|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
150.148.14.5
In Reply to: RE: Aries Cerat...SE OTL from Cyrus!! posted by morricab on February 12, 2015 at 04:24:43
I was interested to read the following as regards their SE OTL:
"All tube OTL amplifiers use SEPP, cyclotron or Mr. Futterman’s variations. All these are push-pull amplification stages. They can claim single ended, but they are push pull topologies. Feedback is mandatory in this designs to cancel out the artifacts generated by the problems of not true complimentary driving of the signal, and of course cross-over distortion. Some clever variations have been presented on the market during recent years, but they are nevertheless push pull topologies with many dbs of negative feedback, some of them use up to 60db of degenerative feedback. Of course, these amplifiers are not Class A devices."
Contrary to what is written:
(1) Feedback is not "mandatory" in a circlotron
(2) Crossover distortion due to phase splitting is not an issue if the amplifier topology is fully balanced from front to back.
(3) A fully balanced circlotron amplifier can operate in Class A for at least the first several Watts of output power, at least as much Class A power as they get from their SE OTL.
I'm just sayin'.
Rather sick of hype.
Follow Ups:
"A fully balanced circlotron amplifier can operate in Class A for at least the first several Watts of output power, at least as much Class A power as they get from their SE OTL."Well, to be fair, their SE OTL is, they say, 20W output. By contrast, a typical 20W circlotron or totem pole (it makes no difference, for this discussion) would be biased at, say, 200mA, and so the maximum power it could achieve into 8 ohms, while remaining in class A, would only be about 640 mW, if I have my sums right. A circlotron or totem-pole OTL really is operating in class AB for all except a rather miniscule fraction of its maximum achievable output power.
I can see other reasons, though, for not being so keen on their SE OTL. First of all, it must be hugely inefficient, and it weighs a ton (well, a ninth of a ton, anyway!). And also, it presumably has a hefty choke loading the SE output stage. So although it has no output transformer, it is certainly not free of iron-cored inductors in its output stage.
Chris
Edits: 02/12/15
After further reading on their website, I did note that at least some of their SE OTLs are considerably more powerful than is typical of the breed. I'd be interested to hear a pair. As to the Class A operating range of a circlotron, I guess it's all about where you can bias the output tubes in terms of current. I use a highly modified Atma amplifier, once marketed as an "MA240". It used six 6C33Cs per monoblock. You can bias those at 300mA per tube, all day long, so, more Class A power than you calculate I think. I am now using four 7241s per monoblock for about the same current. But the larger Atma-sphere amplifiers, like MA2 and MA3, can handle more bias current for more Class A operation.
My real point was that I object to their exaggerated put down of conventional OTLs. Parenthetically, because this is a subjective opinion, the Futterman amplifiers, even though they do use copious amounts of NFB as compared to a typical transformer-coupled tube amplifier, still manage to sound wonderful, better than one would expect given that fact.
The OTLs I have had (Silvaweld and Transcendent Sound) used a fairly large amount of feedback and I believe that only Atmasphere's large amps are feedback free.
I think Aries Cerat's more powerful OTLs are actually circlotron designs and only one is SE OTL that makes 20 watts.
I believe that is because the Silvaweld and the Transcendent for sure, were/are basically Futterman type amplifiers. They typically do require NFB, but I have never heard any negative consequences therefrom in Futterman type amps that I have owned. (Futterman H3aa, Prodigy, and home-made monster Futterman.)
"I have never heard any negative consequences "
Until you compare them to equally good no feedback circuits I can imagine you don't hear the consequences as it has to be in relation to circuits without it. I have no doubt you find the OTLs of this design to sound good...I do too but not as good as the best SET designs in some important ways. What the OTLs have is this sensation of absolute transparency...something that you find only with the best of the best SETs and even then the quality is slightly different...where I heard my OTLs ultimately lacking was coherence and this only with full range electrostats...not with conventional speakers.
This is why the idea of a SE OTL is so appealing because IMO SET has greater coherence and harmonic integrity and the OTL greater transparency and punch...so marry the two and Violà perfect amp ;-).
