Gave up my SET----looking for moderate to higher power replacement options

It does not sound like a set amp. Only set amps sound like set amps. Solid state amps are only going to be an approximation, but leave you lacking.

A push-pull tube amp with a single-ended input will exhibit a prominent 5th harmonic in addition to the 2nd and 3rd. Mathematically, this is the result of this type of amplifier having both a quadratic and cubic non-linearity. So it will tend to favor odd ordered harmonics to a certain degree, and higher ordered harmonics will fall off at a rate similar to that of an SET.


This reinforces the complaints that SET lovers often level at Push-pull amplifiers. Its nice to know that the math and measurements support what the subjective camp (SET lovers) is saying in this case.


However its important to understand that an entirely differential push-pull tube amplifier expresses a cubic non-linearity, which IME is even easier to listen to than SETs. I think this is the bit that the SET crowd misses and its not surprising because I’ve yet to see any designer of SETs talk about a quadratic non-linearity.


What is important here is that the subjective camp, theoretical mathematics and practical mathematics are in agreement about what we experience as audiophiles.


I don’t like to take sides on the objectivist/subjectivist debate. I see it as a bit ignorant; in particular when objectivists talk about ’being measurable’ they usually don’t have a clue about what to measure (although they are otherwise correct). As a result for the last 40-50 years they’ve been pushing amplifiers that have increasing distortion as frequency is increased and so sound bright and harsh, especially when driven to higher output levels (and this due to the simple fact that these amps employ feedback in insufficient amounts). That’s not objective IME; its simply expressing poor taste.

2. Hopefully Ralph will weigh in but unfortunately I suspect @atmasphere is referencing his own Class D amplifier that is currently under development. I'm sure it'll be wonderful, but not yet available and perhaps outside your budget.

We are working on a class D amplifier which has reached beta production. Its intended pricepoint is about 5000.00/pair. Although I think our amp is doing quite well as far as being musical  is concerned, I was not talking about it specifically. I was more talking about how feedback is troublesome if improperly used (which is about 99% of the time). The Benchmark is an example of an amp that does feedback properly. Self oscillating class D amps are another example.


Feedback has gotten a bad rap and deservedly so because if you don't run enough of it (and that's been the case in the last 60 years), it will generate higher ordered harmonics and IMD at the feedback node (unless you are very careful about the latter). This causes harshness and brightness, and this is the main reason transistors are harsh and bright. If you run feedback in tube amps, they get harsher and brighter too (and this will be in spite of the fact that the frequency response is actually **flatter**, because the brightness is caused by distortion, not a frequency response error).

So we've seen designers, including myself, building zero feedback amplifiers to prevent this brightness and harshness. You can think of the use of feedback as being on a bell curve, use a tiny bit and its not bad but doesn't do a lot, use more and the distortion issues increase (all the while suppressing the innate distortion of the circuit, so it will measure well but not sound as good as the specs suggest). But finally you get to the other side of the curve, which is at about 35-40dB (meaning that if the amp had no feedback, it would have to have at least 65-70dB of gain!). At this point there is so much feedback that it allows the amp to compensate for the distortion generated by the feedback, and corrects phase shift too.


One thing I forgot to mention is how distortion can increase at higher frequencies if feedback is insufficient. You really want the distortion to be the same regardless of frequency. Our OTLs can do this, which is part of why they are musical, and the smaller the SET, the more they can do that as well (this is part of why the lower powered SETs are held in such high esteem). To cover up this problem, the distortion specs you see are usually done at 100Hz! If you measure the distortion spectra at 1KHz you get far more meaningful information- look at what the amp is then doing at 7KHz, which is in the area where the ear is most sensitive (birdsong frequencies). The 7th harmonic is unpleasant! If you measure at 100Hz, the 7th harmonic can often appear benign. This is why many solid state amps seem to do bass just fine, because at bass frequencies, they really **are** doing just fine because there's enough feedback. But as frequency is increased, the amp can run out of Gain Bandwidth Product, which is a way of saying that the feedback is being reduced. This is a very common problem!


