SETs and Horns

[tima]

It stands for tapered quarter wave tube and was invented by Voight (was also called a Voight pipe I believe). It is a mix of horn and reflex and because the mouth is mass loaded it gets reasonablly deep bass in a normal sized cabinet. Commercial examples are the Horning speakers, Tune Audio’s smaller horns, some Bastanis, some Supravox designs (I have one of these), Odeon had some designs (I have one of these too), Cube Audio. There is a good website by a Danish designer Troels Gravessen that goes through some nice TQWT designs using a mid he designed for Jansen Audio.
My Supravox are single 8 inch driver, 99db, and get in-room bass flat to 30hz or so. The main drawback is the single driver doesn’t have a Whizzer so it is not truly full range. So I started mating it with a horn mid/high. It has an effortless and huge sound despite not being overly huge.

My Odeon speakers are two way, 97-98db (depending where you read) and mate a 10inch Mid/bass with a wooden horn round tweeter with 1 inch compression driver. Also effortless dynamics in a normal room with a lot of punch. The bass doesn’t go as deep as the Supravox but the mid bass hits harder.
The speed of the bass is a big part of the horn bass appeal but it often can sound lightweight. TQWT sounds fuller on the bottom than most exponential horns I have heard and have the horn bass speed.

Hornings set up right have awesome bass with a ton of texture that is often missing in normal ported designs.

Thanks for the explanation Brad - that helps. As we Americans have not had a lot of exposure to commercial horn offerings, one issue, at least for me, is unfamiliarity with the vocabulary. Is a Whizzer a tweeter?

 

[morricab]

Thanks for the explanation Brad - that helps. As we Americans have not had a lot of exposure to commercial horn offerings, one issue, at least for me, is unfamiliarity with the vocabulary. Is a Whizzer a tweeter?

A whizzer is this little extra cone that sticks out in drivers like Lowthers of Fostex. It extends the highs of these full range drivers mainly it seems through chaotic breakup.

 

[acg]

Still, I think every committed audiophile should try to hear a 45 type driving large efficient horns. It's a bit of a revelation and a mind bender.

The 45 is a wonderful tube when used in the right way. My midrange channel (3kHz - 12kHz) can use either 4V or 2.5V directly heated triodes and the 45 sounds stellar in that position.

 

[Ron Resnick]

Hello acg,

SET does not necessarily equal low power.

We have the NAT Magma New, Trafomatic Drina, and various 833-based amps. These are all 100 watts and up SET amplifiers.

You may quibble with the sound of these amps versus low power SETs, but that is a separate discussion.

 

[acg]

5. What about the question of the oft-mentioned horn colorations? Are they still a concern in modern horns? Are they a concern in vintage horns?

Horn colourations are interesting.

On the one hand you can choose to listen to room distortions from the reflections of a broad dispersal loudspeaker such as a box or waveguide or unity horns, which should be designed to have room reflections as close to identical that coming from the loudspeaker, or you can listen to narrow dispersion front loaded horns which enable you to hear a higher percentage of direct to reflected sound and remove a fair bit of room reflections from the equation. I prefer the latter because the 'sweet spot' in the room is sweeter although much smaller.

Shouty horns are largely ones given too much work to do i.e. too wide a frequency range, particularly trying to go too low for the mouth size. This can be overcome by having more horns which of course comes with its own problems.

 

[acg]

Hello acg,

SET does not necessarily equal low power.

We have the NAT Magna New, Trafomatic Drina, and various 833-based amps. These are all 100 watts and up SET.

You may quibble with the sound of these amps versus low power SETs, but that is a separate discussion.

Hi Ron,

Yes, you are right of course. SPL and watts scale like crazy...double the watts gives 3dB more SPL from the loudspeaker, so from 100w is 6dB louder than 25W, 9dB louder than 12W. So if we say 10% of amplifier output to pay the single-ended zero feedback tax then the 100W amp really is optimal to 10W or 95dB/w/m sensitive loudspeakers.

It is much more difficult to wind a decent output transformer for those mega voltage, high power single ended 211/833/845 etc. amps because of the larger gap to accomodate circa 1000Vdc and the crazy inductance for 100W @ sub 20Hz. These are great big transformers...not so great for mids and highs which are better served by smaller transformers.

To be honest, although I have listened to systems with these bigger SET amps I have never felt compelled to build any, for a number of reasons, some of which I mentioned above.