I have close to this with my SE(Transistor) hybrid from NAT. It is single ended, it is OTL but it uses a MOSFET on the output (only 1 device per channel though...so kind of like a big output tube), it is hybrid, it is direct coupled and it has no negative feedback. Does it sound like a typical SET? No, it sounds less colored than most. Does it have the otherworldly transparency of an OTL? No, it doesn't quite startle you in the same way...althoug after 2 hours of playing it does get kind of psychadelic sounding, where transparency approaches OTL but with less leanness (feedback related, IMO) and superior tone and coherence.
I recently heard the new integrated amp from Ypsilon, the Phaeton, and it was extremely good sounding...also a hybrid but WOW! It REALLY makes me want to hear the SE(Transistor) hybrid mono blocks...although they actually put an output transformer on theirs (a nearly 30KG one at that). The price is unobtanium but really nice machines
You wrote, "Until you compare them to equally good no feedback circuits I can imagine you don't hear the consequences.."
Right you are. 15 or 20 years ago, I encountered the Atma-sphere amplifiers. Run with no NFB, these clearly were and still are superior to even the best Futterman type amplifiers in my experience. Now, whether my preference for a circlotron output stage with no NFB to a Futterman output stage with gobs of NFB has anything to do with NFB per se, we cannot know. However, I do prefer the circlotron types, in my system, to my ears. (Not wishing to criticize the Futterman topology which still also sounds pretty great.)
I think I asked this before, but maybe not: How can an SE OTL work with no NFB and still be able to drive a real world speaker with reasonably low distortion??? I guess the answer from the company you mentioned at the start of this thread is…. use a very large number of output tubes in parallel. That requires a very robust driver stage plus a gaggle of tubes that are at least fairly well to very well matched.
Well, my NAT did it by using a big MOSFET instead of a tube and that gets low distortion without feedback but to get any power and a reasonable impedance it seems you need a forrest of tubes if you want to go feedback free for a pure tube SE OTL design.
..has its own issues, but there are a few very nice sounding amps with tube input and driver stages and solid state output stages, I admit.
LewIMHO original older J.Futterman OTL designs which uses triodes but not pentodes in asymetrical SEPP(Single Ended Push Pull) OPS can probably have almost that SE amp sound signature , without of any of my doubt that Futterman amps are really PP amps and also regardless of use of significant amount of GNFB .
But must say that only some THD analyzer probe can reveal real output THD signature , so if THD spectrum consist from mainly second harmonic followed with much lower level of third harmonic and even almost non existand traces of higher order harmonics only than I can say that Futterman amps perfectly mimics SE amplifiers sound .Best Regards !
__
"Art which does not have the appearance of art is true art."
- Old Roman saying -
Edits: 02/23/15
The only such amplifiers I can think of were those old models made by Fourier. One used 6AS7s and the other ("Panthere", I think) used 6C33C, I think. I owned the smaller version for about two weeks, or until I could re-sell it. It was nowhere near in the league with a Futterman H3aa or needless to say an Atma-sphere amplifier. Its sonic failings may have had as much to do with its poor quality of construction, as anything else about it. But so far as I know, Julius never built a triode output stage for any of his various models, of which I owned at least 3, at one time or another. As you probably know, he eventually settled on the 6LF6 as an output tube. He did build a headphone amplifier; maybe that one used triodes.
LewYou are basically right ,
but at that old time (1956) Harvard Electronics Co. NY under license was made J.Futterman OTL( model H1 , 15W ) amps based on 12B4A power triode tubes for AB class SEPP OPS , actually every early Futterman OTL amp was based on that 12B4A power triode , I have few original J.Futterman articles from 1954 & 1956 which clearly show that .
But of course that later he switch to use only TV horizontal sweep power pentodes probably with main intention to even more rise up the AB class OTL amplifier efficiency .
Any way my original point was that Futterman OTL PP amp under few condition can easy reveal typical SE amps sound signature ,or even to sound much better than any typical low power SE amp .
__
"Art which does not have the appearance of art is true art."
- Old Roman saying -
Edits: 02/23/15
Thanks for that information. Even as I was typing my post, I was wondering also about the very early models, built in the early to mid-50s. I have seen only one or two in the flesh, but I did not realize he used the 12B4A in those days. Too bad Fourier made a hash of it in the later years.
I have an old HiFi Guide somewhere from the early 1960s that lists a Futterman OTL that used 6AS7Gs for power tubes. It had distortion specs that were not matched with transistors until a few years later.