My experience with monostrapping and bridging amplifiers isn't good. The amps almost always get less musical (higher distortion). Lower impedance loads can be an issue as well because the current expected of the output devices might be exceeded, such as into 2 ohms- that leads to failure and shipping costs...


I miss Al as well. He was a friend, a good egg and always steady on.

I’ve heard the Benchmark, Mola Mola and Soulution amplifiers (Examples of High NFB circuits) no SET sonic characteristics in my opinion. I don’t deny that many listeners loves these particular amplifiers and in fact I could be in the minority.

@charles1dad There is a reason that I mentioned the bit about the designer paying attention to human hearing rules (I didn't put it in quite those terms). If he's not careful, and does not permit the lower orders to mask the presence of the higher orders, the amp runs the risk of sounding dry. Now if you get the distortion low enough, this phenomena is reduced, but you have to get really low distortion numbers! 0.0001, that sort of thing. So if for whatever reason you can't get there, then you have to be sure that the lower orders mask the higher orders.

To give you more insight on this, almost any solid state amp has less higher ordered harmonic distortion than almost any SET. But the SET has a huge 2nd order by comparison, which tends to mask those pesky higher orders.


(Here's a fun bit: amps that make the 2nd ordered harmonic as their primary distortion component have what is known mathematically as a Quadratic Non-linearity. Amps that have a 3rd as the primary distortion have what is known as a cubic non-linearity. The difference is, if based on the 2nd the distortion harmonics do not fall off as quickly as they do if the 3rd predominates. Since the 3rd is treated the same by the ear as the 2nd (contributes to warmth and 'bloom') an amp that is based on the 3rd will still have a musical presentation, but otherwise will be perceived as being more neutral since there is less coloration overall (the 3rd will be slightly less than it is in an amp based on the 2nd harmonic) due to distortion. This is why I built entirely differential amplifier circuits, as the simple fact that they cancel even-ordered harmonics in each stage throughout the circuit caused it to have its major distortion component as the 3rd. Since distortion is not compounded from stage to stage, the higher orders fall off at a faster rate as the order is increased.)


Obviously the brightness and harshness of traditional solid state that we've heard over the last 50 years (and why tubes are still around) is a coloration too. I don't agree that we hear all that differently: all humans use the higher ordered harmonics to sense sound pressure and all human's ear/brain systems assign the same values of brightness and harshness to these harmonics when they are distortion. IME people express these differences out of expectations of what a stereo should be, if not sounding real. When you ask for the music to sound real, then (again IME) these differences tend to go away and people start hearing the same things.

It was just more detailed with subtleties that increased realism beyond anything else.  That's what I'm asking about in a possible Class D replacement.

Speaking from experience, the right class D amp can do that. Class D has as many variations as any other type of amp does, so just because one class D might not bring home the bacon says nothing about another.

If your used to your SET tube sound, no class D will ever match the tone.

Again, speaking from experience, I've not found this to be the case. The main differences we hear in amplifiers are two-fold: what is the distortion signature, and how flat is the frequency response.

SETs generate a lot of distortion. At full power its typically 10%. If you're really serious about using one, your loudspeakers should be efficient enough that the SET is never asked to make more than about 20-25% of full power (this will limit the higher ordered harmonic distortion that causes them to sound 'dynamic'). It should be noted that the dynamic nature of music should come from the recording, not the amplification!

Now the advantage of SET is that as power is decreased to zero, the distortion drops to unmeasurable. This is the source of that 'magical inner detail' that so many talk about. Put simply, its that First Watt that is actually very good (unless you have a type 45-based amp, in which case it will be that first 100th of a watt).


But SETs by no means rule the roost in this regard. Our amps have the same property of linearly decreasing distortion. Our amps are push-pull; that sort of quality in a push-pull amp is rare! But there isn't some sort of pixie dust that allows for it; you simply have to be aware of how amps make distortion, and what distortions are going to be heard by the ear. Any zero feedback tube amplifier that is fully differential from input to output will have this quality.


