SET the best?

This is issue I see today with SET in general, some of these guys run higher plate voltage then output tubes can handle, too many instances of premature tube failure. People too often blaming tube manufacturers. Prior to purchasing any SET, I'd research this aspect of that particular amp, know what your in for.

Its probably also a good idea to get some idea of what the tube is rated for. For example a 300b in class A1 (A2 an even A3 are possible modes of operation as well) will be good for 7 Watts. A type 45 is 0.75 Watts. When you get stories of greater power, how are they doing that? Is it by using a variant (there's a graphite plate version of the 300b called the 6300b for example) or are they running plate current (class A2 or A3)?

How anyone can say they know what the tube sounds like when the design of the output transformer easily trumps the tube is beyond me. All you can really say is how that tube sounds with that output transformer.

I was refering to general charachertistics f the 845,

. I’ve read good things about the 2A3,, then here we read the 845 is simply gorgeous. Another comment likes the 300 as driver stage tube,whereas my tech says avoid 300B in any part of the design. He has no sympathy for the 300B, and yet some big manufactuers employ the 300 as driver tube. I guess as with all amp designs everyone will have his particular preferences.

@mozartfan You’re making my point. Its not the tubes, its the transformers. IMO the tubes are all sounding the same- they have similar linearity. But the output transformers vary dramatically!

I have a set of Coral Betas you can have for the cost of shipping. One of the drivers has a damaged voice coil. These are the drivers in Coral cabinets with the passive radiators.

 

how would you describe the sound of a 845 tube?

You can’t- any more than you could a 300b or type 45, because the output transformer and the individual tube’s match to it plays such a critical role.

People often say the tube has a certain sound. But with all the differences in transformers there's no way they could possibly tell.

I can't imagine any benefits from class D for high sensitivity speaker.

I can!! Low noise so low its hard to hear even with your head in the mouth of the horn- and greater detail/better 'focus' so images in the rear of the sound stage are easier to hear- and overall every bit as smooth- in fact smoother.

When you combine smoothness and greater detail at the same time that's when you're getting somewhere.

There are so few threads focused on SET amps....Is it too much to ask that the AGD and GAN and Class D discussions be moved to their own thread or one of many already in existence?

@david_ten There are also a good number of SET threads. The comments are relevant due directly to the question asked in the OP. It appears that if the answer were 'yes' at one time that SETs are rapidly being eclipsed, even by class D amps let alone other tube topologies.

It pointed out that “all” other current GaN based amps use existing modules with chips that are currently used in non Audio devises like power chargers.

@mglik If he's getting semiconductor devices made for him I can see that happening. Usually you have to buy quite a lot of such devices to get any semiconductor manufacturer to pay attention to your order! But he might have an in with one of his former employers. But FWIW the chips used in switch mode power supplies have a lot in common with the requirements you face in building a class D amplifier.

The GaN module used is the only such chip developed specifically for Audio.

AGD has just come out with a stereo amp called the Tempo. It uses a aluminum chassis cut from a solid billet. It is a more conventional design without the “tube” but utilizing the same technology. I believe the price is $5400.

Thus it differs from any other GaN based amps. I believe my explanation is accurate.

@mglik Its not. The first sentence in the post above isn’t right. A ’GaN module’ isn’t a chip. Its a circuit board, likely with chips on it. And none of them are specifically designed for audio, although the module is, but its not the only module designed for audio.

The statement that it ’differs from any other GaN based amps’ is tricky. All GaN based amps are different, so technically the statement is true. But installing a GaN based module in an aluminum chassis is commonplace.

This is not to denigrate Alberto’s designs- far from it. He was involved with GaNFET manufacturing and knows how to use them very well. But no GaNFET is built specifically for audio.

 

All I was trying to say is that the 845 hasa good reputation for delivering the goods.

My tech points in the 845 direction for what he feels will give me best bang.

He also likes most other SET tubes.

Liking a tube is very different from how it can be made to perform 😁

The simple fact is the more power the SET has, the more limited its bandwidth. This is because the higher the power the output transformers designed for SETs are, the more limited their bandwidth. Its inescapable (except in marketing literature).

The more limited the bandwidth, the more phase shift is introduced (unless you have a mess of feedback to control it; not possible with a tube amp). Phase shift messes with the soundstage; not saying you won't have one, but it will be more palpable if the phase shift isn't there.

That is why the 300b was King 30 years ago, why the 2A3 supplanted it 20 years ago and why the type 45 is reigning now. So we went from 7  watts to 4 and from 4 to 0.75Watts. Compared to these tubes in terms of the sound quality you get, the 845 is fun but not in the running.

Speaking in very broad, generalized, terms, I tend to find pushpull amps to have a "tighter" more punchy sound than SET amps.  But, too much of that quality and the sound seems a bit artificial or "mechanical" and less natural than the more relaxed sound of good SET amps.  For me, for example, KT88 pushpull amps tend to be particularly likely to sound too punchy and mechanical for my taste, but, as with everything in audio, there are always exceptions. 

I agree- I tend to see punchy bass as a coloration, since you never seem to run into it in live events. I've played in orchestras, jazz and folk ensembles, rock bands and just not encountered it. I think its the product of an overdamped speaker to a certain degree but I don't think that's all of it.

The distortion signature is what is at play here IMO. Most pentode-based amps employ a bit of feedback, but since they really don't have enough, they have an obvious distortion signature (the amplifier's 'sonic signature'). Some of that includes harmonics of bass fundamental tones, generated by the feedback itself. The ear is interpreting them as 'punchy' (and yes, this happens with many solid state amps too). IMO/IME the way around this is to make sure the amp is as linear as possible, so much so that it will sound reasonably decent with no feedback at all. Only then do you apply the feedback- and then it should be used to control gain more than anything else. With a pentode-based amp I suspect this would mean that the output section is wired Ultra-Linear.

But many pentode output stages really aren't that linear and need the feedback to linearize them. That is a bad move IMO. You can't get rid of all the distortion and if you start with a high distortion amp, the feedback will cause it to have a less musical distortion signature overall when you're done. Amps that have zero feedback have a different distortion signature that tends to favor the lower ordered harmonics which will mask the presence of the higher orders. But if you have no feedback you also have a much more limited range of speakers that are practical with the amp. For more on this see

The Power Paradigm Vs the Voltage Paradigm

 

My 300B amp measures flat well below 30hz.

To play the bottom octave properly you need bandwidth to 2Hz. This is because once the amp starts rolling off, it does so at 6dB/octave. When you have a slope like that it causes phase shift to be present up to about 10x the cutoff frequency. This is basic filter theory. So 2Hz bandwidth allows for 0 degrees of phase shift at 20Hz.

The ear perceives phase shift as a tonality. In the case of bass, its perceived as a lack of impact, as a thinness in the sound.

In the case of a 30Hz rolloff, the phase shift will go as high as 300Hz. As a side note, speakers are not part of this phase shift issue- its something that applies to electronics.

You can get around this problem if you have enough feedback. If you have enough, the feedback can correct for phase shift.

That kind of feedback isn’t available to SETs or for that matter, most tube amps. So bandwidth is the only way to solve the phase shift issue.

AREN‘t you tooting your horn just a little too much? OTLs are great but only address a very small portion of speakers with high efficiency and high impedance. That may or may not be everybody‘s fare and your implied claim of superiority seems therefor to run precisely into the Western gun issues you profer.

I've been careful to not speaker about our amps specifically. But for the record, although over the years it seems as if a lower powered OTL is somehow the holy grail for many, IME such a thing is really difficult to do and impractical. Its quite true that OTLs don't like low impedances but you don't always need a high efficiency speaker to work with them because many OTLs made over the last 60 years made over 100 watts. Back in the early 1960s a set of Quads and Futterman amps were the match made in heaven and SETs simply didn't make enough power for that speaker.


When I first started out 46 years ago I had a lot of audiophile ideas about how things worked- what made a difference in the sound of a circuit and so forth. Over time a lot of those ideas (many of which came from audiophile friends of mine) died an ugly death. It turns out that if you have a command of the appropriate math you can predict how an amplifier will sound and that won't be a matter of taste since all humans use the same hearing/perceptual rules. Whether people want to hear about it is a different matter of course but you 'canna change the laws of physics!' as Scotty once put it.

Especially when we have solid COLBALT out put trans in the mix,
These powerful trans, are not cheap, I should add.

:)  or you could dispense with the output transformer altogether.


My speakers are 97.5dB, 16 ohms and flat to 20Hz. I've yet to hear an SET on that system that can make proper bass. I've got LPs I produced so I know how they are supposed to sound- I was there when they were recorded.


The thing is, SETs can be better than some PP amps but generalizations like saying they are 'better than all' usually don't work out. At the very least its like 'the fastest gun in the West', taken down eventually by the guy that was actually faster. 

So this is a mono set? I assume you recommend your tube pre, or is that coming soon too?

@bjesien We've been making preamps since 1989- the MP-1 was the first balanced line preamp made for home audio.

Intrigued. Could you please point me in the direction of any credible scientific documentation which arrives at this conclusion?

