I think one of the German mags measured similarly. If you look at the plate structure, it is basically the same as an Emission Labs 520B tube, just slightly bigger. I was able to do this comparison because I had the C3 and the Horning SATI amp, which uses 520Bs, at the same time I have also owned the Amare Musica Entropy Diamond that used the Emission Labs 1605 triode (its bigger than the 520B) but sadly, this amp didn't really sound very good, despite being all DHT (the input tube was a 20B DHT from Emission labs). That was one of the bigger disappointments I have owned recently.
SET amp owners thread
- Thread starter bonzo75
- Start date
@morricab
I tried once with Duo Omegas the micro SET of 1 W, barely bigger than a match box.
No problem with driving Duos
I will PM you the list of upgraded items in my C3.
I tried once with Duo Omegas the micro SET of 1 W, barely bigger than a match box.
No problem with driving Duos
I will PM you the list of upgraded items in my C3.
Last edited:
Tube amps vary but are often less than that if they employ feedback.
The Harmon Kardon Citation 2 from 65 years ago was 0.5% at full power. Futterman's OTLs from the early 1960s were 0.05%. The Dynaco ST-70 was rated 0.025%.
Most SETs, owing to zero feedback, are 10% THD at full power (clipping). The Jack Elliano's Ultrapath circuit allowed about half that. Sometimes you see lower distortion in the specs, which is often the amp specced at a lower power rating below clipping.
Can you provide specific empirical measurement statistics for specific amplifiers … Or is this merely generalising once again ?
Last edited:
It would be interesting to see the spec across 20 differently better know manufacturer of SET and highlight the distortion levels.
With which output tube? You realize that this has the biggest effect on how much power can be generated with a given distortion level? What would you do with THD data? Do you think you can assess the sound quality based on that number?
If you want to know more about the new MTA (Multiplied Transconductance Amplifier) by Frank Blöhbaum, you have to pay and read this article. Briefly described, you give a triode the distortion values of a modern transistor without destroying the character of the triode. For example, at full power 0.05%thd.
Article Linear Audio Vol. 6 & 8.
www.linearaudio.net
A little teaser boosted 6sn7
P.S
Blöhbaum Gm 70 SE tube amp 54 Watt at high end munich took longer to find the photo ...sorry.
Article Linear Audio Vol. 6 & 8.
Linear Audio Volumes | Linear Audio
P.S
Blöhbaum Gm 70 SE tube amp 54 Watt at high end munich took longer to find the photo ...sorry.
Last edited:
300b in ideal circumstances
2A3 in ideal circumstances
211 in ideal circumstances
type 45 in ideal circumstances
You'll see that the power tube by itself is making about 5% THD at full power. Any driver and Voltage amplifier will of course add more distortion since there is no feedback. Usually such tubes like the 6SN7 don't have a distortion spec in the same way that power tubes do. This does not include contributions from the output transformer.
So any SET amp circuit is going to have more distortion than merely that of the power tubes. 10% is a very common value- this is just at clipping, when the sine waveform is bumping into the amplifier's limits. Stereophile does not like to show distortion above 1% so most zero feedback tube amps are no-where near clipping in their measurements FWIW.
So my comment is a generalization; based on having designed such amps, knowing the actual numbers the designer is up against (such as the tube specs above) and also from having seen how some amps are rated, which is to say I've seen exaggerated power figures and the like. If you look at these tube specs and then look at the power claimed by some manufacturers, along with the distortion (if they even make that claim) its easy to see the numbers don't always add (or multiply) up. So now you have a tool; for example if you see a 5-Watt 2A3 amp boasting 5% at full power, you know by looking at the specs that something isn't right.
In the 1950s, David Hafler was awarded a patent for his Ultra Linear concept wile working for Acrosound. Landford-Smith did some measurements of distortion and showed that UL is actually lower than either pentode or triode operation:
https://pearl-hifi.com/06_Lit_Archive/02_PEARL_Arch/Vol_01/Sec_02/081_Ultra-linear_Amps_w_Refs.pdf
Its possible to operate a pentode in class A, single-ended UL operation and obtain lower distortion numbers with no feedback. As the patent shows, there is a particular tap percentage for optimal operation. To bypass the patent, other amp companies (including Dynaco where Hafler was employed after Acrosound) deliberately set the tap percentage slightly off, which has become tradition (IOW, everyone does that nowadays despite the patent being long expired). If the tap percentage is set correctly you obtain curves that allow for more linear operation. FWIW this gets you similar sound quality and lower distortion for a lot less money as most pentode tubes tend to be cheaper than most triodes.