Let's talk about the lower ordered harmonics for a moment. Its the 2nd order that makes SETs have that ever-lovin' tube sound- it give the amp a rich, warm quality and assists the ear/brain system in winnowing out detail and soundstage (that latter bit is IMO something that could do with more research as to why this is so). To this end, the 2nd harmonic is fairly innocuous. Other than making the amp a bit more euphonic, you don't hear it.


The 3rd harmonic has this same quality- the ear treats it the same way. The presence of the 2nd and 3rd in sufficient quantity can mask the presence of higher ordered harmonics to a certain extent. This is part of why tube amps sound smoother than a lot of solid state. But that isn't the entire picture.

The ear/brain system assigns a tonality to all forms of distortion. The lower orders mostly richness, the higher orders cause harshness and brightness even in tiny amounts. This is because the ear uses them to sense sound pressure and so has to be keenly sensitive to their presence.


This is why many solid state amps have sounded bright- its entirely due to the higher ordered harmonics making themselves known.


The problem has been feedback in amplifiers. If used in insufficient quantity, it will add distortion of its own while suppressing the innate distortion of the circuit in which its used. But- if you use **enough** feedback this will allow the amplifier to compensate for the distortion caused by feedback itself. Nelson Pass comments on this in an article he wrote about distortion. In it he speculates that you'll need more than 70dB of open loop gain (this is how much gain the amp has if there is no feedback). This is a very difficult number to achieve in traditional solid state amps since oscillation is a clear and present danger due to a phenomena known as phase shift. How that works is that at some frequency above the audio band, the phase shift of the amplifier starts to go up. When it passes about 90 degrees, the amp can use the feedback as positive rather than negative- and so oscillation can result. So many designers don't push their luck.

But Luck has nothing to do with it. Sound engineering does, and some designers have succeeded in making amps of this type. The Benchmark is one example.  There is another way to get around this problem. That way is to use a class D amplifier and use so much feedback that the amp goes into oscillation. But in this case, the oscillation is used as the switching frequency of the amplifier, so we're OK with that. Such an amp can have north of 40dB of feedback, allowing it to compensate not only for distortion caused by feedback but also phase shift caused by the output filter of the amp.


Once either of these approaches are achieved, its then up to the designer to understand how the ear senses distortion and how the ear's masking principle works. If he does this right, then he will see to it that despite the resulting low distortion of the amp, the primary distortion components are still the lower orders.


If he does this, the resulting amp will be smooth like tubes and will be extremely detailed at the same time. So you can get that 'magic'. The only downside is that any amplifier that does this will have a hard clipping character, so it will be important to have more than enough power to do the job- IOW you *never* clip the amp!


The flip side of this is to avoid using any feedback at all, relying on other techniques to control distortion. Do there you have it. You can have an SET which has no feedback, or a class D which uses a ton of feedback, and the two can sound surprisingly similar (although IME you'll hear more detail from the class D, and of course it will have more power)!

And despite the modest power rating, this amp had startling dynamics.

Because the ear uses higher ordered harmonics to sense sound pressure, and because this amp was really being pushed to hard to drive this speaker, higher ordered harmonics were showing up on the transients where the power is. IOW, it was distortion masquerading as 'dynamics'.


The mark of a good system is that it does not sound loud even when it is. This will be because distortion is controlled.


SETs make a particular type of distortion known mathematically as a Quadratic Non-linearity. This results in the primary distortion component being the 2nd harmonic followed closely by the 3rd. This gives you a rich sound, but again its due to distortion. The only class D amps that make that kind of distortion are zero feedback designs. You'll have to ask if they have a quadratic non-linearity; if they don't know what you're talking about don't buy the amp even if you don't know what that is either :)

Amps that make a 3rd harmonic might be interesting to you. The ear treats the 3rd the same as the 2nd, but amps that make the 3rd harmonic as the primary distortion component tend to be overall lower distortion. If you want the musicians to sound real this is a better way to go. Such an amp will still have some romance- it won't be dry.

Have you considered other tube amplifiers?