zener on ct...again

[R.G.]

Now, Steve -
Isn't there some poetic justice in millions of mindless silicon slaves working ceaselessly to make the world a safer place for their hollow-state masters? Taking care of the drudgery so the thermionic overlords can do the audio amplification they do so well?

AS the MOSFET Follies proved, there are certain places where silicon can aid and abet tubes, reserving them for their rightful place at the top of the audio hierarchy.

That's my story and I'm sticking to it. 8-)

[jaysg]

The 50W zeners from Mouser cost $8.55 to $9.41 each.

A 180W power MOSFET(512-FQA19N20C) is $1.74.

No offense, R.G, but so?

No one is going to do this in a production amp. It's a one-off approach to avoid buying a new PT with appropriate specs. The Mosfet solution involves complexity, e.g., a circuit. The 50W zener is stupid simple, and imo, a more robust solution as there are no anti-static handling issues. It also consumes a bit less space.

[Steve Conner]

Now, Steve -
Isn't there some poetic justice in millions of mindless silicon slaves working ceaselessly to make the world a safer place for their hollow-state masters?

Yes, if it weren't for microprocessors, I'd probably be living in a trailer on benefits right now. But a microprocessor in a tube amp, I'm not so sure about

[R.G.]

No offense, R.G, but so?

No one is going to do this in a production amp. It's a one-off approach to avoid buying a new PT with appropriate specs. The Mosfet solution involves complexity, e.g., a circuit. The 50W zener is stupid simple, and imo, a more robust solution as there are no anti-static handling issues. It also consumes a bit less space.

No offense taken.

I think our varying perspectives probably match our past experiences and our perceptions of the problem.

You are correct on all counts down at the bare facts level. What varies is the interpretation.

1. Not a production amp; dead correct, and as we all know, it is possible to afford things in a hand-tweaked personal amp that would give an amp maker's MBA's apoplexy. No dissention here. However, in every problem involving money, there is always a place where you say "Hey, wait a minute... how much more am I paying?" and there ought to be a crisp answer.
2. Mosfet solution requires a circuit; correct again. You need a MOSFET, a gate protection zener, a gate stopper resistor, and either two resistors or a pot. 50W zener needs just a zener and if you already know what voltage you want, you're set. However, the question was about a *string* of zeners for a variable/switchable power supply. Compared to that, the MOSFET approach not only is cheaper, it has fewer parts and wider range as well as simpler adjustment (pot vs a high voltage switch).
3. Anti-static handling issues. Yes, you do have to know how to not kill a MOSFET. I personally have never killed a MOSFET by accident, but then I don't build circuits in a dry environment. Maybe an issue, maybe not. My experience is that this just requires some normal cautions.
4. Space taken; for one zener, about the same volume, but easier wiring. For a switchable array of zeners, the MOSFET approach is much smaller, fewer wires, etc.

There is no "best" until one can define what "best" would be - lowest cost, smallest space, simplest to build, fewest parts, easiest to modify, easiest to maintain, most reliable, or a weighted average of all of the above.

My personal experience is that zeners used for power applications are unreliable and should be avoided if possible and replaced with active equivalents when encountered. Ask the owners of Fender's Blues DeVille and Hot Rod Deville how they like the resistor/zener +/-15V regulators in them. Those are the parts with the burned spot on the PCB under them, and the ones that techs will tell you they're always replacing.

Your experience may be different, and I respect that.

[Pedro Vecino]

Does someone know some schematic with specific components and his values for a typical 40/50 volts reduction? The problem of the power zeners is that they radiates a lot of heat and till now only they have given me confidence to reduce screen grid voltages...
Thanks

[Chuck H]

When it was a 50W amp the EL34s were idling at 19W but now it's a 12W amp with the EL34s idling at 4.5W. The idle dissipation of the EL34s is the same % of the output power for both the 50W and 12W amplifiers so I wouldn't expect the crossover distortion to increase dramatically.

I understand the concept. But it seems arbitrary. When I look at the load line for an EL34 (our example tube) it shows that for 250 volts the tube should be biased to about 60ma to be in a linear operating region. So without knowing more about why or how it's OK to operate a tube outside of it's intended parameters, the 18ma of idle current just seems too cold. And, unless I'm wrong, when you bias a tube that close to cut off the need to reduce drive voltage arises so that you do not drive that tube into cutoff. Which it should reach long before saturation due to the bias condition, causing crossover distortion.

And if that is the case, then what is really being achived by power scaling? If meeting a power tubes required operating conditions is a goal in amp design then running an EL34 at 250 volts and 18ma seems arbitrary. And if that is the case, why do we even need to reduce plate volts. Why not just make the bias colder and colder and reduce drive voltage to avoid asymerty? What's the difference? I'm not being sarcastic or pontificant. I don't understand. What's the difference?

Chuck

[R.G.]

I understand the concept. But it seems arbitrary.
...
And if that is the case, then what is really being achived by power scaling?
Chuck

Power amplifiers are best understood as a power supply with some extra stuff tacked on (i.e. the rest of the amplifier) that releases some of the power under specially controlled conditions.