LewIt is not surprise for me that Fourier do not have big success by replicating original Futterman amplifier concept .
Since even if that early triode based Futterman OTL amp looks very simple on the paper ( schematic ) basically they are not so simple , since surprisingly they are designed almost as some modern SS PP amps , with very high OLG ~60db ,with~ 35 dB of GNFB , with one extra bootstrap loop around of phase splitter and with final target to represent close as possible pure voltage source for at that time standard 16ohm load(speaker).
Even simple wrong layout of passive and active elements on amp chassis can cause unstable operation for that type of amp, sometime inserting of some extra compensation elements is needed to suppress amp oscilation ...
Any way to make long story short ,original Futterman OTL amp concept can not successfully replicate anyone , pretty big knowledge and experience is needed for that , and even bigger for original Futterman OTL amps based on TV power pentodes.
__
"Art which does not have the appearance of art is true art."
- Old Roman saying -
Edits: 02/24/15
" I use a highly modified Atma amplifier, once marketed as an "MA240". It used six 6C33Cs per monoblock. You can bias those at 300mA per tube, all day long, so, more Class A power than you calculate I think. I am now using four 7241s per monoblock for about the same current. But the larger Atma-sphere amplifiers, like MA2 and MA3, can handle more bias current for more Class A operation."Yes, I absolutely agree; if the output tube complement can support a higher bias current, then you can certainly get more class A power output. I was taking the specific example of a typical circlotron or totem-pole OTL with 200mA quisecent bias, as one might have in a 20 or 30 W amplifier (probably like the Atmasphere S-30). Then, I think, it remains in class A only up to the 640 mW or so that I mentioned.
An MA-1, with 14 output 6AS7 tubes, is probably biased at about 560mA, I would guess, and then one could get about 5W in class A, I suppose. But that is again rather humble compared with the 140W it can put out in its normal class AB operation. Working on the same assumption of 40mA bias per triode section, I would estimate about 10W class A for the MA-2 (versus 220W in its full class AB operation); and 45W in class A for the MA-3 (versus 500W in full class AB operation).
I probably wouldn't feel comfortable biasing my 6C33C tubes above about 200mA each, but certainly, if you have six in the output stage then that would allow a comfortable 600mA bias current, which would then be just a bit more than an MA-1, so 5W or a bit more in class A.
All the OTLs I've made, whether totem-pole or circlotron, use quite a lot of negative feedback. I don't personally share the ideological objections that some audiophiles have against negative feedback. I feel that quite a lot of the objections are based on the recycling of outdated anecdotes about poorly-implemented early solid-state systems, and some early studies taken somewhat out of context. But nonetheless, I am actually interested to try a low-feedback OTL, and I'm planning sometime to build something like an Atmasphere M-60.
"My real point was that I object to their exaggerated put down of conventional OTLs..."
I guess most manufacturers tend to give exaggerated put downs of all the other products on the market!!
Chris
Added: Thinking about it, maybe the class A powers I estimated for the big amplifiers could be a bit on the high side. It occurs to me that the voltage swing needed for getting 500W into 8 ohms is pretty big (about 90V peak), and maybe that means the power supplies for the output stage have to be higher voltage than those in the more modest power amplifiers. Correspondingly, the quiescent bias current per triode section would then have to be set lower so that the plate dissipation wasn't too big. In which case, the maximum power achievable in class A mode would be lower than I estimated.
Edits: 02/13/15 02/13/15 02/13/15
Max plate dissipation is, I think, 60W.
If you bias at 300mA and plate voltage is around 130V (in my Atma-sphere output stage), then you're running around 40W, which should not overtax the tube. (It's close to the Golden Mean espoused by you know who.) FWIW, I never had a tube failure in 8-10 years using these parameters.
"Max plate dissipation is, I think, 60W.
If you bias at 300mA and plate voltage is around 130V (in my Atma-sphere output stage), then you're running around 40W, which should not overtax the tube. (It's close to the Golden Mean espoused by you know who.)"Ah yes, the golden mean and "you know who"! Far be it from me to defend anything that YKW says, but there is perhaps a difference in that in true class A, like an SET, the mean power dissipation in the tube increases by a much small factor when the music gets loud, in comparison to the increase in a class AB amplifier. And I would still maintain that practically-speaking, a circlotron or totem-pole OTL is operating in class AB when the music is loud. So one probably ought not to be taxing the tubes too much in their quiescent state, in order to leave plenty of headroom for the loud passages.