Anthony

 

[morricab]

No doubt single ended circuits are euphonic. But its important to understand what's happening in a differential circuit!

Both halves are single-ended circuits, but they communicate and become differential because their emitter circuits are made to be common, and then tied to either through a constant current source or minus voltage of enough magnitude that the circuit is caused to be differential.

So that means both sides make a 2nd harmonic and have a quadratic non-linearity. This in turn means that the most significant harmonics it makes will be the 2nd and 3rd, and the 3rd will be at the level seen in a single-ended circuit. But the even orders are cancelled at the load of the differential circuit, leaving the 3rd as the primary and most significant distortion component.

At this point (just as if the 2nd were present) the ear's masking principle comes in. The 3rd harmonic masks the presence of the higher ordered harmonics. If the differential circuit is not perfect (and none are) there will be some even ordered content too, along with a 2nd to mask their presence. But over the decades I've been making differential circuits, over and over again its been the 3rd that is predominant.

The ear treats the 2nd and 3rd the same way. So we're dealing with a harmonic that adds warmth and bloom, but at a lower level than if there were a 2nd harmonic. But because of its relationship to the higher orders (its got more amplitude) its masks those higher orders.

The result is sound as smooth as any single-ended circuit, but considerably more neutral owing to less distortion (and so there is more detail too). When you get smoother sound and more detail, that's a plus in my book; its not brightness masquerading as detail, its the real thing.

I've not read Boyk and Sussman's works (if you have a link I'm happy to look into it) but if your interpretation of their work is represented here correctly and is in absolute form, I'm going to say that they missed something or are outright incorrect. What that 'missing' thing is has to do with real music that does not contain extra harmonic information other than those imbued to the instruments themselves. If the higher orders made by the reproduction are rendered inaudible as I described, then what is left is music, plus the 3rd. At that point our ear/brain system is not going to be worried about exponential (quadratic) decay of ascending orders of added harmonic content.

To be clear I'm not saying that SETs sound bad! But I am saying they are not and cannot be neutral. Specifically with regards to the OP I am responding to these bits of that post:

3.: Yes
5.: I hear more colorations in SETs than I do in modern horns. I own horns (Classic Audio Loudspeakers T-3.3 with field coil drivers) and I do that because they are some of the most neutral and revealing speakers I've heard. They operate at the same level as ESLs, but without the need for more power, with less sensitivity to humidity and with greater overall dynamic range.

https://www.its.caltech.edu/~musiclab/feedback-paper-acrobat.pdf

The masking of higher harmonics is only in relationship to the fundamental frequency and the further away from the fundamental you get the less it is masked by that that fundamental. 3rd order is not masking higher harmonics, nor is 2nd masking anything....in relation to the fundamental they are still miniscule. Studies have shown that 2nd order below 1% is essentially inaudible and this perceptability changes exponentially. That is why an exponential decay of the harmonics with increasing order is important because the autiblity is increasing exponentially with increasing order. This is why early attempts at an equation to determine this degree of audibility by the likes of D.E.L Shorter failed as they were only quadratic with increasing order and this was not steep enough to capture the real impact of audibility.

Cheerver and Geddes both came up with metrics that are more on point for the real impact of higher order harmonics. However, it also seems that the pattern matters. For example, an all odd harmonic progression that falls away exponentially will not sound the same as one that has all harmonic orders, even and odd. How the brain does this is anyone's guess at this point I would say. So, by cancelling 2nd (and other even orders, it won't be just 2nd) you will change the percerption of the sound...I would argue for the worse.

All amps that are feedback free (SS, OTL whatever) will have a steady increase in distortion with increasing power output. This is not a SET thing, it is a feedback/no feedback thing and then the same restrictions about using the power bandwidth and staying in the lower quadrant applies to all such amps.

Finally, it seems the SET has a greater dynamic headroom than other amp types in general (of course an amp can be engineered specifically for dynamic peaks and have a huge headroom...I think there was a Proton amp once that had nearly 6db of dynamic headroom). This was shown in a series of Stereophile articles by Peter Van Willenswaard where a 300B amp had 2.6x voltage headroom. PP tubes also had more than the SS amp on test but not nearly the 300Bs range. Also, you should well know that tubes act like compressors before actually hard clipping, which can have the perception of more headroom.

Of course bad examples abound but the best ones are, IMO, the best amps that can be had in the here and now. I know you can't come right out and say that you think your OTL designs are the best sounding amps on the planet but I am know you think it. There might even be some merit to your claims, but at least from my perspective, OTL is only my 2nd favorite amp type...I have owned and loved a couple of them...who knows, maybe I will get some Novacrons one day!