Once you say "350Vdc power supply in Class AB1 push-pull" you have largely dictated what the output of the power amplifier can be, without ever saying anything about the tubes, circuit,etc.

To a large extent, the B+ voltage (given that it can supply enough current) is, all on its own, the determiner of the power output of the amp. So simply reducing the B+ and rebiasing properly for the lower B+ gives a smaller maximum output for the amp.

The concept of biasing to X% of a tube's power rating is an OK-ish rule of thumb (not my personal favorite, but OK) but it's only useful for small ranges around some pre-selected B+. It does tend to get in the way of understanding what happens when you change B+ a lot.

More thoughts that might help.

If you have, for example, a B+ of 450V and get 60W out of a pair of 6L6s, the 6L6s are producing all the current swing they can. If you do nothing but lower the B+, the 6L6s can't produce more current swing because they're already maxxed out. So the output power (proportional to the product of current and voltage) must go down as B+ goes down. You can't lower B+ and get the same power as at the higher voltage because you are limited on how much current swing the output tubes can do. So the power available must be less.

And the biasing point under these less-than-max B+ conditions is no longer related to a % of max tube power. A pair of 6L6s producing 30W from a 300V B+ is under less stress than a pair producing 60W from 450V when biased equally into class AB. To even get more dissipation, you have to change the bias into the class A direction to heat them up more. At some point you can bias them to both be on all the time - true class A push pull. When you reach that low B+, you can bias them from pure class A to AB1 to B just by changing only the bias voltage, and they work essentially the sam at all bias points, except for the changing amount of crossover distortion. The output voltage and currents are the same, limited by the power supply and the max current of the tubes and output transformer.

[Chuck H]

R.G., (holy cow, I've got "the guy" responding to my post and I'm still not getting it !?!)

Power amplifiers are best understood as a power supply with some extra stuff tacked on

I've always believed that and tried to work around that concept (even with my limited understanding).

So simply reducing the B+ and rebiasing properly for the lower B+ gives a smaller maximum output for the amp.

Yes. But "power scaling" reduces current proportional to the B+. It doesn't adjust the current to the new B+ congruent with the tubes optimum operating range. Why isn't the reduced current called for in "power scaling" considered as biasing the tubes too cold, as it would be in any other scenario.

The concept of biasing to X% of a tube's power rating is an OK-ish rule of thumb (not my personal favorite, but OK) but it's only useful for small ranges around some pre-selected B+. It does tend to get in the way of understanding what happens when you change B+ a lot.

This is clearly where I need more info. Because the data shows how the tube "should" operate and by that demonstration, where it will fail to perform satisfactorily. I'm sure if I understood tube operation better that I would have an intrinsic feel for why "power scaling" works. But according to the specs, it's just a way to operate tubes poorly under bad circumstances.

You can't lower B+ and get the same power as at the higher voltage because you are limited on how much current swing the output tubes can do. So the power available must be less.

Sure, good. I get that. But "power scaling" is still biasing the tubes cold at the decreased voltage. At least, that's how I understand it. If I ran my EL34s at 400 volts I wouldn't expect the same power and headroom as if I ran them at 500 volts. But I would still be biasing them such that they should reach cutoff and saturation at appropriate intervals for class AB (or even class A push pull) operation.

And the biasing point under these less-than-max B+ conditions is no longer related to a % of max tube power. A pair of 6L6s producing 30W from a 300V B+ is under less stress than a pair producing 60W from 450V when biased equally into class AB. To even get more dissipation, you have to change the bias into the class A direction to heat them up more.

Right. But more dissapation doesn't mean more power in class A, where you may even have LESS power at full output than you do at idle. Not less volume, but less current. That would be an indication that the tube is biased close to the middle of it's operating range. Again, "power scaling" doesn't do this. The current is set proportional to the voltage. Unless that was explained to me wrong by someone else in another post.

R.G., I've benefitted greatly from your freely given advice and web info numerous times. And I mean no disrespect. But it still seems to me that "power scaling" is just biasing the tubes cold and reducing drive voltage so that the tubes won't clip asymetrically into cutoff (since at such bias they will never saturate). At which point the power tubes are contributing NO distortion or benefit to the tone at all. And may even be a detriment. So why not just use a post PI master volume.

Either I'm completely confused, or alot of fundimental flaws about "power scaling" are being overlooked because it's so cool. I mean, it's about time we tried to effect the power tubes in our quest for cranked amp tone...Right? But does it really work? Or are we just glad to be doing something new. No doubt, the Suhr Badger clips sound great. But I'll bet Peter Thorn could sell a Peavy Bandit.

I feel like I've wandered into Bizzaro World or the Twilight Zone.

Chuck

[loudthud]

IIRC power scaling requires a little circuit that makes the bias track the B+. I'm not near my TUT library today but that's how I remember it. Probably two or three transistors, six resistors and maybe two pots. To reduce drive, the phase inverter is usually tied to the variable B+ also.

[Steve Conner]

Yes. But "power scaling" reduces current proportional to the B+. It doesn't adjust the current to the new B+ congruent with the tubes optimum operating range.

Who says? If I was designing a power scaling system, I'd have it adjust the current to give the best sounding overdrive at all levels, whatever that adjustment characteristic happened to be. And I'll bet biasing the tubes into cutoff for that sputtery, funky "broken amp" feel wouldn't be a good start. If you make the bias voltage track the screen voltage, the tube will maintain approximately the same current, and that's what I'd try to start with.

Why isn't the reduced current called for in "power scaling" considered as biasing the tubes too cold, as it would be in any other scenario.

I can't think of any answer to this except "Because Kevin O'Connor says it's not"

I'm sure if I understood tube operation better that I would have an intrinsic feel for why "power scaling" works. But according to the specs, it's just a way to operate tubes poorly under bad circumstances.

Why shouldn't this be the case? What if power scaling was just a viral fad to sell more TUT books?

I'll bet Peter Thorn could sell a Peavy Bandit.

Exactly. I remember when the Peavey Bandit was the hot metal amp that I and all my mosher friends wanted but couldn't afford. I played one recently and they sound better than a post-CBS solid-state fender.

I feel like I've wandered into Bizzaro World or the Twilight Zone.

It's worse than that. You've wandered into the Internet. Get out while you still can.

[hasserl]

No doubt, the Suhr Badger clips sound great. But I'll bet Peter Thorn could sell a Peavy Bandit.

Props to Pete Thorn, but those Badger clips don't sound that great to me, no better than a decent attneuator. There is definitely coloration of the tone at reduced volume levels with the PS on that video, at least as much as when using an attenuator.

[R.G.]

Does someone know some schematic with specific components and his values for a typical 40/50 volts reduction?

Look at the bottom of the page in the
Mosfet Follies at Geofex.

The values are typical for a 40-50V zener. The MOSFET must be rated for about 800V, and at least 2A, as well as 75W or more. The IRF820 and IRF830 work, and there are certainly others that will work. You will need to remove the same amount of heat from the MOSFET as you did from the zener, as the MOSFET is doing **exactly** the same thing the zener does. However, the MOSFET is more convenient to put on a heat sink than a huge zener diode.

Be sure to watch whether or not isolation from the heat sink is needed. If you make it exactly as shown, the drain of the MOSFET can be bolted directly to a grounded heat sink, and only the source needs high voltage insulation.

There is one component which is not shown, as some MOSFETs do not need it. This is a 9V to 12V, 1/2W zener diode, with cathode to the MOSFET gate, anode to the MOSFET source. The exact voltage and power rating of this zener is not critical. It protects the MOSFET gate from electrical transients. Some MOSFETs have this internal to the device, some do not. The IRF820/830 do not have it, so it is needed if you use these devices.

[Pedro Vecino]

Thanks a lot, R.G.
Very interesting articles for the experimentation. Thank you for sharing it!

[Dave H]

And if that is the case, then what is really being achived by power scaling? If meeting a power tubes required operating conditions is a goal in amp design then running an EL34 at 250 volts and 18ma seems arbitrary. And if that is the case, why do we even need to reduce plate volts. Why not just make the bias colder and colder and reduce drive voltage to avoid asymerty? What's the difference?

Chuck

I don't think that will work like power scaling. To reduce the bias current without reducing the screen/plate voltage it has to have a larger negative voltage on the grid, which means it needs more signal voltage swing to drive it to saturation but you've also reduced the drive voltage so the preamp will probably clip before the power amp reaches full power. i.e. it's just like a normal master volume amp. A class AB amp isn't symmetrical even when biased hot. The peak current for the 50W EL34 amp is about 200mA so even if it's biased at 50mA the tube only has to swing 50mA to turn off but 150mA to turn fully on.

The power scaled amp works by reducing the plate/screen and bias voltage which makes it a lower powered amplifier requiring a smaller input signal to drive it to full output. When you turn down the power scaling control to reduce the power the power tubes will be literally running colder but the bias point isn't necessarily closer to cut off from the smaller amp's perspective when you think about it in terms of conduction angle rather than absolute current magnitude. e.g. you could set it up to make the conduction angle overlap (where both tubes are conducting) increase as the power scale control is turned down. If you were to make the overlap increase to 360 deg wouldn't it then be class A with reduced crossover distortion? I'm not sure. It sounds too good to be true or maybe not? It could be class A and still sound horrible.

Dave H.

[jamez]

Does someone know some schematic with specific components and his values for a typical 40/50 volts reduction? The problem of the power zeners is that they radiates a lot of heat and till now only they have given me confidence to reduce screen grid voltages...
Thanks

IIRC power scaling requires a little circuit that makes the bias track the B+. I'm not near my TUT library today but that's how I remember it. Probably two or three transistors, six resistors and maybe two pots. To reduce drive, the phase inverter is usually tied to the variable B+ also.

Not necessarily>>>



From Tube Amplifier Attenuation

This will give any voltage drop you want down to around 30VDC.

Loving all this power discussion. Learning loads. Great forum