|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
192.25.142.225
In Reply to: SET and OTL posted by Paul Folbrecht on December 22, 2006 at 06:51:59:
>Probably a dead horse, but..
Really digging the SET sound for awhile now.>At RMAF, I discovered that OTLs could have that same extreme-purity thing going. Walking into the Atma-Sphere room, I was sure it was SET initially. Those things are sweet.
>Theory is theory, but, theoretically, I've been wondering about the relative advantages. Does an output tranny do more harm than the paralled tubes that are needed for OTL?
>When you're class A, triode, zero feedback, like A-S, certainly very little, if anything, is given up circuit/simplicity-wise to any SET. I am not sure I understand this part fully, but it seems that a fully balanced circuit is NOT push-pull, anyway. No phase split. No extra component on the input.
>And OTLs give that speed that, I believe, no SET can match.
I have a few comments. I used to own a pair of Atma-Sphere M60 monoblocks and have designed many SET and PP amps.
First of all, A-S amps are not fully class A, they are better described as class AB with high current bias. The M-60's perform as a class A amp to about 5W and then switch over to class AB operation. If it were truly biased up to class A those tubes wouldn't last long and you probably couldn't stand the heat.
They are zero NFB amps. They do have a phase splitter, even if you don't see it. If you're using RCA single ended input jacks, then the diff amp input stage operates with one side of the diff amp pair grounded which then operates as a phase splitter. If you're using XLR inputs, then there is no phase splitter going on in this amp because you're feeding it a pre-split differential signal.
The balanced circuit of the diff amp input and the circlotron is indeed a push-pull configuration. It's not the same as a typical push-pull, with dual output power supplies, but it still acts as a push-pull in all regards including the odd-order harmonic distortion cancellation. This distortion cancellation does not occur in a true SE amp.
The fact that there are many parallel triodes in an OTL does seem to be much less harmful to the quality of the sound than in a transformer-coupled amp. Parallel triodes trying to get together to drive a transformer primary seems to be a recipe for smearing and overall mushy sound. Such is not quite the case when the transformer is removed, or at least not nearly so much.
The A-S M60 sounds nothing like a SET to me. It sounds nothing like a transistor amp to me. It has its own sound. It is very agile, very detailed, but does not have the harmonic richness nor depth of soundstaging of a true SET. Well, that's probably because it is a push-pull.
I decided in the end after many mods to my M-60 that I could never make it sound like a SET amp which is the way I like it to sound. So I switched over to SETs again. With my high efficiency speakers, it's all I need.
Follow Ups:
Thanks, Kurt. I was surprised to read that AS are not true class A.
Are you quite sure about that?You may be right that soundstaging is a key difference 'twixt OTL PP and transformer-coupled SET.
"Thanks, Kurt. I was surprised to read that AS are not true class A.
Are you quite sure about that?"
Okay, let's do the math. The individual bias currents for each 6AS7 output triode section is set to 60 mA. There's 2 sections per 6AS7, and 4 6AS7's per phase on each channel. With the circlotron, you have two power supplies out of phase that can contribute to the overall bias, which doubles that again to a total of 16 times 0.06 amps = 0.96 amps. That's 0.96 amps of bias current available as a peak current source for class A operation on each channel. The peak output wattage into 8 ohms is thus (0.96 amps)^2 times 8 ohms = 7.4 W. For RMS class A output power, that's (0.96 amps * 0.707)^2 * 8 ohms = 3.7 Watts, a little under my estimated 5 Watts.Okay, so how much power is being consumed by this output stage at idle? Well, the voltages at the 6AS7 plates are about 180 VDC each. So the total bias power consumption is 180 V times 0.96 A = 173 W per channel. And the power dissipation on each 6AS7 section is 0.06 A * 180 V = 10.8 W, which is about right for that tube.
If it were competely class A to 60W as claimed, the current would have to be biased to 3.88 ADC per channel, with a total power consumption of 698 Watts. For both monoblocks that adds up to 1396 Watts which is not even including the heater power and the input section. Heater power is an amazing 126 watts per monoblock just for the 6AS7's. That means each monoblock would idle at a power dissipation of about 824W, or 1648W for a pair. Now that's a lot of heat for 60 W/ch, and maybe a little too much for the circuit breaker for the outlet.
