Power Scaling...again

[Chuck H]

Sorry I'm all over the board right now. I just seem to have alot of Qs lately.

I don't understand how power scaling reduces output (I just know it does?!?)

To do it right you should track the bias to mate up with the reduced plate voltage, sooo...

How does running, say, an EL34's at 500 volts and 38ma current make more power than running the same tube at 250 volts and 76ma current? They're both 19 watts. Plus thers a max current on that tube of 150ma. So what happens below 130 volts or so when you must exceed the max current rating to bias the tube? Power scaling claims "down to 1/100th" power.

I just want to understand this.

Thanks

Chuck

[BackwardsBoB]

If you cut the B+ voltage in half, you need to cut the bias current in half too. This will give you 1/4 the output power, and sound more than half as loud. The ratio of bias voltage to B+ voltage is maintained, somewhere less than -1/10 for an EL34. I guess that this is only an approximation of the ideal ratio function, but it seems to work, and it's fairly easy to implement.

[mooreamps]

Well, then my brake works a little bit differently. Lets say the plate voltage to cathode voltage remains at 500 volts, but plate current is reduced to 10 mills. So, what's that, a plate dissipation of about 5 watts ?

-g

[Chuck H]

If you cut the B+ voltage in half, you need to cut the bias current in half too. This will give you 1/4 the output power, and sound more than half as loud. The ratio of bias voltage to B+ voltage is maintained, somewhere less than -1/10 for an EL34. I guess that this is only an approximation of the ideal ratio function, but it seems to work, and it's fairly easy to implement.

So if my EL34 is at 500 volts and 38ma current for an idle of 19 watts, and I cut the voltage down to 250, then I have to cut the current to 19ma. So I'd have an EL34 running at 4.75 watts idle?!? Why in the world doesn't it suffer from crossover distortion? Cold bias is cold bias at any voltage. I'm still confused.

Chuck

[BackwardsBoB]

You won't get really nasty crossover distortion in class AB until your bias current gets near zero. You're normally counting on a bit of negative feedback to compensate for the tubes' inability to source current and correct for the switch from both tubes in the pair conducting at low output voltage to just one pulling anyway. The substantial bias current recommended for normal operation has at least three benefits:

1. The range in which the amp operates class A increases.
2. You insure that when only one tube is pulling, that tube is far from cutoff and at full gain.
3. It's easier to drive the tube to saturation (rock on!).

Since you need a master volume or elaborate circuitry to reduce the drive into the scaled circuit when you scale anyway, all three effects pretty much scale too.

The output voltage range over which both tubes are sinking current will get smaller, but it will scale with the output voltage of the circuit.

For that rare Class A push-pull circuit or a single-ended circuit, the tubes can drive well swinging from near zero current to twice the bias current. Halving the output swing with power scaling halves the bias current requirement. For Mr. Moore, who appears to be scaling screen voltage but not plate voltage, the tube power dissipation will halve. If you scale the plates too, tube plate dissipation is half (V) of a half (I), or 1/4.

I feel uncomfortable going into too much detail on this. There's little value in me misrepresenting O'Connor's work, and that's what I'm basing everything on. TUT4 from London Power goes into detail on power scaling, among other things, and they are taking orders for TUT6, which provides discussion of additional scaling circuit solutions.

[Chuck H]

I feel uncomfortable going into too much detail on this. There's little value in me misrepresenting O'Connor's work, and that's what I'm basing everything on. TUT4 from London Power goes into detail on power scaling, among other things, and they are taking orders for TUT6, which provides discussion of additional scaling circuit solutions.

That's cool,

It's obvious to me that the general macro understanding of power tube operation, while functional for homebrewers, isn't the whole picture. If I work from only that approach :

low voltage on power tubes + low idle current+ reduced drive voltage = Woosy a$$ thin sounding power amp with scratchy distortion character.

But I've heard the power scaling demos. And while they all seem to be done by outstanding players (which is suspect) they sound good.

I guess I still don't get it. If you run the power tubes at low voltage and reduced current. Then you trim the drive voltage to avoid exceeding the low level of bias, are the output tubes really overdriving at all. Or is this circuit just producing the same effect as a post PI master but with a much more complex circuit.

Like I said, there's plenty I don't understand. But on the macro view it doesn't make sense at all.

Chuck

[Steve Conner]

low voltage on power tubes + low idle current+ reduced drive voltage = Woosy a$$ thin sounding power amp with scratchy distortion character.

I'd go one further and argue that *any* distortion device played at bedroom levels will sound like that. Pure lack of volume is what makes it sound wussy, thin and scratchy.

Like I said, there's plenty I don't understand. But on the macro view it doesn't make sense at all.

I totally agree. The most gratifying "quiet cranked tone" I ever got was with a speaker simulator, an EQ and a reverb box going into headphones. I got some quite good tones that way using a plain load resistor and an Alesis PEQ-450, that had enough flexibility to dial in a reasonable speaker simulation, with its two low- and high-pass filters.

I have no qualms about misrepresenting Ko'C's work, since I've never seen it and don't ever intend to...

[mooreamps]

Supposedly, it's about the "cranked tones" at lower volumes. Yes, can be done with an external attenuator. The thing with electronic power scaling, it's less current stress on the power tubes, and less current stress on the output transformer.

When you get that "cranked tone" is when the PA tubes break-over into saturation. At this point, the dc component of the waveform starts to collapse the electromagnetic field in the primary coil of the output transformer, and you get a momentary dc current spike through the power tubes. If there is no external attenuator, the same thing happens in the voice coil of the loud speaker. . Perhaps it's been said it's possible to burn up a 30 watt speaker with a 5 watt amp. Well, this is how it's done.

-g

[booj]

When I think of power scaling, I think in terms of power reduction. In my variable power 2204 clone, I might say, reduce the plate (and screen) voltage from 400 volts to 200 volts. That means I have to INCREASE the current going through the tube to maintain the same idle dissipation. In my case, I use cathode biasing, which means I have to DECREASE the cathode resistor, to put the grid closer to ground, so the tube will concuct more.

[Dave H]

I have to INCREASE the current going through the tube to maintain the same idle dissipation.

Why increase the current to maintain the same idle dissipation? You could let the dissipation reduce along with the plate and screen voltage or better still adjust it to maintain the best tone as the power is reduced.

[Dave H]

I guess I still don't get it. If you run the power tubes at low voltage and reduced current. Then you trim the drive voltage to avoid exceeding the low level of bias, are the output tubes really overdriving at all. Or is this circuit just producing the same effect as a post PI master but with a much more complex circuit.

No, it's not like a post PI master volume, the power tubes really are over driving just as they were in the full power amplifier. In the power scaled amplifier the power tubes are operating at a reduced plate/screen voltage which means they require less signal swing to drive them from cut off to saturation.

Dave H

[mooreamps]

When I think of power scaling, I think in terms of power reduction. In my variable power 2204 clone, I might say, reduce the plate (and screen) voltage from 400 volts to 200 volts. That means I have to INCREASE the current going through the tube to maintain the same idle dissipation. In my case, I use cathode biasing, which means I have to DECREASE the cathode resistor, to put the grid closer to ground, so the tube will concuct more.

Why do you "need" to maintain the same idle plate dissipation? I believe one does not. The power tube will still work just fine.

-g

[Chuck H]

In the power scaled amplifier the power tubes are operating at a reduced plate/screen voltage which means they require less signal swing to drive them from cut off to saturation.

Yes. Exactly. But if you look at the tube data it seems that they still need to be biased at higher current than is used in power scaling to be at a linear operating point. It looks to me as though power scaling idles the tubes closer to cutoff. So close in fact that it doesn't seem possible to reach saturation without substantial crossover distortion. Thats where my confusion is.

Why increase the current to maintain the same idle dissipation? You could let the dissipation reduce along with the plate and screen voltage or better still adjust it to maintain the best tone as the power is reduced.

Again. I agree. But does the best tone for a low voltage operated pentode happen closer to cutoff?

Look, there's obviously something I'm not getting. But I give. I'm being told that it works by enough people. Well, R.G. Keen said "it seems to work" in the context that people are using it. I'll guess he hasn't looked it over that closely then, or his response might have had conviction. Like "it does work" or "it works".

I'll just have to give it a try and see for myself. And I may not have but for your persistance. So thank you for that.

Chuck

[Dave H]

I'll just have to give it a try and see for myself. And I may not have but for your persistance.

Chuck, I've never made or played through a power scaled amp so you can ignore anything I've said about it.

I was only trying to figure out how it could work as well as everybody says it does. I've just done a few simulations and from looking at the waveforms I can't understand how it can sound as good as it's supposed to down to 1/10 power. I was hoping to be able to make the 1/10 power waveform look like a miniature version of the full power waveform but I can't. I was using a continuous sinewave and that's not a guitar signal so it probably sounds better than it looks.

Dave H

[booj]

Why do you "need" to maintain the same idle plate dissipation? I believe one does not. The power tube will still work just fine.

-g

But it puts out less power. The idea actually is to maximize the idle current for the applied plate voltage supply. Sweet saturated momma tones.