You can also reduce distortion by things like the Ultrapath. But on looking at SET amp websites, I don't see a lot of mention of that technique. This leads me to believe its not being used that much. So yeah, 10% at full power before clipping is pretty close to most SETs.
Who drives their *carefully* matched SET amplifier into their *carefully* chosen transducer into Full Output ? You are incessantly referencing worst case operating scenarios !
If I read this correctly it seems to contradict itself.
To answer though, apparently no-one; its worse than that. From what I've seen most transducers (speakers) are variable impedance (and phase angle) loads and the tubes in the amps these days are lessor reflections of the tubes made in the classic era, except for the boutique tubes. This puts the amp at a disadvantage in either case, as both conditions exist in almost any system unless someone has been really careful about tubes. Not much they can do about speaker load variance though!
When you see people talking about how 'dynamic' their system is and how the little power they have is all they need, that's an indication they may well be clipping the amp (SETs have arguably the most graceful clipping characteristic of any amp made) a lot more than they think, simply because the speaker they are using lacks the efficiency. When the amp is at or near clipping, the SET will sound 'loud' due to the presence of higher ordered harmonics rather than actual sound pressure. A sound level pressure meter shows the truth of this (available as an app for almost any phone).
One mark of the best systems is they don't sound 'loud' even when they are; relaxed at any volume.
Who says the speakers that are being used with SETs "lack efficiency"? Mine certainly do not. Easy 105 dB/16 Ohm load played in a small room at 75-85 dB volume. My system does not sound loud even when it is. It is a calm, relaxed sound and I choose the volume that is best for a particular recording and music when listening.
All you do is criticize SETs, on a SET amp owner's thread. List all the examples of these systems that are clipping that you keep describing. What amps, what speakers, what volume? When? How much distortion? Get specific, otherwise your generalizations mean very little to the end user on this thread.
He's giving it to you straight. With no, feedback an SET's waveform just gets more and more rounded with increasing power until a tube along the way is either in saturation or cutoff, so it's next to impossible to point to the amp actually "clipping". All we can really do is to point to an expected power output, such as 7 watts for a 300B, and note the distortion at that point.
As a point of reference, below are two measurements which duplicate those Stereophile usually does, on my recently built SET. The distortion readout is at the top of the graphs. This SET uses 5dB of negative feedback.
0 to 1kHz THD at 1w into 8 ohms, unweighted:
Full bandwidth THD at 1kHz at 1w into 8 ohms, A weighted:
I have a system which uses active crossovers; the low frequency horn has a sensitivity of 101dB/w and the high frequency horn has a sensitivity of 107dB/w. The speakers do not utilize passive crossovers of any type - the drivers are directly connected to their respective power amplifier. I use a 2A3 SET for the high frequency horn - a very good match of transducer and amplifier. The low frequency horn, despite having a very high sensitivity, routinely peaks at 30 watts or so from a McIntosh MC-240 when playing full range orchestral (or modern electronic) music at healthy levels due to the high levels of deep bass, either from bass drums or synthesizers. A 300B SET in this situation would have no chance whatsoever to avoid high levels of distortion - it just can't be done.
That isn't what I do. I've mentioned a number of ways SETs could be improved if you have DIY expertise to execute some of them (like the Ultrapath, which is a pretty easy thing to set up). Also tips for getting the most out of the amp (such as keeping speaker cables short and connections tight- that can have a big effect on the bass as well as clarity in the mids and highs).
We've already talked about the fact that you in particular use higher efficiency speakers and so avoid the problem of clipping or coming anywhere near close. As I said, 'relaxed at all volumes'. I feel like your picking a fight when I've already acknowledged what you're doing right.
I find it odd that on a thread where measurements are considered nearly meaningless (ignoring for the moment that engineering brought single-ended amps to life...) that you want specific numbers. I already pointed out in various places on this thread what happens at clipping or simply asking too much power: Again, the system takes on a sense of 'dynamics' beyond that of the recording or other amps running at the same power on the same speaker. That is a very specific description that does not require measurement. If you read about SETs (particularly in reviews) this is a very common comment and it means that the speakers used with the amps lacked efficiency for the amp to really show off its stuff.
I get why some people on this thread are being testy with me. I literally stepped into the lion's den. I've been studying SETs longer than most people on this thread (since about 1991 or so, if anyone is keeping track). One thing I've seen over and over is how people make poor comparisons and then use them as gospel. For example when comparing to PP, its usually to a much more powerful amp than one of the same power despite the well-known trope that lower powered amps usually sound better. So I built SETs and PP amps of the same power. Another thing I noticed is that PP amps often use different power tubes in comparisons to SETs. So I built SETs and PP amps using the same tube complement. I was able to get the PP amps to sound better in every way when they were of the same power or used the same tubes. A class A zero feedback type 45 based PP amp can do music quite well.
