bias voltage, but that's typically a handful of watts. Note I said 'programming voltage', ie you can't just tie the cathode to a voltage divider - it takes way too much current for that. But consider the following.



(Once I arrived at the basic circuit and started analyzing it I realized it's essentially the same thing as the sink portion of Art of Electronics figure 6.47)

This was designed to take a 2.5v to 5.0v programming voltage, from an AVR using a PWM for DAC output, but you could just as easily take the programming voltage off a divider padded on either end to set the desired range. In this case it's an AC coupled, then filtered and diode clamped to the range above, with a pullup to 5.0v. If PWM ever stops because the micro hangs, bias decays to a safe voltage (cutoff and then some - this should scale from about 40v min to 80v max. AC coupled PWM can't reach 100% duty cycle, and I didn't want to rule out tube swaps, so I left a little surplus bias - I couldn't find any maximum grid-cathode voltage limits beyond cutoff. )

Note that the feedback loop is negative feedback even though it's going into the non-inverting input. Low bias voltage lowers the gate drive, reducing current through the mosfet, and raising the cathode voltage. Another consideration is the current sense resistor - bigger values will make the sensed voltage bigger, but this voltage subtracts headroom from the gate driver's range.

I realized after posting that I left the grid grounded - the grid is the signal input as you would expect, it's just muted because I grabbed this from a version of the schematic intended for simulation. (i.e. it's not a "Grounded Grid" configuration)

[Alan0354]

So it is nothing simple!!! If you talk about uP, this is not simplification in any stretch. There are much easier ways than that.

[NateS]

You don't need the micro for that circuit. It only needs one opamp stage, a HV FET such as you'd use for any FET in the cathode scheme, a voltage divider in the feedback loop, and a reference voltage to set the bias point for a minimal implementation. If you want to say 'no sand in my tube amp' that's fine, but the principle is not that complex. After all.... this IS the theory section.

[Steve Conner]

That circuit will be unstable because of the extra voltage gain of the MOSFET in the opamp's feedback loop. It'll need some compensation.

[NateS]

That circuit will be unstable because of the extra voltage gain of the MOSFET in the opamp's feedback loop. It'll need some compensation.

Horowitz & Hill actually show the compensation. As I said, this was from a version intended for [DC] simulation. It also doesn't show any bypassing on the output sinking cathode current, which is probably a good idea as well, but it should convey the principle. (It was reassuring to find a scheme I cooked up, already in Art Of Electronics and not in the 'Bad Circuit Ideas' sections at the chapter ends)

Their design is actually using this section to drive the pass transistor in a high voltage regulator, but mentions in passing that it can sink current. (In the last sentence of the first paragraph of p368)

[Steve Conner]

Oh, cool, I used the same basic circuit for the power supply of my Toaster amp, except I replaced the opamp with a TL431 cascoded into the sink FET's source. I didn't reinvent it, I saw the Horowitz & Hill circuit first.

[jpfamps]

Fender, Marshall and Peavey don't have it. They cover 70 to 80% of the Market.

Marshall and Peavey make amps with "Power Scaling". The just can't call it "Power Scaling" as "Power Scaling" is a trademark of London Power.

Fender have used power reduction by reducing power amp voltages since the 80s with their 1/4 power feature.

The first mention I've found of "power scaling" is from a patent by Gut Claret filed in 1979.

Control for electronic amplifiers (US4286492)

I expect it took a while after this to take off because the high voltage MOSFETs needed to effect power scaling where either horrendously expensive, or unobtainable.

[jpfamps]

This may be controversial, but I also think a properly implemented master volume can give you much of the benefits of power scaling with none of the cost.

Another issue with the HT regulator circuits is how to protect them against tube shorts. It can be done, but the protection components add even more cost and complexity. The circuits I've seen tend to just ignore the issue.

Speaking to people who have used MOSFET HT controllers the mode of failure tends to be the MOSFET going short circuit, so the power control no longer functions, but at least the amp is still useable. Not ideal, but I can think of worse failure modes.....

Regardless, I always though that having a MOSFET HT controller should make it very easy to implement an over current detection system that kills the HT.

My idea was to have a current sensor in the valves' cathodes that would short out the control voltage to the MOSFET HT controller if a fault condition was detected.

I'd not got around to testing anything yet..........

[Alan0354]

Marshall and Peavey make amps with "Power Scaling". The just can't call it "Power Scaling" as "Power Scaling" is a trademark of London Power.

Fender have used power reduction by reducing power amp voltages since the 80s with their 1/4 power feature.

The first mention I've found of "power scaling" is from a patent by Gut Claret filed in 1979.

Control for electronic amplifiers (US4286492)

I expect it took a while after this to take off because the high voltage MOSFETs needed to effect power scaling where either horrendously expensive, or unobtainable.

This one control the screen grid instead of the B+. Believe it or not, I designed a full function power scaling using variac with separate filament and preamp HV transformer in spring 1979. I went all passive as I didn't know anything about SS electronics at the time. That's how I started my career more than 30 years ago.