 
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
|
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
75.168.143.116
In Reply to: RE: Transmitter Tube SET. What Are Your Overall Experiences? posted by Mister Pig on September 08, 2021 at 14:32:21
I'm sure the tube itself is fine. But as you increase power in an SET the OPT is progressively harder to get bandwidth, to the point that its a stretch to call it hifi.
You're going to need some power with those speakers- 60 watts at least unless you have a smaller room or never crank things up. So barring the two caveats SETs seem out of the picture unless you replace the speakers.
Keeping in mind of course that if you really want to hear what an SET does don't play it past about 20-25% of full power. Otherwise the distortion will cause it to sound very 'dynamic'.
Follow Ups:
SE amps need High-Power, low DCR power supplies.
Ordinary power supplies that use time-honored
power supplies that were historically designed
for push/pull amps were also used for SE amps,
which is a severe power mismatch for SE operation.
Properly designed SE amps don't have the
described distortions mentioned.
-Dennis-
Properly designed SE amps don't have the
described distortions mentioned.
Single-ended circuits have a quadratic non-linearity. That's a thing unaffected by the OPT. I'd be very interested to see a 'properly designed' SET that behaves as you describe.
I agree with the quadratic non-linearity.All topologies have some kind of problem(s).
Really good push/pull-- really good--- sounds
just like the best SE, but they're very different
animals as you point out.I would point out that the P/P or any balanced topology
is processing the signal (if it had started out as a SE signal--
there are some sound labs that use all SE topology-- ),
applying Common-Mode rejection to it-- the
distortions & noise are lowered.Just how much of what is filtered out was music? Is
perhaps the lack of this filtering why we love SE amps?Yet I have heard all balanced systems that sound exactly
like the best SE systems-- NO audible differences!It comes down to the designer-- what you get in your
listening room. It is very possible that the best
SE amps require certain speakers, and those speakers
may offer some sort of corrections---?.High power SE's have never interested me because of
the output trans limitations and the power required
to drive the large output device.It's an amp driving another amp, so why not just
listen to the stage before the output device?
You get fewer signal quality losses due to excessive parts,
better sound-- so why not just get a speaker that needs less power?As your top-notch work indicates, the two
examples are different anyhow-- the handling of
musical data will be different depending on how many
stages are used. I certainly agree with your work
on P/P amps-- different coupling methods, etc.,
will deliver different handling of waveforms.In the end, one may decide to change or even
compromise one topology in some way in order to
get a better overall result. For instance, one may
choose to drive a P/P output stage with a single triode
that is driving an interstage transformer-- using
the transformer as a phase inverter.You would see a flaw right there, and you would be
correct. Someone else, however, might look at it
differently-- see that it is very simple, has a truly
balanced phase inverter, and has
few signal losses in the circuitry even though
things look better if that amp is just balanced all
the way-- giving a far different handling of the signal.One could point out-- why not trash the transformer
phase inverter and just install two triodes-- that's
better, isn't it? SURE-- for about 10 minutes.One need only wait until the two tubes decide to
differ from each other. Then, the single-triode
driving the transformer/phase inverter will CREAM it.Now, which one is going to be the most fun for the
user? Especially if he keeps it for long time, and
puts many hours on it?Why wear-out tube sockets-- that's another signal
compromising problem for sure....Well, which one is going to sound the best?
That would depend on what the designer did with
the rest of the amp, and what he did to minimize the
effects of the flaw.Kudos to you and your fine work, Ralph!
-Dennis-
Edits: 10/04/21 10/04/21 10/04/21
Just how much of what is filtered out was music?
None, although if poorly executed you might wind up with a poor distortion signature (what most people call the amp's 'sonic signature').
Operating fully differential (and therefore balanced) there is nothing specifically acting as a filter unless you include the fact that all audio circuits are filters in that they have limited bandwidth.
Distortion cancellation won't remove actual musical information, just distortion. Its never perfect of course. But what that means isn't that some music got thrown out; it means that some distortion got left in the signal.
In the case were a transformer is used to convert from single-ended to balanced (which is one of the things they do really well), if you are using a single-ended circuit to drive it you'll still get some of that pesky 5th. So you'll want to drive it with a balanced circuit rather than single-ended.
The way I get around drift in the triode driver is done by using a triode differential cascode with a highly effective constant current source. With the extra gain possible and with the current in the circuit highly regulated, the drift is really minimal and tends to drift to a design center rather than away. That solves a number of problems at once- the only issue then is driving the power tubes. I use direct coupled cathode followers for that. Over a period of weeks and months the drift is surprisingly minimal- no adjustments needed in that time unless a tube failed.
"Distortion cancellation won't remove actual musical information, just distortion. Its never perfect of course. But what that means isn't that some music got thrown out; it means that some distortion got left in the signal. "
I just wanted to add to that for clarity. The distortion cancelled by a differential circuit is just the distortion that would otherwise have been created by the circuit, not distortion present in the signal from the source. On a related topic, the hum/noise picked up by the cable is the only thing "rejected" by a balanced connection between units, not any hum/noise present in the signal on the recording. Either way, none of the music is being filtered out by these methods.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
TRE and Ralph:Thank you BOTH for the excellent discussions.
You've given readers a good grip on how the
circuits you describe are supposed to work.I agree with you both, but I DO hear ANYTHING
that complicates a circuit, adding and subtracting
it's own artifacts, no matter what it is.For instance, CCS. Lots of people use them in many
ways and swear by them-- no argument there.Trouble is, I and others can hear the extra circuit
operating on the music. With the best circuits, and
speakers, it's subtle, but obvious. I really hate CCS circuits
because I can spot one operating anytime I hear it--
in anyone's amp-- two sounds for every one that's really
there. It's very small, and it's subtle. But over months of
use, CCS circuits eventually become musically boring.Another most excellent performing Band-Aid. BUT,
you can HEAR the thing doing it's thing....A super simple amp with a huge power supply can feed
the load without a CCS circuit better, and not color
the musical timing cues.So------- a few of us break a few rules now and then.
And why do we do that, when everybody knows we're WRONG?It's only because that's how we get really good
performance.I've been doing this all my life. When racing Motocross bikes,
I always had more usable power than the other guys in the same class.
They ran more RPM's= more horsepower.BUT-- I had more low-end torque AND decent hi-rev power, but
not as much as theirs.. TRE would call my engine linear and point
out that my amp isn't truly linear, only my Motocross engine is!
(the engine was putting better power to the ground, sooner).Most people today listen to a lot of single-taste music,
and build amps for their tastes, which is fine with me.I don't like that because I want to play anything I want
and have IT sound right-- no matter what it is.So----ooooo---- what is necessary gets done, trying to
keep it all simple, transparent, and honestly dynamic.It is perfect? No, it's not perfect. But it is reliable,
and you never want to replace it with something else you've
heard somewhere else.Overall, not to any particular taste. It's still not
perfect, it's still not absolutely linear, it's still
got some forms of distortions-- like everything else.It does do one thing right: it plays music like it really
sounds. With your kind understandings, it could make some
sense to you. The reviewers and the guys who play with it
agree-- even if you don't, and I am sometimes at a loss
to explain it all in engineering terms because I don't
groove-out on bringing-up online data-- I figure people
can easily already have that..After all, when a tube is operated in the lower 3/4 of
it's linear curve, why would it sound good? It might not--
unless the system is transparent enough to convey what's
taking place-- the sound of unstressed physics playing
through minimal interference circuitry..Anything-- everything-- that you do to force a vacuum
tube to comply with YOUR demands for some coveted thing--
linearity, loading perfection, etc., is audible as
something or somebody tampering with music if the audio
system playing it is sensitive enough so you can tell.Vacuum tubes-- above all-- operate best when as little
as possible is attached to them, and their operating
temperatures don't stress them in any way.-Dennis-
Edits: 10/04/21
I don't use a CCS in single-ended circuits so I can't comment, but I use them in differential amps all the time and I do have this to say:
Most CCS circuits I've seen are detrimental, because they leave performance on the table and often have linearity problems. I solved that issue about 25 years ago by implementing a 2-stage CCS in our circuits. By having 2 stages it could be referenced to ground at one end- so it knew what 'quiet' was supposed to be; at the other end it was referenced to the power supply, so it could suppress noise from the power supply as well. You really do need a 2-stage CCS to get the job done.
In our amps the power supplies were not regulated. So the performance was affected by line voltage. With the CCS installed as described, the performance of the voltage amplifier circuit was unmeasurable with the line voltage variable from 105VAC up to 130VAC. The CCS over that range varied in performance by 17 parts per million. That's pretty sweet- works really well to reduce noise and distortion while increasing gain.
I can't speak to use in single-ended circuits.
GOOD STUFF. Wonder how many DIY guys are NOT paying attention to what you typed here. Thank You, Ralph.
The Mind has No Firewall~ U.S. Army War College.
Thanks, Ralph. I like that it "sees"-- both sides--
that's an excellent idea, and a good implementation of it.
A two-sided CCS. Better-- much more so,
it's a far more stable circuit.
-Dennis-
-
"Keeping in mind of course that if you really want to hear what an SET does don't play it past about 20-25% of full power. Otherwise the distortion will cause it to sound very 'dynamic'."
My 211 SETs can output 30W in Class A2, and distortion is very low all the way out to that maximum capability. I can't speak to the OP's speaker characteristics, but it is certainly possible to design a SET well above 300B power levels with low distortion and the nuance SETs are known for.
Much of the credit for this performance admittedly goes to the OPTs - Magnequest FS-100 with 10K primaries. Don't know if anything comparable is still available.
--------------------------
Buy Chinese. Bury freedom.
~!
The Mind has No Firewall~ U.S. Army War College.
> > > Much of the credit for this performance admittedly goes to the OPTs - Magnequest FS-100 with 10K primaries. Don't know if anything comparable is still available. < < <
We can get custom-wound CC cores, nanocrystalline, really high B, or partial Ni, wound in OCC Cu or Ag that could easily replace those and more. There are still some good winders out there.
Output transformer quality certainly matters... and the bigger they get, the tougher they are to wind for a given performance. I'm not sure where the point of "not worth the compromise" sets in... and it probably differs between folks.
"Confusion of goals and perfection of means seems to characterise our age." Albert Einstein
My 211 SETs can output 30W in Class A2, and distortion is very low all the way out to that maximum capability.
Class A2 (and now class A3) do help get more power at low distortion out of any tube amp. We use class A2 in our amps quite a lot.
I'd be interested in seeing what the distortion signature of your amp is really like. I'm not disputing your claim; I've just not seen an SET that does not make more higher ordered harmonics when you get past about 20-25% of full power- somewhere in there anyway; it varies from amp to amp.
Even though you might not consciously hear them because they are masked by the lower ordered harmonics, the ear does respond to the higher orders because the ear/brain system uses them to sense sound pressure (fundamental tones without overtones being pretty rare in nature...). Since this distortion is initially turning up on transients as you run the volume up (since that is where the power is) it can give the amp a very lively 'dynamic' presentation.
We've all read about this frequently in comments about SETs, particularly in reviews where the reviewer simply didn't have speakers of sufficient efficiency.
Another way of putting this is that your amplifier in particular does not sound as 'dynamic' as other SETs (which IMO is a good thing, since real dynamics come from the recordings, not the electronics). Do I have this right?
"ven though you might not consciously hear them because they are masked by the lower ordered harmonics, the ear does respond to the higher orders because the ear/brain system uses them to sense sound pressure (fundamental tones without overtones being pretty rare in nature...). "
I don't think this is right. Can you provide a reference for this?
My understanding is that you detect the "loudness" only if it is unmasked...masked is masked and you don't sense it. Otherwise, SET amps would probably not sound as good as they do (and they DO sound good when done right).
You are right that high order harmonics not being masked will impact sensation of depth perception and loudness.
BTW, the low order harmonics do no mask all higher order harmonics. It only masks relatively near neighbors. A bit more masking occurs when the ears own distortion rises with SPL level and this expands a bit to even higher orders but if you have significant higher order harmonics at low volume (by this I mean 7th and above), these will not be masked and will show up in the "character" of the amplifier...even at extremely low levels.
"ven though you might not consciously hear them because they are masked by the lower ordered harmonics, the ear does respond to the higher orders because the ear/brain system uses them to sense sound pressure (fundamental tones without overtones being pretty rare in nature...). "
I don't think this is right. Can you provide a reference for this?
Which part?
My understanding is that you detect the "loudness" only if it is unmasked...masked is masked and you don't sense it. Otherwise, SET amps would probably not sound as good as they do (and they DO sound good when done right).
OK- now that part doesn't sound right to me? Are you really saying that if the higher orders are completely masked, that you'd not be able to tell how loud the sound is? I'm not talking about filtering out the higher orders, I'm talking about how an amplifier makes distortion just so we're clear.
You are right that high order harmonics not being masked will impact sensation of depth perception and loudness.
I don't recall making a comment about depth perception with respect to the higher orders. If I did do that, it was a mistake.
BTW, the low order harmonics do no mask all higher order harmonics. It only masks relatively near neighbors. A bit more masking occurs when the ears own distortion rises with SPL level and this expands a bit to even higher orders but if you have significant higher order harmonics at low volume (by this I mean 7th and above), these will not be masked and will show up in the "character" of the amplifier...even at extremely low levels.
If this is correct then the reason why tubes are smoother than solid state goes away. Tube amps make more higher ordered harmonics than solid state amps do. By rights they should be harsher. But they aren't and the big difference being that in a solid state amp the lower orders are suppressed.
If you refer back to Jean Hiraga though I think you can see why he says an exponential decay is important- the lower orders are thus able to mask the succeeding orders. What this comes down to is what is the formula for the exponential curve describing this decay? You and I differ in this regard (I think...). I prefer the cubic since the decay is more rapid.
Sorry Ralph this is kind of a dumb comment on your part. Loudness perception is only in small partially determined by high order harmonics. Masking unnatural ones (as all non_linear artifacts are unnatural) only prevents the unnatural excess of loudness that has the effect of also making things sound closer than they should because depth perception is dependent on the relative HF content. High order distortion impacts HF perception the most and pushes the sound forward in an unnatural way.
Masked is masked, it means you don't hear it. 2nd and other low orders are largely masked by the ear\brain's own self_distortion. The extension of that masking does not go beyond the next couple harmonics, so all the high order stuff is largely unmasked.
Cubic is wrong. It is the pattern as well as the decay...Keith Howard demonstrated this as well. One might speculate why...I personally think it has to do with the gap in the harmonic spectrum breaks up the masking effect... and our ear\brain evolved to hear natural sounds in an exponential decay if harmonics...that is how the ears own distortion pattern looks...not cubic.
I totally concede that I say dumb things. Do it all the time. But in this case, I've no idea to what you're referring- do you have a link? I'm also interested in the Keith Howard thing, but there are a lot of Keith Howards out there.
So far, a cubic non-linearity seems to have a good pattern. Since the harmonics drop off at a faster rate, and since the overall distortion is considerably lower at any given power level than you see in an SET (quadratic), the result is more neutral and smoother. Its easy to hear too. So you have measurement, the subjective experience and the math all on the same page. But I would like to see what Keith Howard was/is up to.
The distortion products are always there regardless of the power...they are just not visible below the noise floor. What you mean is that they rise above the noise floor with > 20-25% power (maybe...again depends on the specific amp). Does this mean they are now audible? Not necessarily. The sensitivity to higher order harmonics is dependent on the SPL level and so just because they rise above the noise floor doesn't make them immediately audible. If the rise is linear with power increase then they could easily remain inaudible as the low orders are also rising and will provide masking. The problem comes if the high order harmonics are rising at a faster rate than the lower order harmonics thus disrupting the pattern that was set at low power.
The problem comes if the high order harmonics are rising at a faster rate than the lower order harmonics thus disrupting the pattern that was set at low power.
Correct! And that is why SETs have such a reputation for being 'dynamic'. Its also why they sound 'loud' like they are making a lot of power when they aren't. People that are into SETs often talk about things you can hear that are not measured, amps that sound good but measure poorly. This is one of those things you can easily hear.
IMO/IME when audiophiles talk about 'dynamics' you can safely replace the word 'dynamics' with the word 'distortion' without changing the meaning of the conversation.
No, this is not why Ralph...look at the experiments from Peter Van Willenswaard. He found tubes do things dynamically that SS does not. He found it was more pronounced in SET.
Sounding loud and sounding dynamic are not the same perceptually. Sounding loud pushes the sound in your face, which is a sign of distortion taking over. Many amps sound dynamically lifeless at all levels but a good SET sounds dynamic at all levels. Sounding dynamic doesn't depend on SPL and the perception happens even at lower volume levels and perception of soundstage is unaltered.
There are many SS and some tube amps (usually those with a fair amount of feedback) that sound dynamically constipated, where one feels the need to turn them up to try to get dynamics but all they get is loud and still lifeless sound.
But if I had to say, those amps that you try to turn up to get them to play ball have too little feedback (20dB isn't near enough), with no way to add the amount needed to get them around that problem.
Sounding loud and sounding dynamic are not the same perceptually.
Yes and no... both come from higher ordered harmonics. Music has a lot of transients- that's where the power usually is needed. In an SET, when these transients occur, if they happen at a power level where the amp's linearity is reduced (usually about 20-25% and above) then the higher orders show up on the transients and nowhere else. That makes them sound dynamic. At higher levels though it becomes 'loud'. This is why so many SET owners seem to think that 85dB is all the 'louder' they need.
Not sure what you're referring to in your first statement. I googled Peter's name but don't know what I'm looking for. Do you have a link?
When you get rid of that issue, you can't even tell that they system is playing that loud. Now the dynamic structure is coming from the recording rather than the electronics. When sitting right next to someone, you may find you have to yell to be overheard. IOW, its lot harder to tell how loud the stereo is playing.
"Sounding loud and sounding dynamic are not the same perceptually.""Yes and no... both come from higher ordered harmonics. Music has a lot of transients- that's where the power usually is needed. In an SET, when these transients occur, if they happen at a power level where the amp's linearity is reduced (usually about 20-25% and above) then the higher orders show up on the transients and nowhere else. That makes them sound dynamic. At higher levels though it becomes 'loud'. This is why so many SET owners seem to think that 85dB is all the 'louder' they need."
The answer is just no. You may have heard the term "micro"dynamics? This is where little, seemingly subtle shifts, in intensity can make or break realism. Things like action of a pedal on a piano or a clack on the rim of a snare. The initial draw of a bow on a string etc. These are not loud in the context of SPL but there is a relatively large shift in level.Dynamics is the shift in level, loudness is the level.
For a car analogy: Dynamics is like acceleration and loudness is like velocity.
Speakers and especially amps that are dynamic show this prowess at ALL listening levels. Because the realism is retained throughout the SPL range. Quiet sounds shifting in level, loud sounds shifting in level...dynamics is about the shift, the speed of the shift and how accurately the shifts final level and speed is captured... not the actual SPL.
High order harmonics from the electronics has nothing to do with this perception. It has to do with accurate tracking of the recording of all these subtleties and tracking the shifts accurately as they would have occurred on the recording...or in real life.
One of the most disturbing effects of negative feedback that I have heard is this loss of natural dyanmics. The sound is so tied down that it no longer breathes like it should. I think that when you are feeding back the output to the input you are also somehow blunting these dynamic cues from the music...you are not only removing distortion (or creating new harmonics).
All of the best amps I have heard from a dynamic expressiveness perspective were zero feedback designs.
I would have thought, given your bigger models at least seem to be feedback free (the S30 at least is not I believe) that you would understand what I mean here and that it has nothing to do with loudness per se but the perception of a shift in level that reminds one of what one hears live. This could mean a very high instantaneous SPL but one that would be gone almost before it is perceived. However, it is prevalent (or absent) at all SPL.
No SS with feedback and no Class D I have owned or heard has the same "acceleration" as a good no feedback amp.
I realize that not everyone hears this properly so they are not bothered by it's absence, but once heard and understood it distinguishes true high end from mere hifi.
Edits: 10/18/21
'I think that when you are feeding back the output to the input you are also somehow ... '
I have to agree ... you DO think that!
it has little to do with reality but you're certainly entitled to your opinion
be well,
OUTSTANDING!!
Thank you, Thank you, and Thank you!
This is the most useful post EVER on this forum...
-Dennis-
One of the most disturbing effects of negative feedback that I have heard is this loss of natural dyanmics. The sound is so tied down that it no longer breathes like it should. I think that when you are feeding back the output to the input you are also somehow blunting these dynamic cues from the music...you are not only removing distortion (or creating new harmonics).
Yes- I've heard that too. That is why the feedback in our smaller amps (S-30 and M-60) is really minimal (2dB) and zero in our larger amps.
But that isn't a function of feedback as it turns out. Its a function of not enough feedback. And its not enough to say its not enough, because you can point to a Futterman OTL which claims to have 60dB (which it does at bass frequencies). At higher frequencies (like so many other amps out there) it has considerably less, owing to a lack of enough Gain Bandwidth Product.
Dynamics should come from the recording. The amp should not mess with that in any way. And they don't; the problem is distortion affects how we perceive the dynamic contrasts. If you don't have enough feedback you can have proper presentation in the bass, but as frequency goes up, things get messed up. There's more higher orders so while the bass might right, the highs are not.
If you have enough gain bandwidth product you can get around this problem. What you're looking for at any rate is a distortion figure that is the same at all frequencies. You can do that with zero feedback if you have enough bandwidth. Or, if you can get enough GBP you can do it that way, but you have to have enough GBP to support 35dB or more at 20KHz.
At that point the 'dynamic compression' you seem to hear with lessor amounts of feedback goes away. The sonic signature of the amp is another matter; it still needs to have the proper distortion signature (lower ordered harmonics as the dominant distortion product) even though that distortion might be quite low.
I've seen many 'objectivists'(?) (people who simply look at the specs) denigrate high end audio amps as 'tone controls'. In a way they are right, the problem they are having is they are not being pragmatic to understand that distortion is never going away. So if its not going away, you have to make sure that its as innocuous as possible. That will allow the amp to sound like music.
I have heard ultra high feedback amps (mola mola for example) that still sound relatively dead dynamically. Sure they play loud and clean but not dynamic.
If you feed a large % of the signal back you are in fact combining fedback signal with "fresh" incoming signal...this must create some kind of phase shift and "noise".
In fact, Norman Crowhurst noted that one of the most insidious aspects of feedback was a signal correlated "noise" floor. Which is another way of saying a myriad of distortion products, that covers the whole spectrum none of which is large enough to stick up as a clear harmonic peak. What it does do though is modulate with the signal level because, as you know, true noise is not correlated with the signal.
This is why you can easily hear "below the noise floor" on a recording with tape hiss, which is uncorrelated with the music signal. Your brain can pick out the correlated signal and reject the uncorrelated signal (a bit like an organic lock-in amplifier).
What happens though with a correlated "noise floor" though is that you can no longer hear below that noise floor because it is correlated with the signal, which has the effect of truncating low level information and I would argue probably has a deleterious effect on the dynamics as well.
An amp with huge amounts of feedback would also have a very strong and correlated response to the noise floor and this has negative impacts on perception...I can't see how adding even more feedback would magically take you out the other side for this particular problem.
I think Ralph has a way to balance two
feedbacks against each other, cleaning up
the mess that is created by using only one NFBK .
(By doing NFB ALSO in the opposite direction, we balance-out
most of the unwanted aspects of a single NFB loop).
I like it better, but I still don't like it much because
it still truncates dynamics-- simply by lessening
amp sensitivity to "microsignal" information.
The 2-stage amp is made possible-- partially-- by
not using NFBK.
If a balanced NFBK system is employed, many desirable
items in music are restored. But if we do this, the NFBK
loops still reduce gain. When we do that, the gain must be
restored-- add another stage of amplification if you please.
Now, what do we have? GAIN restored. Microsignal musical
information further reduced by what?----- the extra stage AND the
NFBK loops.
Not too bad when driving a medium-eff speaker using several
watts from the amp.. Sterile (by comparison)
when driving a large surface area, high-eff model that is wired
with silver, and has a great crossover system..
That speaker will tell you that your amp is OK-- better than
most-- but in comparison it lacks quite a bit of life-- it's
got power, it's got bandwidth, it's got pretty good phase alignment.
But there's one thing-- in spite of all of it's "A" scores, that
earns it a "D" score when playing music. It doesn't sound ALIVE--
like music dancing in an actual real life listening area.
-Dennis-
When we do that, the gain must be
restored-- add another stage of amplification if you please.
The gain of a class D amp is the relationship between the triangle wave and the incoming signal, if the amp uses pulse width modulation. You can get quite a lot of gain this way. In practice, we get enough so that we can drive the module directly with our preamps (which have no worries driving low impedances, even though they are all-tube). Most preamps don't have that kind of output voltage or current, so we boost the input signal by about 6dB using a set of low noise opamps. When opamps run with a gain of only 2 (6dB) the feedback is so high that the distortion is really stupid low so they tend to be very neutral, even at 20KHz. We're using some pretty nice opamps for this service, so the class D amp will be low noise on horns.
We do not run any significant feedback in our tube amps, because you have too many problems with phase margins and insufficient gain bandwidth product to support the feedback needed. So the distortion product tends to be a simpler kind with more lower ordered harmonics and less higher ordered, much like an SET in this regard although usually a couple of orders of magnitude less.
Thanks, Ralph! An excellent discussion,
as usual.
I break rules whenever I think that it
might work-- maybe it won't work.
I have been lazy in that regard because
I always wanted to play with a high-gain
solid-state device, and run it Class A, with zero FBK..
Everyone I ever talked to on this said mostly
the same thing-- it won't work with solid-state
devices-- they're not tubes-- they require all that
FBK.
Someday I think I'll try it anyhow-- I just want
to know for myself--- hah!
Who knows what that will sound like!