 

[Al M.]

I would say "horn-loaded" as the driver feeding the horn could be any of several different types, including conventional cones. Separately, a driver can be back horn-loaded, as in the Tannoy Westminster Royal GR.

I am much more inclined than most to use higher than typical power on high sensitivity loudspeakers. So I am answering your questions more from the point of view of horn loudspeakers than from the point of view of SET amplifiers.

I am answering these questions having in mind the very best and most ambitious speakers of each topology I have heard.


1. In which areas to SET/horn systems excel?

midrange realism, naturalness and presence

reproduction of the sounds of brass instruments

dynamics and "jump factor"

wider listening zone (no narrow "sweet spot")
Horn-loaded bass can sound very "fast."

An overall sonic presentation which is different from any other speaker topology in that horn speakers project the sound towards you in an energetic way that contributes to "aliveness," presence and realism.

2. In which areas can they not even be matched by other types of systems?

the reproduction of the sounds of brass instruments

I think there is something about the way horn-loaded loudspeakers reproduce the sounds of brass instruments which is consonant with the way those instruments themselves produce sounds.

If my primary musical genre interest were jazz, I definitely would have a horn speaker system. If my primary musical genre interests were jazz and classical, I probably would have a horn speaker system (unless I could have a Wilson XVX + Subsonics in which case it would be very difficult to make a decision).

3. Are there any sonic drawbacks to SET/horn systems, in the sense that there are other system types that just can do certain things better?

I find almost all contemporary design horn speakers to be a bit aggressive-sounding and fatiguing. The Tune Audio Avaton and the Cessaro Zeta and the Pnoe are my only three exceptions to this personal subjective experience. For example, I have yet to hear any Avantgarde or Acapella speaker system which I did not find a bit fatiguing.

I continue to feel that, for me, there is something slightly more transparent and in-the-room real about the way planar speakers reproduce vocals than any other speaker topology I have heard.

I think big cone speakers systems with a lot of driver surface area in the midrange and from the midrange on down can reproduce piano and big classical symphony orchestra at least as well as big horn speaker systems.

4. How much money do you have to spend on an excellent SET/horn system? How much money do you have to spend on a good SET/horn system?

I don't have the time at the moment to sit down and price out actual specific comparative systems. The point I wish to make here is that small horn speakers don't do the trick for me. I feel the same way about planars.

Unfortunately, with contemporary designs the horn loudspeaker system has to be big (and this inevitably means expensive) to achieve the sonic attributes I am attributing to them. (I have never heard from a 30" tall, two-way JBL or similar what I have heard from big horn systems.)

5. What about the question of the oft-mentioned horn colorations? Are they still a concern in modern horns? Are they a concern in vintage horns?

I don't know what you mean here by "colorations." (Is ruler-flat frequency response which sounds unnatural and unrealistic a "coloration"?)

I think some people refer to horn "colorations" as the horn "shoutiness" phenomenon or the "cupped hands" phenomenon. I don't hear either of these on the big horn loaded speakers I have heard.

I will define "coloration" here as frequency anomalies. Other than the Shindo system Keith took me to hear in Hollywood, which sounded to me like bourbon dripping down a warm brownie, I do not think horn-loaded speakers have any ubiquitous infirmity regarding colorations.

Thanks, Ron, for taking the time to answer all my questions!

 

[Ron Resnick]

Hi Ron,

Yes, you are right of course. SPL and watts scale like crazy...double the watts gives 3dB more SPL from the loudspeaker, so from 100w is 6dB louder than 25W, 9dB louder than 12W. So if we say 10% of amplifier output to pay the single-ended zero feedback tax then the 100W amp really is optimal to 10W or 95dB/w/m sensitive loudspeakers.

It is much more difficult to wind a decent output transformer for those mega voltage, high power single ended 211/833/845 etc. amps because of the larger gap to accomodate circa 1000Vdc and the crazy inductance for 100W @ sub 20Hz. These are great big transformers...not so great for mids and highs which are better served by smaller transformers.

To be honest, although I have listened to systems with these bigger SET amps I have never felt compelled to build any, for a number of reasons, some of which I mentioned above.

Anthony

Yes, the gigantic Trafomatics are basically all transformers.

Mixing amplifier topologies on the same loudspeaker system is, of course, controversial, but I love the idea of employing an SET amplifier to drive the midrange and treble drivers of a complex loudspeaker system, while a solid-state amplifier drives the woofers for the low frequencies. Taking the low frequency pressure off of the SET amplifier resolves some of the SET distortion issues, I believe.

 

[VladB]

Yes, the gigantic Trafomatics are basically all transformers.

Mixing amplifier topologies on the same loudspeaker system is, of course, controversial, but I love the idea of employing an SET amplifier to drive the midrange and treble drivers of a complex loudspeaker system, while a solid-state amplifier drives the woofers for the low frequencies. Taking the low frequency pressure off of the SET amplifier resolves some of the SET distortion issues, I believe.

Sorry for re-posting the video from another thread, but it works well as an illustrative example of this combo.

Here a Border Patrol 300B SET with 8W Takatsukis drives the 4-channel horn towers while plate AB amplifiers drive the 2X16 inch TAD bass horns.

300Bs works from 100 Hz upwards so do cover the mid-bass as well.

One might argue it is not the most organic way to drive a large system (different topologies as you say etc.), though I have not found a better solution so far to get the best out of this 5-way combination of compression and dome drivers/speakers.

 

[Gregadd]

It is sort of a chicken and egg argument. Does the horn necessitate the SET? Or does SET beget the horn? Probably the latter.

 

[morricab]

It is sort of a chicken and egg argument. Does the horn necessitate the SET? Or does SET beget the horn? Probably the latter.

Try big SS amps on 100+dB horns and see how you like the sound quality...most will move quickly back to SET...

 

[KeithR]

There has been a lot of enthusiasm expressed on WBF about SETs and horns. Some passionate advocacy in discussions might even cause an impression that there are only advantages to this type of systems, no drawbacks.

So apart from questions of practicality, such as size of horns, WAF and money to be invested, my questions are:

1. In which areas to SET/horn systems excel?
2. In which areas can they not even be matched by other types of systems?
3. Are there any sonic drawbacks to SET/horn systems, in the sense that there are other system types that just can do certain things better?
4. How much money do you have to spend on an excellent SET/horn system? How much money do you have to spend on a good SET/horn system?
5. What about the question of the oft-mentioned horn colorations? Are they still a concern in modern horns? Are they a concern in vintage horns?

3. deep bass is poor with SET (and most horns that don't have active bass). i also hear too much 2nd harmonic that distracts my listening - i close my eyes and it feels like distortion is leaping out at me, singers dripping in it in front of me. i much prefer push/pull amps which are more dynamic with better bass.

5. many horns have frequency anomaly (horn coloration, cupped hands) and integration woes. you pick your poison to get that efficiency and jump factor. also, they can be fatiguing and have spatial woes. but i will say horns as a speaker class are just as different as dynamic drivers - subsets of each have strengths and weaknesses. for instance, I don't hear horn coloration on JBL or AGs.

i don't think i would own horns unless i was fully ok with tube amps. SS has never sounded right on them and I think horns need tube decay to make the "whole note" sound correct. otherwise, too much leading edge.

horns will always be a niche business in audio. really only AG is a worldwide, Wilson type manufacturer over the past 20 years. Klipsch i guess is making a comeback with its more value priced Heritage line. i think people who only believe in SET amps will gravitate to them out of necessity, but if you don't require that than its not worth the drawbacks.

 

[acg]

Mixing amplifier topologies on the same loudspeaker system is, of course, controversial, but I love the idea of employing an SET amplifier to drive the midrange and treble drivers of a complex loudspeaker system, while a solid-state amplifier drives the woofers for the low frequencies. Taking the low frequency pressure off of the SET amplifier resolves some of the SET distortion issues, I believe.

It would often resolve some of the distortion issues for some of the reasons I talked about earlier.

Personally, I have gone to quite some lengths to have a SET amp also drive my lowest bass channel <100Hz. I paralleled 8 x 10" Scanspeak drivers a side in individually sealed boxes (alas no room for a bass horn) which presents a 1.1ohm load to the amplifier, had a custom output transformer wound that a very smart man was actually able to design and build, which present the full 9w power at 6Hz -1dB. It's a beast. The transformer, being such a high inductance unit made on a huge core actually will transmit frequencies up to about 5kHz before rolling off. The design brief was to get a frequency response of 10Hz to about 1kHz at full power but it does a few octaves better than that. Still, given how it is built it would probably sound rubbish at those higher frequencies...those are much better served with smaller transformers that do not need to transmit the bass frequencies nor that much power.

 

[Al M.]

3. deep bass is poor with SET (and most horns that don't have active bass). i also hear too much 2nd harmonic that distracts my listening - i close my eyes and it feels like distortion is leaping out at me, singers dripping in it in front of me. i much prefer push/pull amps which are more dynamic with better bass.

5. many horns have frequency anomaly (horn coloration, cupped hands) and integration woes. you pick your poison to get that efficiency and jump factor. also, they can be fatiguing and have spatial woes. but i will say horns as a speaker class are just as different as dynamic drivers - subsets of each have strengths and weaknesses. for instance, I don't hear horn coloration on JBL or AGs.

i don't think i would own horns unless i was fully ok with tube amps. SS has never sounded right on them and I think horns need tube decay to make the "whole note" sound correct. otherwise, too much leading edge.

horns will always be a niche business in audio. really only AG is a worldwide, Wilson type manufacturer over the past 20 years. Klipsch i guess is making a comeback with its more value priced Heritage line. i think people who only believe in SET amps will gravitate to them out of necessity, but if you don't require that than its not worth the drawbacks.

Interesting, Keith, thanks.

As for the 2nd harmonic distortion of SETs, I haven't heard it yet. I have only experience with a Tron SET amp, and another vintage SET amp, on open baffle Bastanis speakers (100 dB sensitivity). But perhaps there any such distortion was masked by a slightly bright and thin tonal balance (the owner reportedly has toned down the balance since). Or, as others have said on this thread, the distortion depends on how hard the SET is driven.

As for horn colorations on JBL, following your comments I dug a bit into this. It may depend on which JBL speaker. An otherwise relatively positive review of the JBL S3900 horns on 6Moons states:

Speaking about distortion, I'd like to mention how the horns affect the S3900. Let's not pretend they don't. Part of the crossover range between 800-900Hz is emphasized. This range is responsible for midrange body and so-called presence. The JBLs didn't sound aggressive in that they did not irritate with these transients. Even so the lower range of female vocals and part of the violin's bandwidth especially when recorded slightly hot as by Deutsche Gramophone were somewhat elevated and s nasal. There was no trace of brightening or glassiness however. The Americans behaved coherent and well thought-out to embarrass many an expensive hornspeaker.

The change in tonality emphasized the violin, enlarged it and accented its nasal characteristics. This happened to Gil Sham’s instrument on Paganini For Two and Hilary Hann’s on Bach Concertos even though the former was released by JVC as XRCD24, the latter by Universal Japan as SHM-CD (in other words two theoretically top CD formats). In part this was the result of certain DG recording conventions but also it was due to the JBL character which deviates from neutrality in this particular range.

It was different with lower-pitched string instruments on Lachrimae or Seven Tears by Jordi Savall and Hespèrion XX. The latest Alia Vox sounded equally thick and accented on the lower midrange as the Haden album. This was due to another JBL trait, namely their ability to build a full mature midrange particularly for male vocals and instruments having a similar tonal centre. Playing Nat 'King' Cole, Dominic Miller’s Fourth Wall or even Mike Oldfield’s Tubular Bells on the latest Platinum SHM-CD release proved convincing enough. If you don't own these discs, use others characterized by a saturated lower midrange that's 'opened up' by a treble that won't allow it to close off in a thick pulp. What you get is something that happens with the best of warm tube amps and fast speakers: the unity of tonality and micro dynamics, tangibility and soundstage depth.

On the other hand, the dynamics are described as stellar in another review:

Trumpets and other brass instruments really come alive on horn speakers, you haven't lived until you hear Miles Davis over a decent pair of horns. His expressive trumpet over the S3900 was a huge leap in breaking the sound reality barrier! It's not just louder, or lower in distortion, you feel more of what Davis was pushing through his trumpet.

Drums impact and shading of dynamics are superior over the S3900. Not just loud drumming, the nuance of a cymbal shimmer or the tautness of the head of a tom-tom: the S3900 puts you in touch with those sounds.

With Robert Plant and Alison Krauss's brilliant Raising Sand, the S3900 has me rethinking the sound of that album. I've always liked the music, but the sound was opaque and hazy, now it feels more direct. Hearing those two singers pushing each other is a real thrill, and I never felt that way before!

The 4429 and S3900 are extraordinary speakers, they can humble many far-more-expensive audiophile speakers in terms of dynamics and "feel it in your gut" realism. If you've never heard what some of the better horn speakers can sound like you're missing out.

I'm in no way saying that horns sound better in every way, just that they're more fun to listen to than box and panel speakers. The best box speakers are smoother sounding, with flatter frequency response, punchier bass and their treble may be less aggressive than the horns I've heard. Box speaker imaging is also superior, and better focused.

***

So I guess, pick your trade-offs.

In any case, the JBL S3900 do not seem to be able to be driven by SETs, given their 92 dB sensitivity.

 

[Atmasphere]

Is there no cancellation in SET configuration, and is that a reason why the transformer cannot have high power handling and wide bandwidth at the same time?

Yes. The current through the transformer causes a magnetic field. At some point the transformer saturates, which is to say that the field is as strong as it can get. In a push-pull amp, the two power tubes draw current in an opposing manner, and so the two resulting magnetic fields are in opposition, which is to say that they cancel. This allows for a lot more audio power and wider bandwidth.

SET's are ridiculously dynamic when paired with the correct speaker, but it is not the type of flawed dynamics that you correctly describe when overdriven, they are no-compromise dynamics.

IMO/IME dynamics are a thing of the recording. The electronics should not impart any of their own.

(b) Run the amp at a low percentage of its potential power output. This means very high sensitivity loadspeakers.

I've been telling people for years that 20-25% of full power is going to be the upper limit on most SETs, so you need to plan accordingly with the loudspeakers (meaning horns are very likely in the mix) if you want to get the most out of the amp.

So, by cancelling 2nd (and other even orders, it won't be just 2nd) you will change the percerption of the sound...I would argue for the worse.

In practice it works the other way 'round. I have heard this trope a lot, but as best I can make out, its a myth. Thanks for the link BTW; I read the article and it does not seem to address my comments. In the paper there is some comment about using degenerative feedback in the differential amplifier. I used to do that decades ago but found that it really didn't work. In a differential amplifier you want as much differential effect as you can get; adding degeneration resistors runs counter to that. As you get more and more differentail effect, distortion goes down, gain goes up (to the limits of the tube) as does bandwidth. This is quite the opposite of what was described in that paper, but they seemed to only use one tube type. I suspect that they didn't have a really effective constant current source either, and IME that is crucial if you want the circuit to perform. Failing that you leave a good deal of performance on the table. Years before that paper was written, we were using CCS circuits that had no more drift than 17 parts per million even when the incoming AC voltage was varied from 105 to 125Volts and used in an otherwise unregulated circuit. Since then we've improved the CCS by 2 orders of magnitude.

All amps that are feedback free (SS, OTL whatever) will have a steady increase in distortion with increasing power output. This is not a SET thing, it is a feedback/no feedback thing and then the same restrictions about using the power bandwidth and staying in the lower quadrant applies to all such amps.

While this is true, it might be interesting to see how this pans out in real life. With almost any SET, 10% THD is typical at full power. With our amps (which are an example of a fully differential zero feedback triode amplifier) you'll typically see between 0.5% to 1.0% THD at full power; keeping in mind that full power might be 60 watts or 140 watts or even more. As you point out, as power is decreased so does the distortion component; but in an amplifier that is fully differential (and uses the differential circuitry to drive the output tubes) the distortion does not rise below a certain minimum power as seen in a lot of push-pull amplifiers (none of this is addressed in that paper you linked FWIW). So by the time its making the same power as an SET, the distortion is vastly lower- hence, also more neutral and musical. This is why our amps work well on horns despite often having much more power than needed in many cases. IOW its still all about that first watt.

Finally, it seems the SET has a greater dynamic headroom than other amp types in general (of course an amp can be engineered specifically for dynamic peaks and have a huge headroom...I think there was a Proton amp once that had nearly 6db of dynamic headroom).

Headroom in a class A amplifier should be 0dB. Class AB1 amps should have more headroom and class AB2 even more. So I have to assume that the word 'headroom' is being used in an alternative way which hasn't been defined for amplifier measurement. A class A amp has 0dB of headroom because it does not matter if you run it at full power for a while or if you give it a transient peak at full power, it will make the same power. Class AB amps might make more when given a big transient, due to fluctuations in the power supply (which does not happen in a class A1 amplifier).

As for the 2nd harmonic distortion, I haven't heard it yet.

If you heard an SET, you also heard the 2nd harmonic. The ear tolerates quite a lot; in the 1960s GE demonstrated that listeners would not object even if there was 30% distortion if it was only the 2nd harmonic.

 

[Al M.]

Yes. The current through the transformer causes a magnetic field. At some point the transformer saturates, which is to say that the field is as strong as it can get. In a push-pull amp, the two power tubes draw current in an opposing manner, and so the two resulting magnetic fields are in opposition, which is to say that they cancel. This allows for a lot more audio power and wider bandwidth.

Thanks for the great explanation, Ralph, very helpful.

 

[Ron Resnick]

i don't think i would own horns unless i was fully ok with tube amps. SS has never sounded right on them and I think horns need tube decay to make the "whole note" sound correct. otherwise, too much leading edge.

+ 1

 

[microstrip]

IMHO each of these items should be a

(...)
I find almost all contemporary design horn speakers to be a bit aggressive-sounding and fatiguing. The Tune Audio Avaton and the Cessaro Zeta and the Pnoe are my only three exceptions to this personal subjective experience. For example, I have yet to hear any Avantgarde or Acapella speaker system which I did not find a bit fatiguing.
(...)

I don't know what you mean here by "colorations." (Is ruler-flat frequency response which sounds unnatural and unrealistic a "coloration"?)

I think some people refer to horn "colorations" as the horn "shouty-ness" phenomenon or the "cupped hands" phenomenon. I don't hear either of these on the big horn loaded speakers I have heard.

Ron,

Admitting they do not suffer from "colorations" as you say, what should be the real reason for, in your opinion, most modern horns sounding "a bit aggressive-sounding and fatiguing"?

 

[morricab]

Yes. The current through the transformer causes a magnetic field. At some point the transformer saturates, which is to say that the field is as strong as it can get. In a push-pull amp, the two power tubes draw current in an opposing manner, and so the two resulting magnetic fields are in opposition, which is to say that they cancel. This allows for a lot more audio power and wider bandwidth.


IMO/IME dynamics are a thing of the recording. The electronics should not impart any of their own.


I've been telling people for years that 20-25% of full power is going to be the upper limit on most SETs, so you need to plan accordingly with the loudspeakers (meaning horns are very likely in the mix) if you want to get the most out of the amp.


In practice it works the other way 'round. I have heard this trope a lot, but as best I can make out, its a myth. Thanks for the link BTW; I read the article and it does not seem to address my comments. In the paper there is some comment about using degenerative feedback in the differential amplifier. I used to do that decades ago but found that it really didn't work. In a differential amplifier you want as much differential effect as you can get; adding degeneration resistors runs counter to that. As you get more and more differentail effect, distortion goes down, gain goes up (to the limits of the tube) as does bandwidth. This is quite the opposite of what was described in that paper, but they seemed to only use one tube type. I suspect that they didn't have a really effective constant current source either, and IME that is crucial if you want the circuit to perform. Failing that you leave a good deal of performance on the table. Years before that paper was written, we were using CCS circuits that had no more drift than 17 parts per million even when the incoming AC voltage was varied from 105 to 125Volts and used in an otherwise unregulated circuit. Since then we've improved the CCS by 2 orders of magnitude.


While this is true, it might be interesting to see how this pans out in real life. With almost any SET, 10% THD is typical at full power. With our amps (which are an example of a fully differential zero feedback triode amplifier) you'll typically see between 0.5% to 1.0% THD at full power; keeping in mind that full power might be 60 watts or 140 watts or even more. As you point out, as power is decreased so does the distortion component; but in an amplifier that is fully differential (and uses the differential circuitry to drive the output tubes) the distortion does not rise below a certain minimum power as seen in a lot of push-pull amplifiers (none of this is addressed in that paper you linked FWIW). So by the time its making the same power as an SET, the distortion is vastly lower- hence, also more neutral and musical. This is why our amps work well on horns despite often having much more power than needed in many cases. IOW its still all about that first watt.



Headroom in a class A amplifier should be 0dB. Class AB1 amps should have more headroom and class AB2 even more. So I have to assume that the word 'headroom' is being used in an alternative way which hasn't been defined for amplifier measurement. A class A amp has 0dB of headroom because it does not matter if you run it at full power for a while or if you give it a transient peak at full power, it will make the same power. Class AB amps might make more when given a big transient, due to fluctuations in the power supply (which does not happen in a class A1 amplifier).


If you heard an SET, you also heard the 2nd harmonic. The ear tolerates quite a lot; in the 1960s GE demonstrated that listeners would not object even if there was 30% distortion if it was only the 2nd harmonic.

You seem to have missed the point that the data in the paper is all from computer models with ideal devices and voltages etc. This means they had a generic triode, generic MOSFET and generic BJT. THe voltages would be perfect because they are computer simulations. Have a look again as they show the circuits they are modelling.

"With almost any SET, 10% THD is typical at full power. " No, this is completely depending on what you are defining as "full power" since clipping is usually soft there is no hard knee to say "Ah, there it is clipping". Many of them just keep rising in the same more or less linear manner with power. For example, I had a parallel 300B amp (JJ-322) that was rated at 20 watts. At that power it made about 3% THD (based on measurements in a Polish magazine). If they had defined their rated power based on 1% THD, then the amp would have been rated at around 15 watts instead. It all depends on what is the AUDIBLE onset of clipping. Probably you would not hear if it hit 1% distortion (it had a very nice looking FFT spectrum... I probably should have kept the amp as it had quite good output iron...huge double C cores...probably 6kg each) but probably it would start to get a bit fuzzy sounding at 3%...probably not objectionable yet but probably noticeably less pure sounding. Bass was tight, controlled and quite deep because the core saturation of the transformers was minimal.

My current SET, the Aries Cerat Genus, is an enormous amp (70KG) with vast power supply reserves (courtesy of super capacitors that have 1/100th the ESR of normal electrolytics...so it delivers that current fast). It uses around 12Kg output transformers so there is no core saturation to speak of (bandwidth 11-60Khz) and bass is deep and powerful. Distortion is very low for a SET at the normal power levels one would use for normal listening levels with sensitive speakers (mW to say 5-10 watts). Dynamic power is almost double the rate power. Transparency and lack of coloration is the name of this amps game.

"So by the time its making the same power as an SET, the distortion is vastly lower- hence, also more neutral and musical. This is why our amps work well on horns despite often having much more power than needed in many cases. IOW its still all about that first watt."

That depends on the SET. At 1 watt my amp make -62db (0.08%) 2nd harmonic and -73db (0.02%) 3rd and the rest fades gradually down into the noise floor (a true noise floor as there is no negative feedback to create other harmonic components that are signal correlated). I would argue that -62db of 2nd is utterly and completely inaudible. You are right though that lesser SETs will have significantly higher distortion than this...even some very expensive ones.

Your MA-1, as measured by Soundstage shows about the same distortion at 1 watt for the third harmonic (around 0.08%) as my amp and your 2nd is similar (around 0.01%) to the third in my amp. The problem I see is that the 4th and 5th are also at about 0.01% and the 6th and 7th are still well above the noise floor. I have extrapolated the values thus: The power vs. THD curve in the review shows that the THD at 1 watt is about 5 x less than the distortion at 10 watts (0.08% vs. 0.4%). The harmonic spectrum data was taken at 10 watts. So I took those values and divided by 5 to get the 1 watt result. The other problem I see is that the distortion vs. power flattens out a 1 watt. It doesn't keep going down with less power below 1 watt. In this a good SET will do...it will keep going down as the power goes down.

Here is an interesting amp from a designer that I know can design good stuff (I have had his DACs for over 20 years now).
SoundStage! Measurements - Monarchy Audio SE-160 Mono Amplifiers (12/2003) (soundstagenetwork.com)

A hybrid amp that has a near exponential harmonic distortion decay and a perfectly linear distortion vs. power curve. It almost out SETs a SET. I may try to find one some day...

I am not saying your MA-1 from that time (I guess you have improved them since then) is better or worse than my amp but you can't say the measurements are necessarily better.

This one sounded horrible: SoundStage! Measurements - PS Audio HCA-2 Stereo Amplifier (9/2002) (soundstagenetwork.com). Took it on a trade in...made a great subwoofer amp .

This one I thought would sound great...but was somehow boring...SoundStage! Measurements - Einstein Absolute Tune Integrated Amplifier (10/2007) (soundstagenetwork.com). Sold it after trying for about a year with it.

This was a nice sounding amp the couple times I heard it: SoundStage! Measurements - Tenor Audio 75 Wi Mono Amplifiers: Measurements (2/2002) (soundstagenetwork.com)

In response to the headroom argument read this series of articles: Tubes Do Something Special | Stereophile.com
Note the SET had the biggest instantaneous power of the amps used...that is no means exhaustive or conclusive but it does show they can go way beyond their rated power without audible issues...perhaps class A2?