However, having said that, real-life experience such as yours with the 6C33C tubes run at 300mA is probably a much more reliable indicator of what one can get away with than mere theorising!
"FWIW, I never had a tube failure in 8-10 years using these parameters."
Ironically enough, just this morning I have been trying to track down a problem in one of my OTLs, which uses 6C33C tubes. One of the output tubes seems to have developed a problem; it goes into a run-away mode after a few minutes and starts conducting more and more. This has happened after quite a few years use, and I was only biasing them at something like 180 mA quiescent. So my low bias current prescription is certainly not a panacea! I did have one other 6C33C failure a couple of years ago, but in that case it was the opposite kind of problem; it stopped conducting completely.
Chris
Edits: 02/14/15
I guess the speaker is a big factor in whether the tube will be overtaxed on transients, when it surely does go into AB mode. (So far as I know, the 6C33C cannot be driven into A2. Why that is so, I don't know.)
You can't push it into A2 very easily due to the immense grid currents involved. If we went to a semiconductor driver we might be able to get a few extra watts.
The 6AS7 goes into grid current while the grid is still about 15 volts negative with respect to the cathode- its onset of grid current is much more gradual by comparison. We can drive the tube to about +15 volts on the grid before the saturation currents become so immense that they saturate the driver (which in turn saturates the voltage amplifier).
I was taking the specific example of a typical circlotron or totem-pole OTL with 200mA quisecent bias, as one might have in a 20 or 30 W amplifier (probably like the Atmasphere S-30). Then, I think, it remains in class A only up to the 640 mW or so that I mentioned.
One thing that is not taken into account is the impact of the load impedance. As with any amplifier, the lower the load impedance, the less class A operation can be had. In the case of the M-60 (with 1 amp of idle current) it remains (barely) in class A2 operation up to full output into an eight ohm non-inductive load (insofar as a 'scope on the cathode resistors is able to show anyway; cutoff does not occur until the amp is into clipping). But change the load to 4 ohms and the amp is class AB2.
The S-30 does considerably more than just 640mW of class A power (its idle current is around 400mA). But it is best running into 16 ohms rather than 8, and this should probably be taken into account.
"In the case of the M-60 (with 1 amp of idle current) it remains (barely) in class A2 operation up to full output into an eight ohm non-inductive load (insofar as a 'scope on the cathode resistors is able to show anyway; cutoff does not occur until the amp is into clipping)."Just for fun, I've been playing around with a Spice simulation of an M-60 this afternoon. Up to a point, I think I see now what you are saying. Namely, that the current in the bank of tubes that is heading towards non-conduction in a given half cycle of the audio signal goes to zero rather more slowly than one might perhaps have expected. ("Slowly" in the sense of requiring a more negative audio signal voltage to drive it into non-conduction than one might have expected.)
The quiescent current per tube bank was 533mA in my sim. I think this is roughly in line with what you were saying (about 66mA per triode section). At about 5W output into 8 ohms, I(max) through one bank of tubes is 1.27A, and I(min) is 0.125A (so not cut off yet). At about 10W output, I find I(max)= 1.65A and I(min)= 0.04A, so maybe one could still say it has not cut off. At about 17W output I get I(max)= 2.10A and I(min)= 1mA. Does one call this cut off or not? At an output of 20W, I get I(max)= 2.33A and I(min) really now shows as zero.
So, in summary, I suppose the question of what power level it transitions from class A to class AB really comes down to a matter of technical definition, and at what point one draws the line. If one were to say that it is class A if even one picoamp of current is flowing through the tube bank that is heading to cutoff, then I suppose one could claim that it *always* operates in class A. But this would certainly not be in the spirit of what class A is supposed to mean, I think!
It seems to me that the whole notion of class A for any kind of push-pull amplifier essentially assumes that the transfer functions for the two halves of the output stage are pretty much linear. If one is depending for calling it class A upon gross nonlinearities that mean the output tube banks are difficult to push into non-conduction, then it's not clear that there is any useful class A quality to the behaviour. It really seems to me that one would then be just claiming class A behaviour on a technicality of a definition pushed beyond its regime of sensible applicability.