-Dennis-
-
I have some idea on this because I had in the past an amp from the company NAT out of Serbia.
The amp was the Symbiosis SE, which was a single ended hybrid that had a tube input, tube driver and an output stage with a single large MOSFET that was strapped to the bottom of a huge heat sink...per channel. There were two other heat sink towers that also had a single MOSFET that were regulating the output of the other MOSFET. Being single ended it was pure Class A where it consumed 800 watts even at idle.
This amp in some ways was the most amazing sounding amp I ever heard...if you could wait 2 hours for it to fully warm up (it was 155lbs. so there was a lot to warm up). It got spookily transparent...like an OTL and yet had much of the tonality of a good SET. It didn't quite have the same inner resolution and holography of a top SET but it wasn't absent either.
Of course there was not Feedback on this amp at all and it sounded that way. The most interesting part was that it actually put out 100 watts into either 8 or 4 ohms and was 2 ohm capable.
So, while a MOSFET doesn't sound quite like a triode it could in many ways sound quite convincing when used like a triode.
The only reason I sold it was the very long warm up to where it sounded almost psychedelically good...before that it was kind of like you knew it could do much better so it was a bit disappointing until it really came on song.
Most interesting, and confirms some of
my suspicions about it.
The question for me is-- is that long warmup
a function of zero FBK, or is it just power supplies,
etc., finally getting warm enough to clean-up.
I suspect that, but don't know it, so I'll just
have to build one. Mine would be much simpler--
maybe I'll get away with something-- it could work.
Thanks! -Dennis-
you know, I see where you were going with your earlier post that I ['tongue in cheek'] 'snarked' on and have to say it does resonate with me morricab
thanks for expanding on it here!
with regards,
-
nt
I have heard ultra high feedback amps (mola mola for example) that still sound relatively dead dynamically. Sure they play loud and clean but not dynamic.
Yes- something is wrong with the distortion signature. Its not enough to simply get the distortion low.
I agree (and quite a lot) with most of what you wrote.
An amp with huge amounts of feedback would also have a very strong and correlated response to the noise floor and this has negative impacts on perception...I can't see how adding even more feedback would magically take you out the other side for this particular problem.
Its not magic... its just when you get to that much feedback (and it has to be that much even at 20KHz) then it gives the amp the ability to clean up the mess that feedback otherwise leaves. It corrects phase and even corrects the harmonic and inharmonic noise floor (which sounds like hiss, but cannot be penetrated by the ear the way regular hiss can) as well as the bifurcated harmonics it otherwise creates. Back when Crowhurst was writing, this kind of feedback wasn't possible- the gain bandwidth product needed simply was decades out of reach.
But even though that's different today, you still have to pay attention to the distortion signature even though the distortion is much lower. Its got to be right or the amp will still get shot down.
Thanks again, Ralph. From you we learn more
about feedback. I try everything I can
think of to avoid it, but you point out more--
something else-- that people should pay
attention to.Negative global FBK is a TIMING issue, above
all else-- as far as music sounding real is
concerned.This is what makes your post important.
Read between the lines of your post and
one can deduce:The amplifier and the speaker should be viewed as a single
part-- should follow the dynamic events occurring with the
music sources and their components,, their interconnects, etc.Some setups will require a bit of NFBK in order to have
the timing cues come out -- shall I say-- unmolested.IF, however, one can invent ways to get these timing
cues ("microdynamics", etc.) to come out of the speaker
intact, WITHOUT any global NFBK being necessary, then
you've just won the sonics lottery.-Dennis-
Edits: 10/18/21 10/18/21
"But as you increase power in an SET the OPT is progressively harder to get bandwidth, to the point that its a stretch to call it hifi."
I've had this suspicion with higher powered SET amps..... It seems like the flea-powered units ( < 5 watts per channel) provide the most "magic"..........
I find this true with push-pull solid state as well.....
As a rule of thumb, yes, as power requirements of output transformers increase, it becomes more difficult to get good performance.
I think really, really good transformers can be wound for amplifiers producing greater than 5 Watts, and the point at which size becomes a problem depends on the winder, the design, and materials used.
I order transformers with oversized cores because considering the design compromises, this provides better lower mids and bass with little to no compromise of other areas of performance. But, these transformers are custom wound and I'm in the realm of being able to accomodate maybe 10 Watts (and the amp is really more like 4 Watt). Headroom has its benefits.
"Confusion of goals and perfection of means seems to characterise our age." Albert Einstein
~!
The Mind has No Firewall~ U.S. Army War College.
-
I've had this suspicion with higher powered SET amps..... It seems like the flea-powered units ( < 5 watts per channel) provide the most "magic"..........
I find this true with push-pull solid state as well.....
This is why the type 45 is the King of SETs and has been for a while. You almost have to try to mess things up to not get bandwidth.
Just for fun I designed a little 5 watt P-P amp using EL95s to see how it might compete against SETs of the same power. At least with the amps I used (which use 2A3s) the EL95 amp was better in every way: more 3D, more detailed, smoother and more relaxed, especially at higher power. But it was better at low power levels too.
I think the reason SET lovers like them so much is they don't really get a chance to compare apples to apples. IMO/IME you can do that one of two ways. Either compare a PP amp using the same tubes as the SET, or compare a PP of the same power as the SET. I've done both so I have a pair of PP type 45 amps that I built up to compare against a type 45 SET. It wasn't even fair- the PP was better in so many ways (def.: 'better', adj to have greater detail, smoother presentation, greater impact, overall more musical and organic sound).
I did all this in order to sort out what kind of amp I wanted for my bedroom system. Since I never push the volume hard SETs were in the running until I tried this comparison. The little 5 watt amp demonstrated that technology had indeed progressed since the time when DHTs were the only game in town.
Do you know of an Amp or Kit similar to what you designed (EL95) that is for Sale? - I have used the SET 45 Amps for over 15 years now.AND ..... Why do you think that this is not widely known, especially with us Low Power Devotees?
Edits: 09/10/21
AND ..... Why do you think that this is not widely known, especially with us Low Power Devotees?
That one is easy! I think you're aware of how much care most SET aficionados put into their designs. But how many really high quality low power PP amps have you ever seen? I've been in the business for over 46 years and personally (so this is anecdotal) I can count them on one hand.
So when people are comparing SETs, probably the lowest power PP amp they could get their hands on used either 6V6s or EL84s. In either case those amps probably were not optimized nearly as much as the SETs were in comparison.
One issue is the way the amps make distortion. SETs tend to have a quadratic non-linearity while PP has a cubic non-linearity. But if you have a PP amp with a single-ended voltage amplifier, you get both. This algebraically tends to enhance the 5th harmonic. That gives such amps a harder edge which I think most SET lovers point to as wrong, and I agree with them.
But if you avoid any single ended circuitry and make the amp fully differential, then you get the cubic non-linearity by itself. IOW the even orders are cancelled, but as you increase the order of the harmonic they fall off at a fast rate than seen with a quadratic non-linearity, since distortion isn't compounded from stage to stage in the amp.
The 3rd is treated by the ear the same as the 2nd (its innocuous) and masks the higher orders so such an amp can be nice and smooth, but because its overall lower distortion, also more detailed. Its easy to hear.
Regarding my little EL95 amp, I put up a post about it on the amps and preamps section of this website months ago, maybe even last year. It uses (FWIW) one 12AT7 and a pair of power tubes in each channel. The 12AT7 is wired differentially so each plate is driving the grid of a power tube, and the EL95 is designed to be very easy to drive. A constant current source is used in the cathode circuit of the 12AT7 to linearize its operation and get the most differential effect- its also tied to a -300V supply. Only one grid is used for input; the other is used to receive feedback, which is mostly used to reduce the gain of the amp since its quite linear with no feedback at all. The advantage of doing it this way is there is no IMD introduced to the amp at the feedback node.
The result is a low distortion amp that expresses a nearly ideal cubic non-linearity with good masking of the higher orders. So it sounds very smooth but its also nicely detailed. I used an off the shelf output transformer which was claimed to have 50KHz response; I got full power 100KHz response so that part turned out quite well.
My point of making this amp was simply to build a low power amp that was properly designed and constructed- something I'd not seen in PP amps of that power level prior; they nearly always have some compromise built in on account of being built to a price.
I don't know of such an amp that's actually for sale, although when I designed this amp it was nearly the same price to make 10 chassis as it was to make one, so I can do it again. I'm working on one now that's done in chrome, just for fun.
Link Below ....
-
As for a powerful true SET the 845 or 211 tube sums it up.
Edits: 09/16/21
I have zero interest in More Power, as the 45 Amp provides more than I could ever need with my speakers (115db Horns)
nt
| FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
||||||||||||||
This post is made possible by the generous support of people like you and our sponsors:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2025, Audio Asylum, all rights reserved ]