@pesky_wabbit  This fact that the ear/brain system uses higher ordered harmonics (5th and up) to sense sound pressure is well known and you can prove it to yourself with simple test equipment. All you need is a speaker, an amplifier, a VU meter and a sine/square generator. Set the generator to sine. Run the signal into the amp and then to the speaker. Use the VU meter to show the level. Set the level to 0VU. Then cover up the meter and turn down the volume. Set the generator to squarewave output. Keep the meter covered; run the volume up until its as loud as it was before. Uncover the meter and you'll immediately see what's going on- typically the meter will be showing -20dB or less. Square waves have lots of higher ordered harmonics and sine waves have none.

On page 31 of the Radiotron Designer's Handbook (3rd edition, from the 1930s) we see that it was understood back then that the higher orders were more audible and so should not be created in large amounts for the critical (or casual) listener. Here's a link to a pdf of that tome:
https://worldradiohistory.com/BOOKSHELF-ARH/Handbooks/Radiotron-Designer%27s-Handbook-3rd-Edition.pd...You might want to read the chapters on audio power amplifier design; its quite interesting to see how well were understood the principles we use today. Keep in mind that the Radiotron was meant as a guide and is not greatly in-depth.

General Electric did a study on this in the 1960s which I read in college but I've yet to find that study on line.

When is yours coming out and what’s the price point?

@bjesien We're doing limited production right now. They are $5200/pair.

I've not heard the SPEC amp. But I have heard class D amps that are boring; I know what you're talking about. There's as much variance with class D amps as there is with tube amps.

There might be a revolution in the making…

There is.

Its now possible to make class D amps that are just as smooth as the best tube amps (whatever you have in mind what that might be) and even more detailed (owing to lower distortion).

I mean at 8 oclock, the SPL would be over whleming.

SETs generate up to 10%THD at full power. If you have a lower powered SET on a speaker of insufficient efficiency, the result will be a lot of distortion, which the ear translates as 'loudness'.

But if you use a sound pressure level meter, you see the truth of the situation.

SETs have just as much if not more bass SLAM as any PP amp. 

This is demonstrably false. SETs have a lot of trouble making bass because of limitations in the output transformer. If you really want to get the bass right, to prevent phase shift (since you have no feedback in most SETs) you have to have full power bandwidth to 1/10th of the lowest frequency to be reproduced (20Hz). That means you need bandwidth to 2Hz and SETs simply can't do that.

For that matter **most** PP amps can't either, but they get a lot closer. There are class A triode tube amps that do have full power to 2Hz FWIW.

I learned the lesson of speaker efficiency a long time ago. When I hear systems that have low efficiency speakers and a high power amp, they sound 'little' compared to what I'm used to...

I think a low sensitive speakers fashion is a huge delusion of Hi-End market .

I totally agree!

This drives my 99db Heresy IIIs to as loud as I need to hear 'em with plenty of headroom left.

Its this phenomena of *sounding* loud that's what I'm talking about. My speakers are 1 dB less and my room is not particularly large; I find 30 watts to be a nice minimum power. This means the amp is loafing all the time. When an SET or SEP is making over 20% of full power, the distortion is causing it to sound 'loud'. A sound pressure level meter sorts things out pretty quickly. They are available as an app for smartphones.

@luisma31   Yes.

The loudness cues to which I was referring are the higher ordered harmonics- the 5th and above. The human ear uses these harmonics to sense sound pressure. But it also assigns a tonality to any distortion- the higher orders cause brightness and harshness even in tiny amounts, since the ear has to be keenly sensitive to them in order to gauge sound pressure over a 130dB range.

If you know these facts then that gives you a good idea of how to approach amplifier design. Since the ear is arguably more sensitive to the higher ordered harmonics than **anything** else, it follows that for electronics to sound natural that is the kind of distortion that should be minimized (BTW it is this type of distortion that characterizes most solid state designs). When this is done, you can run much higher power levels and yet the system will not sound loud as all those loudness cues aren't being generated by the stereo.


The result is to naturally turn the volume up to get a realistic sound pressure. IMO, this is at the heart of the shortcomings of SETs, since they don't make much power and these harmonics are audible at any power level over about 20% of full power (which is why so many SETs are described as 'far more dynamic than their low power would suggest- its simply distortion that the ear interprets as loudness, showing up on transients that have the most power). The result is that even at relatively low sound pressure levels they sound loud. IMO/IME it is the mark of a good system that it always sound relaxed and never sounds loud until it really is (+95dB). The only way to get around this limitation with SETs is to use really high efficiency loudspeakers, which from what I've seen over the years doesn't happen that often- so many people are not experiencing the best that SETs have to offer. 

Saki70, the speaker should be designed for tubes. See

http://www.atma-sphere.com/papers/paradigm_paper2.html

for more information.

I spent a few months tinkering with some 45-based SETs that I inherited last year. In analyzing the circuit, I realized some performance gains were possible through some simple mods and so got the amps to sound a lot better. Ultimately though I got curious what would happen if I made a P-P 45-based amp and did that as well (it makes about 4 watts and is class A). It was **instantly** better on the first attempt. IMO/IME most P-P amps are too complex and if a minimalist approach is taken SETs will not compete.

Phaelon, SETs are usually not very powerful. A big one is 15 watts, maybe 20. If you really want to hear what they do, the speaker has to be efficient enough that you never bring the amp anywhere near full power.

At full power, the THD is often about 10%! At low power levels, the distortion might be unmeasurable, and what of it there is will usually be lower orders (2nd, 3rd and 4th). At high power the high orders become involved- that is the range of the amp that tells you not to turn the volume up any higher- it *sounds* loud due to to the presence of the 5th and 7th harmonics, which tell the brain how loud the sound is.

So the efficiency of the speaker is paramount and no matter what anyone tells you, to really hear what SETs do you need high efficiency: 97-98 at a minimum if your amp makes 15-20 watts. A 45-based amp needs a speaker that is more like 107 (if you really want to hear what it does anyway). This power limitation is one reason why I don't use SETs. The bigger you make them, the less bandwidth and detail.

My speakers are 97 db and 16 ohms. I find that 30 wpc is not enough- its much easier to play the system if I have 60 watts. But the amps I use are designed to not generate any loudness cues to the human ear (unlike SETs, which rely on enhancing the 5th and 7th harmonics to get the 'dynamic' sound that SETs are known for), so I use the power without the system sounding loud. My room is 17'x 22' and is moderately lively with lots of diffusion. I do like to listen at higher levels, often reaching 100-105 db, but the system is relaxed at those levels and lacks any hint of being 'loud'.

I put 'dynamic' in quotes here and above because IME, many audiophiles when using that word are often referring to distortion of the 5th and 7th harmonics instead of actual dynamics. The way this works in an SET is the 5th and 7th harmonics don't show up until you really push the amp, which is usually on transients. So the loudness cues exist on the transients, but not in between. This is why SETs are well-known for being so 'dynamic' in a way that is out of proportion to their power. IOW its a psycho-acoustic effect.

This is why so many SET users play their systems at relatively low levels- the loudness cues created by distortion prevent them from wanting to turn it up. But if the system is designed to avoid these harmonics and has the ability to play louder, it will be very natural to do so.

Johnk, the post was worded right. The upper power limits of SETs in a way are more of a function of the output transformer that it is the tube! The bigger you make the transformer, IOW to handle more power, the less bandwidth it will have. If you try to optimize it for bass, the highs will be rolled off and vice versa.

Since many high efficiency speakers (99db+) have limited bass response, SET builders usually sacrifice the bass performance in favor of the highs, since speakers that go below 40 Hz in the high efficiency world are few and far between.

IME experience the bigger SETs also have less detail. This is based on my personal listening experiences, but you will notice that one of the revered 'king' tubes in the SET world is the type 45, which usually makes less than one watt at full power. IOW, the smaller you make them, the better they sound, which is a common theme with amps in general (although OTLs seem to be one of the few exceptions), both SET and P-P.

I've also noticed that it is exceedingly rare to compare like technologies in the SET world. People often compare say a 300b SET against a P-P EL-34 amp (with the results in favor of the SET), but how often is a 300b or other SET compared to a P-P DHT class A amp of the same power rating? IOW to be scientific, this is where one would start if looking for real answers.

The cutoff frequency of the amp does play a role. Generally speaking, you can hear artifact associated with the cutoff (-3db point) at frequencies up to about 10 times the cutoff frequency. So if the amp starts to roll off at 20Hz, you can hear artifacts up to 200Hz.

The same is true of the HF rolloff- except that now its 1/10th the cutoff frequency. So if the amp rolls off at 20KHz, you will hear artifacts down to 2000Hz.

This is why designers try so hard to get bandwidth!

IMO if there is a subsonic rolloff, you hear it as a diminution of bass impact. It can also be measured quite easily- square wave tilt is the measurement. Having played around with LF cutoff frequencies a lot (our amps are full power to either 1Hz or 2Hz), all I can say is its easy to hear once you know what you are looking for provided you have speakers with bandwidth into the low 20s.