IOW for about 30 years I've been trying to figure out the attraction and if it can be duplicated or surpassed.
Pop quiz: what happens when the output transformer lacks inductance? If you were paying attention you know the answer. So is that an attack or merely the truth?
101 dB and 30 watts is like 115db and probably well over 110dB at the listening position!! That is an ear splitting level in a home. I know no one who listens this loud at home…even my friend who is a concert organizer and likes it very loud at home.
The issue with OTLs is usually output impedance. If you apply enough feedback you can get it pretty low, but even so they don't like making power into lower impedances. There are only a few power tubes that suit. If triode, the 6AS7G (the Russian equivalent, which we use is the 6H13C) or the 6C33. Neither tube is in production.
The other power tube to use is the EL509 or equivalent. These tubes are being produced. But to use them you'd need to monitor the bias of each power tube independently and they really should be matched (unlike the 6AS7G). So you have some real pains to overcome to make that tube easy to use for the consumer.
We used feedback for a while in our early stuff. I think that needs to be revisited since our techniques for using it have improved. So for most of the time our OTLs have been zero feedback. While that's got a nice fuzzy round feel to it, the truth of the matter is the speaker you use with it is every bit as critical for best performance as with an SET, only instead of raw efficiency you need to avoid anything that is 4 Ohms unless its a really big OTL. Our OTLs, because of their high output impedance, tend to work well on speakers that also work with SETs.
I think our driver worked out pretty well but actually most of the distortion in our OTLs comes from the Voltage amplifier, not the output section. To minimize setup issues we developed a circuit that is pretty well plug and play, but getting the Voltage swing we need out of it is a pretty big ask. Other OTLs use more stages of gain (and so have more coupling caps) and employ feedback, which often makes them unstable as the feedback loop isn't very well designed.
So for OTLs not our own, likely the single biggest thing to improve it is to make sure the feedback loop operates properly and does not leave the amp open to oscillation at some ungodly high inaudible frequency. It would be really nice to find a way to use less tubes too. The ideal candidate for that is the 7241 (an enormous power triode) but they are quite rare and expensive, and the later Cetron production tends to have a high failure rate. But a pair of them in class A2 can make 50 Watts into 8 Ohms. People tend to not like a lot of tubes in their amps and there's really nothing for it.
So yes, when we sorted out how to make a class D sound as good it was an easy choice. Took 5 years to sort out the math and what techniques weren't going to cut it.
All this is engineering. All amps have some kind of weakness. The trick is to work around the weak bits and take advantage of the strengths. Fortunately we're not stuck with technology being stagnant- it continues to march along and new solutions appear over time.
Point well taken, but I was specifically talking about bass frequencies, which if you look at the Flecher-Munsion curves, reveal that the ear is far less sensitive at low frequencies than at higher frequencies. A bass drum or synthesizer note at >100dB is far, far less intrusive to the ear as that same SPL at 1kHz.
And it is those bass frequencies which eat up the watts, whether your ears perceive it as 'loudness' or not.
Point being, those very low frequencies are far more demanding of power for a given perception of level by the ear, and consequently can eat up relatively large amounts of power, even with very sensitive speaker, such as my 101dB/w low frequency horns.
Obviously this depends on the music you listen to - if you listen to string quartets exclusively, none of this matters, but if you listen to digitally recorded orchestral music with significant low frequency content or electronic music with no limitation on low frequencies, then it matters a lot.
Either way, if you are using a 7 watt SET on a speaker with a passive crossover, you ARE going to encounter high amounts of distortion on demanding bass passages. Woofers simply are harder to drive, efficient or not, horn or not. Look at my profile picture of my speakers - they are efficient as hell, but still a 40 watt amplifier on the low frequency horns is mandatory.
Last edited:
I agree when you listen that loud then you need that kind of power. I never listen that loud.
If you read carefully, the low frequency range is where there is a lot of discrepancy between various curves, of which Fletcher Munson is the oldest. So, it is likely not as dramatic as these curves suggest, otherwise somewhat flat speakers would sound horribly thin. The B&K finding that a 1dB/octave downward slope gives the most natural sound, means that your bass will be 3-4 dB higher if done right but not 15 dB like F-M curves show.
Similar threads