Of course, the Spice simulation I did is principally limited by the accuracy of the tube models, but I think it has probably captured the essence of what one would find in an actual version of the amplifier.
Chris
Edits: 02/15/15 02/15/15
What load impedance did you use?
I used an 8 ohm load impedance in my Spice simulation of the M-60.
Just to be absolutely sure I'm using the right magnitude for the bias currents for the output tubes, could you confirm that something like 60 mA or so per triode section is what you are using?
By the way, calculating the output impedance for the M-60 simulation in what I would regard as the "standard" way (putting an AC current source across the output terminals and calculating Z as voltage over current), I get results that vary between about 4 ohms and 9 ohms, depending upon what the output impedance of the signal source is. (This is reasonable, because the degree of feedback provided by the 2M ohm resistors from the output to the input stages depends on whether they just "see" the 100K ohm grid-to-ground resistors at the input, or whether they see lower impedances in the case that there is a low output impedance signal source connected to the input.) Anyway, the 4 ohm figure I get for the output impedance, corresponding to having a fairly high impedance signal source, seems to accord with the figure you quote for the M-60 on your website. I guess that gives some confidence the simulation is reasonably trustworthy.
Chris
60ma is about right. Actually in the M-60 it often tends to be slightly north of that number, but not by much.
"The S-30 does considerably more than just 640mW of class A power (its idle current is around 400mA)."I thought the S-30 has 5 6AS7 tubes per channel, and hence 5 triode sections for each "half" of the output stage of a single channel? Surely the quiescent current for each triode section can't be as high as 80mA? That would push the plate dissipation uncomfortably high, I would think. I was assuming 40mA quiescent for each triode section, and hence 200mA for each of the two banks of tubes.
The maximum power that keeps within class A will be when the current in one bank drops (essentially) to zero at the peak of the waveform, with the current in the other bank therefore reaching twice the quiescent current; i.e. 400mA under my assumption of 40mA quiescent per triode section. A peak current of 400mA means an rms current of about I=280mA, and hence a power into R=8 ohms of P=I^2 Z, hence giving the 640 mW I estimated. Did I underestimate the quiescent current per triode section?
"In the case of the M-60 (with 1 amp of idle current) it remains (barely) in class A2 operation up to full output into an eight ohm non-inductive load (insofar as a 'scope on the cathode resistors is able to show anyway; cutoff does not occur until the amp is into clipping)."
Again, I'm puzzled by your figures. The M-60 monoblock has, I think, a total of eight 6AS7 output tubes, and hence 8 triode sections for each "half" of the output stage? An idle current of 1 amp would mean 125 mA per triode section, which with, say, 140V B+ supplies would mean 17.5W plate dissipation per triode section, which is far too large. (13W maximum is quoted in the datasheet I looked at.) So I am suspecting perhaps the idle current you are quoting is the *sum* of the idle currents in the two halves of the output stage? If we assume that, then my calculation for the M-60 would be that the peak current through the 8 ohm load, while staying within class A, would be 1 amp (i.e. twice the idle current in one half of the output stage). Thus the rms current would be 1/(sqrt2) amp, implying a maximum of 4W into 8 ohms while staying in class A. That compares with the M-60's 45W achievable power into 8 ohms.
I'm not quite sure how to account for the discrepancy here, since you are saying it stays in class A, for an 8 ohm load, right up until the 45W maximum output power is reached. That requires about 3.3 amps peak current into 8 ohms. If the quiescent current per half of the output stage is 0.5 amps, then by the time one bank of tubes has reached 3.3 amps in its conducting part of the audio cycle, the other bank of tubes will have well and truly reached zero. Or are you saying that the "non-conducting" bank never truly turns off, because of a stray microamp or two (or even a milliamp or two) of current, and that therefore technically it could still be called class A because neither bank of tubes ever completely literally stops conducting?
Chris
Edits: 02/13/15 02/14/15
In the case of the S-30, it is class AB2 into 8 ohms. While the amp can sound just fine into that impedance, it really seems to shine into higher impedances. Like any amplifier, it has lower distortion into higher impedances, and like any other amplifier not due to class of operation so much as simple output impedance.
The quiescent current is for both halves, much like you would measure in any push-pull amplifier.
I've not worked with simulation. I just put an ungrounded scope probe across the cathode resistors and then looked at what was going on.
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: