SET amp owners thread

[Atmasphere]

I did not write that with the same tubes and components, I repent.

Yes, you are right, there are other OTL.

OTL is a combination of tubes, active elements that work with weak currents and a speaker that requires a large current. The combination, in my opinion, is unnatural. Tubes are used in modes where their service life is short. In addition, they are used inefficiently: in the SET amplifier, I can get an output power of 20 W from the 6?33?, for the OTL amplifier, at least 4 such tubes are needed for this.

In OTL amps there is a problem of preventing direct current through the speaker. And this is only feedback or an output electrolytic capacitor. In both cases, the sound becomes worse.

Four 6C33s can make up to 100 Watts in class A, depending on the load (16 Ohms). A pair of them easily make over 20 Watts! As you describe it, this means the OTL is more efficient??

You'd be surprised how much current can be passed by a properly designed OTL. Were this not the case I doubt we'd have been able to stay in business the last 45 years. Your last statement above is outright false. Our output circuit has no feedback and is direct coupled. If the amps were damaging speakers, again, we'd have been out of business long ago. It is true that if you put a speaker on the amps that can't handle the power of the amps, you might be able to damage the speaker. We've figured out how to keep our amps relatively safe over the last 45 years, if you get my drift

Push pull by its very nature provides a large degree of even order distortion cancellation, leaving the odd order harmonics behind and exposed, if the two halves of the push pull are perfectly matched. The even order that creeps into the measurements of most push pull amps is due to the non-perfect execution and/or device matching. Of course this could also be done deliberately. So, the pattern is quite different between the two topologies and this has audible consequences. Keith Howard did experiments on these patterns by adding the distortion digitally to recordings and he found that a predominantly odd order pattern was by far the most offensive sounding. The same level of odd order distortion with the even order added back in was significantly more palatable. So much for your hypothesis that the 3rd order also masks like the 2nd. You need the preceeding harmonic to be higher than the following harmonic to provide the masking. Remove the evens and you remove the masking as the gap is too large to the next harmonics. The odd harmonics above 3rd are also well known to be unpleasant to humans, often metallic and edgy sounding. Expose them and that edge creeps in and sounds unnatural.

It is also wrong to say that only 20% of the power is usable. If a SET is rated at 20 watts with 3% distortion and nearly all of that is 2nd order, then nearly all of that 20 watts is certainly usable without much in the way of AUDIBLE distortion. AND if your speaker is sensitive, say like 95dB or higher, then you are probably never using much more than 1 watt anyway and you have a lot of headroom for peaks, which at those higher SPLs then 3 or probably even 10% THD (if mostly low order) won't be audible (Cheever makes it clear the audibility of distortion scales with SPL, which is why he has an SPL term in his equation).

If the SET is rated 3% distortion, that won't be at full power prior to clipping. Try it and see.

Its not hypothesis, its theory, and its not mine. Wherever you got information as you describe in your first paragraph, is simply incorrect. What you have to explain is why the 3rd can mask higher orders in a loudspeaker, while somehow this does not happen with an amplifier.

I've explained a bit of this before, but here it is again... : Norman Crowhurst pointed out in the late 1950s that when you combine PP output with a single-ended input, you get a prominent 5th harmonic. He didn't explain why, but I will:- you are combining two non-linearities, quadratic and cubic. So you get an enhanced 5th. If you have only differential throughout, you get only the cubic, which has a very desirable exponential decay of succeeding harmonics above the 3rd. The 3rd itself will be found to be lower than it appears in an SET. The succeeding harmonics are falling off at a faster rate; the circuit is inherently lower distortion than you can get with an SET as less distortion is compounded from stage to stage. The desirable masking phenomena occurs, the distortion is 1 or 2 orders of magnitude lower at any given power level- you get a better 1st Watt.

Now if the only amps you've heard combine PP and single-ended, I'm not going to argue. But please, think about the variables!! When you talk about how PP is inferior, are you using the same power tubes as the SET? How about class of operation? Feedback? and so on. When you eliminate these variables a very different truth is revealed.

 

[morricab]

Four 6C33s can make up to 100 Watts in class A, depending on the load (16 Ohms). A pair of them easily make over 20 Watts! As you describe it, this means the OTL is more efficient??

You'd be surprised how much current can be passed by a properly designed OTL. Were this not the case I doubt we'd have been able to stay in business the last 45 years. Your last statement above is outright false. Our output circuit has no feedback and is direct coupled. If the amps were damaging speakers, again, we'd have been out of business long ago. It is true that if you put a speaker on the amps that can't handle the power of the amps, you might be able to damage the speaker. We've figured out how to keep our amps relatively safe over the last 45 years, if you get my drift


If the SET is rated 3% distortion, that won't be at full power prior to clipping. Try it and see.

Its not hypothesis, its theory, and its not mine. Wherever you got information as you describe in your first paragraph, is simply incorrect. What you have to explain is why the 3rd can mask higher orders in a loudspeaker, while somehow this does not happen with an amplifier.

A loudspeaker, unless it’s push pull driven, doesn’t cancel even orders…that’s how it works…no 2nd then the third is unmasked…no 4th, then the 5th is unmasked etc.

You mean a 50 watt OTL into 8 ohms? Or a 10 watt OTL with two tubes? Maybe. I have owned 100 watt (8 ohm) monos with 4 6C33c per mono and they were most definitely Class AB and with feedback to get a reasonable output impedance.

 

[Atmasphere]

A loudspeaker, unless it’s push pull driven, doesn’t cancel even orders…that’s how it works…no 2nd then the third is unmasked…no 4th, then the 5th is unmasked etc.

You mean a 50 watt OTL into 8 ohms? Or a 10 watt OTL with two tubes? Maybe. I have owned 100 watt (8 ohm) monos with 4 6C33c per mono and they were most definitely Class AB and with feedback to get a reasonable output impedance.

I think you need to clarify what is meant be the first six words above. If I read them right, the statement is nonsense which isn't something I expect from you.

'Reasonable output impedance' and output power are two different things! Into 8 Ohms our Novacron amp made 60 Watts operating class A2. With a pair of 6AS7Gs into 16 Ohms we can make 10 Watts. Works better into a higher impedance though. I don't know why a low power OTL is such a thing, but it is. If you want success with lower powered OTLs stick with 16 Ohm speakers. Of course, that's really true of SETs too, as it is with most tube amps FWIW.

Your idea of masking doesn't seem to make sense. Masking is where a louder sound prevents a quieter sound from being heard. Hiss is one of the very few exceptions where it is possible to hear something quieter than the loudest sound. So a 3rd harmonic can easily mask the 6th or 8th harmonic. All it has to do is be loud enough.

I think one of the issues you might be struggling with is none of the people you tend to quote have actually worked with fully differential tube amplifiers. That sort of thing was really rare until we came along and its remained rare (I know they didn't use one of our amps in such studies- you'd have told me, right?). So when you reference those papers, its sort of like the digital guys referencing studies about LP playback that were conducted in the 1960s (ignoring any improvements that might have occurred since then) or such with no provenance about how the tonearm was set up; often without even mentioning what arm or cartridge was used, whether they bothered to load it properly (if a high output MM) and so on. That tends to limit the scope of such studies or render them useless altogether. FWIW digital guys really hate it when I point that out...

 

[morricab]

I think you need to clarify what is meant be the first six words above. If I read them right, the statement is nonsense which isn't something I expect from you.

'Reasonable output impedance' and output power are two different things! Into 8 Ohms our Novacron amp made 60 Watts operating class A2. With a pair of 6AS7Gs into 16 Ohms we can make 10 Watts. Works better into a higher impedance though. I don't know why a low power OTL is such a thing, but it is. If you want success with lower powered OTLs stick with 16 Ohm speakers. Of course, that's really true of SETs too, as it is with most tube amps FWIW.

Your idea of masking doesn't seem to make sense. Masking is where a louder sound prevents a quieter sound from being heard. Hiss is one of the very few exceptions where it is possible to hear something quieter than the loudest sound. So a 3rd harmonic can easily mask the 6th or 8th harmonic. All it has to do is be loud enough.

I think one of the issues you might be struggling with is none of the people you tend to quote have actually worked with fully differential tube amplifiers. That sort of thing was really rare until we came along and its remained rare (I know they didn't use one of our amps in such studies- you'd have told me, right?). So when you reference those papers, its sort of like the digital guys referencing studies about LP playback that were conducted in the 1960s (ignoring any improvements that might have occurred since then) or such with no provenance about how the tonearm was set up; often without even mentioning what arm or cartridge was used, whether they bothered to load it properly (if a high output MM) and so on. That tends to limit the scope of such studies or render them useless altogether. FWIW digital guys really hate it when I point that out...

Some planar magnetic drivers have magnets on both sides and are therefore push pull and cancel even harmonics.
Your understanding of masking is flawed. Proximity in frequency to the louder harmonic dictates whether something is masked as does the amplitude. So, a relatively low amplitude 3rd will not mask a 6th or 8th etc. This is just wrong.

 

[Atmasphere]

Some planar magnetic drivers have magnets on both sides and are therefore push pull and cancel even harmonics.
Your understanding of masking is flawed. Proximity in frequency to the louder harmonic dictates whether something is masked as does the amplitude. So, a relatively low amplitude 3rd will not mask a 6th or 8th etc. This is just wrong.

A planar with PP magnets will have a better chance of being more accurate, especially when you crank up the volume. When the magnet is on one side, you probably get more 2nd harmonic. I know it makes such speakers a whole lot harder to drive!
And I think your understanding o masking is flawed as well. Its just wrong. We've butted heads over this to no avail for years now, can we agree to just drop it? Or would you prefer to butt heads?

 

[morricab]

A planar with PP magnets will have a better chance of being more accurate, especially when you crank up the volume. When the magnet is on one side, you probably get more 2nd harmonic. I know it makes such speakers a whole lot harder to drive!
And I think your understanding o masking is flawed as well. Its just wrong. We've butted heads over this to no avail for years now, can we agree to just drop it? Or would you prefer to butt heads?

Harry Olson’s studies and plots make it clear that what you say about masking is false…look it up in his applied acoustics book. Or look at a plot in the Cheever thesis from that book. At the typical third harmonic level the masking extends essentially nowhere…certainly not several harmonics higher.

 

[Atmasphere]

Harry Olson’s studies and plots make it clear that what you say about masking is false…look it up in his applied acoustics book. Or look at a plot in the Cheever thesis from that book. At the typical third harmonic level the masking extends essentially nowhere…certainly not several harmonics higher.

So butt heads then...

The pragmatic observer will realize there's something wrong with either the studies or the conclusion drawn. If you were correct, our OTLs would be unlistenable. But they are smoother than SETs. And more detailed. If you are being pragmatic, you'll realize that something is wrong somewhere for it to work like that.

In this case, Olson's book (starting about p455) doesn't support your conclusion. If you are to take him at his word the 2nd harmonic doesn't mask that well either.

 

[morricab]

So butt heads then...

The pragmatic observer will realize there's something wrong with either the studies or the conclusion drawn. If you were correct, our OTLs would be unlistenable. But they are smoother than SETs. And more detailed. If you are being pragmatic, you'll realize that something is wrong somewhere for it to work like that.

In this case, Olson's book (starting about p455) doesn't support your conclusion. If you are to take him at his word the 2nd harmonic doesn't mask that well either.

Exactly, it has to follow an exponential decay with increasing order or the harmonics above become audible. You know what exponential means, right?

As to OTL being smoother, well strongly disagree.

 

[bonzo75]

After listening to the dual FLH videos posted over the weekend, Bill has bought himself a JBL 4550 cabinet. It is massive at 1.5m wide, and deeep. The one in the video is 1.2m. Alongside the pnoe he will continue a dual FLH build.

 

[Rob181]

After listening to the dual FLH videos posted over the weekend, Bill has bought himself a JBL 4550 cabinet. It is massive at 1.5m wide, and deeep. The one in the video is 1.2m. Alongside the pnoe he will continue a dual FLH build.

WOW - now that came from left field - not overly surprised though - as I posted previously - the dual FLH sound had a sensational sense of realism. Please keep us up to date on the progress & wish Bill all the best from me - his build posts were THE BEST...

 

[Solypsa]

After listening to the dual FLH videos posted over the weekend, Bill has bought himself a JBL 4550 cabinet.

Given his superb woodworking skills I would have thought he might make his own. Time I suppose...

 

[acousticsguru]

Given his superb woodworking skills I would have thought he might make his own. Time I suppose...

Was thinking the same.

Greetings from Switzerland, David.

 

[Solypsa]

I see a full circle happening maybe. If the Pnoe / AER is hpf up around 800hz then it might not longer make sense to have such a huge horn for them. Bet there are some beautiful diy wood AER horns from past iterations laying about

 

[Atmasphere]

Exactly, it has to follow an exponential decay with increasing order or the harmonics above become audible. You know what exponential means, right?

As to OTL being smoother, well strongly disagree.

I'm guessing you've not read much of my posts in the last year or two. Even on this thread, even just a few posts back. Put simply, an amp with a cubic non-linearity will express a dominant 3rd harmonic with succeeding harmonics with an exponential decay based on a cubic function rather than quadratic. So fulfills the requirements you've been talking about.

 

[bonzo75]

Given his superb woodworking skills I would have thought he might make his own. Time I suppose...

the cabinets are quite inexpensive and building them will take much longer

 

[bonzo75]

I see a full circle happening maybe. If the Pnoe / AER is hpf up around 800hz then it might not longer make sense to have such a huge horn for them. Bet there are some beautiful diy wood AER horns from past iterations laying about

it will be a separate speaker. He is building a big garden room so old and will have pnoe the other the dual FLH. He will experiment with drivers with the dual FLH, so that will fuel his build itch

 

[bonzo75]

as I posted previously - the dual FLH sound had a sensational sense of realism.

Good video reading.

 

[SVS]

Four 6C33s can make up to 100 Watts in class A, depending on the load (16 Ohms). A pair of them easily make over 20 Watts! As you describe it, this means the OTL is more efficient??

You'd be surprised how much current can be passed by a properly designed OTL. Were this not the case I doubt we'd have been able to stay in business the last 45 years. Your last statement above is outright false. Our output circuit has no feedback and is direct coupled. If the amps were damaging speakers, again, we'd have been out of business long ago. It is true that if you put a speaker on the amps that can't handle the power of the amps, you might be able to damage the speaker. We've figured out how to keep our amps relatively safe over the last 45 years, if you get my drift

I have one tube 6C33C gives a power of 20 W, you have two tubes. Q: Where is the best performance?

I don't care much about your words about the possibility of giving a giant current of your tubes. I read manufacturers' datasheets and draw conclusions. I believe them, not you, because I also have experience using tubes.

If you have no feedback, then there must be an output electrolytic capacitor. I wrote about it. It does not add quality to the sound.

 

[SVS]

The succeeding harmonics are falling off at a faster rate; the circuit is inherently lower distortion than you can get with an SET as less distortion is compounded from stage to stage. The desirable masking phenomena occurs, the distortion is 1 or 2 orders of magnitude lower at any given power level- you get a better 1st Watt.

If you're looking for lower distortion numbers, go with transistors.

 

[SVS]

'Reasonable output impedance' and output power are two different things! Into 8 Ohms our Novacron amp made 60 Watts operating class A2. With a pair of 6AS7Gs into 16 Ohms we can make 10 Watts. Works better into a higher impedance though. I don't know why a low power OTL is such a thing, but it is. If you want success with lower powered OTLs stick with 16 Ohm speakers. Of course, that's really true of SETs too, as it is with most tube amps FWIW.

Output resistance and speaker impedance are completely different things. But they are interconnected in tube amplifiers. Read the tutorial, don't be shy.