Also, for true class A operation, each 6AS7 triode section would have to be set to 0.24 A, with a plate dissipation of 43 W. That would quickly burn out the tube since it's max rating is about 12 W.
Scaling up to the MA-3, if there was 200W per channel class A watts, each monoblock would be idling at 2750 Watts. For two channels, that would be 5500 Watts. Even if you had a 20A outlet, that's not enough for the outlet's required draw of 46 Amps. As you can see, it's not going to work that these are true class A amps in the classic sense of the term. However, clever marketing words make this sound like class A when they write "Class A means the current is never completely cut off at the extremes." If that's how you want to play the game, then all amps are class A as there is always some trickle of leakage current at the extremes.
Kurt
Kurt,Your math makes sense (tho I didn't follow all of it - your knowledge here is greater than my own) - I had always read that the simple "rule of thumb" for Class A operation is that the amp draws at least 3x the max power output, at idle. The Atma-Spheres exceed that threshold by a wide margin.
I guess you are saying either that this is not an appropriate rule of thumb or that it doesn't apply to this topology. Bear with me, if you don't mind - are you saying that the output devices will not each conduct 100% of the signal?
"Bear with me, if you don't mind - are you saying that the output devices will not each conduct 100% of the signal?"To be precise, I am saying that not ALL of the output devices are conducting for 100% of the signal to full rated output power of a sine wave. If it did, it would meet the definition of class A. It does meet this definition for a little under 4 watts of output power. Above that and then one half of the output devices are cut off while the other half is further conducting to generate more power. This class of operation is defined as class AB, where there is appreciable amount of power operating as class A, but if it will go beyond that power to its rated power it is then acting mostly like class B (one side is now cut off). A large bias current makes a big overlap in the crossover and it's therefore pretty much a hybrid, AB. Perfect class B operation would have no overlap, where at the 0V output point both sides just barely cut off simultaneously. No real practical amp does that because you really can't make it work well and not cause bad "notch distortion". So push-pulls in the real world are either class A or AB, with varying amount of bias for how deep the class A operation is set. For the M60, that level is a little under 4 watts, again.
Now 4 watts of class A operation is pretty good, because that is more power than the full output power rating of a typical 2A3 SET, which is a true class A amp that clips at about 3.5 watts. The M60 is still a class A amp at the point the 2A3 quits, and then it will go on to 60 watts if need be, but not as a class A amp anymore.
One could also label the M60 as a 3.7 watt class A amp with 60 watts for dynamic headroom.
Hi Kurt, All Atma-Sphere amplifiers are Class A. This is easy to demonstrate, since all you have to do is put an ungrounded scope across one of the current limiting resistors in the output section and you will see.The tubes never go into cutoff. Since they don't cut off (until well past clipping) the amp is considered Class A.
There are some assumptions that throw your otherwise correct math. First, plate B+ is no where near 180V! We don't even run the tubes that high if they are operated in AB with the optional AB switch (which, no surprise- no-one orders). Class A B+ is varies depending on the vintage of the amp, anywhere from 130V to 145V.
Also, the amps operate A2. This is a form of class A wherein grid current flows through part of the waveform. If the tubes were to go into cutoff without clipping, it would be defined as an AB2 amplifier. But- the tubes do not cut off. The driver circuit is designed to accommodate this operation and it turns out that 6AS7s are bit unusual in that they go into the grid current window at about -15V on the grid with respect to the cathodes. We are able to drive the grids about +15V before the tubes saturate, so our useful grid current window is about 30V wide. We chose the 6AS7G for these and a few other properties that make them the ideal power tube for an OTL, for example the total current that the output section can produce is much higher than most people think!
It is the fact that the amps are class A2 that throws people. Most people produce math similar to yours, not thinking that we might have studied the issue at some point and taken a slightly different course!
As the load impedance is reduced, the amplifier will eventually be operating class AB2. For example, the M-60 is class A to full power into 8 ohms, but is not at all into 4. The MA-1 and the bigger amps are class A into 4 ohms. As the load impedance increases, dissipations decrease and efficiency increases.
There are of course other tubes that are more efficient, but they are pentodes and consequently of no interest. Within the constraints our our design, the 6AS7G provides the widest flexibility for power vs being triode vs being able to operate class A.
Call if you have any questions 651-690-2246 and have a Happy New Year!
"Hi Kurt, All Atma-Sphere amplifiers are Class A. This is easy to demonstrate, since all you have to do is put an ungrounded scope across one of the current limiting resistors in the output section and you will see."Well, if I still had my M60's I would do that measurement right away to see how that looks. I do have an idea of how it actually does look, and my guess is that it's really distorted. Now my impression of an ideal class A amp, A1, A2, it doesn't matter, is that the distortion of each contributing phase would have no appreciable distortion at full power, less than say 5% THD, and mainly even ordered of course. If it's well over this, then I think this is moving out of the realm of a true class A amp in the spirit of what a class A amp is supposed to be - that is, an amp whose opposing phase contributions are always about equal and always there up to the full rated power.
I can only believe that the signal you would see on that current limiting resistor would have huge compression toward cutoff and a pointy end where it is conducting heavily in "class A2", or AB2, whatever. How else can a 6AS7 biased to 60 mA act? It can only swing down to near 0 mA and up to the full 484 mA of current it's required to deliver 60W with the help of its paralleled friends into an 8 ohm load. Its zero power crossing point is a quiescent 60 mA. If this doesn't cut off sharply, but rounds out at the bottom just perfectly like a badly distorting remote cutoff triode could do - like a 6AS7 might do, then the 2nd harmonic distortion alone is:
100 * ( 0.5 * (484mA + 0mA) - 60mA) / (484mA - 60mA) = 43%, BEST CASE.
(That is from equation (6), section 2, chapter 13 of the Radiotron Designer's Handbook for a simple calculation of 2nd harmonic distortion of a sinewave.)
That means that when one side is pulling and the other side is pushing, there is a real extra effort on the side that's pulling than on the side that's pushing. In fact you can calculate how much that is: the pull at max power is 484mA - 60mA = 424mA and the push is 60mA, best case scenario. That means at full power there is 88% contribution from the pull and 12% contribution from the push. An ideal class A amp is 50% - 50%. A class AB amp is ideally 100% - 0%. But also, there is some trickle contribution from any class AB amp in the push direction, so it might be more like 99.99% - 0.01%.
So while one might argue that since this 88% - 12% pull-push contribution is "class A" since there is always that 12% still helping out, it seems to me a very "distorted" class A (not to be confused with the final output signal distortion). I will of course let you have your claim that this is class A, but I still think in my mind that it really resembles class AB in the spirit of its definition more than in some technical definition.
To me a properly designed class A amp for 50%-50% push-pull class A operation needs to have a bias at the quarter-power point of current, in this case 242mA. But you can't have this with a 6AS7 and not burn it out, even at 120VDC.
Now that's only my opinion.
Thanks, and have a happy New Year to you all at Atma-Sphere.
Kurt
I still think in my mind that it really resembles class AB in the spirit of its definition more than in some technical definition.I've always been a fan of the technical definition myself :)
If you are the same Kurt that I am guessing you are, you ought to take a listen to what we've been up to in the last 8-9 years. The kit that you built really only hints at what the amps do now days.
I think you might find it interesting also if we had the discussion about Voltage and Power Paradigms. Over the years it seems that you have dealt with a lot of Power Paradigm stuff (in fact we are having this discussion on the SET forum which is clearly Power Paradigm technology), yet you use Voltage Paradigm nomenclature (which is more suited to transistors). If you're interested give me a call 651-690-2246.
"If you are the same Kurt that I am guessing you are, you ought to take a listen to what we've been up to in the last 8-9 years. The kit that you built really only hints at what the amps do now days."Yes, I am probably that same Kurt you knew way back then. I contributed to ASOG and all that when I built the M60 in kit form and modified it ad infinitum and was a rather prolific writer back then. Yes, those are great amps then as I am sure they are better now, especially now that there are teflon caps to replace those polystyrenes. I wish I could afford your products today. But times have changed and I cannot afford any of your products at this time, and I was only able to afford what you had because it was in kit form back then.
And anyway, I am really quite satisfied with what I already have now. I really do believe in the SET sound for me. Even with the OPT's. But, of course, realize that I have direct-coupled passive-equalized amps with cobalt OPTs in a 45 SET package with ultrasonic AC DHT heating. And that just about trumps everything I have heard to date, P-P, SE, OTL, anything. You're not going to make me believe there's much better out there, at any price. Even if it were true, it seems inconsequential to me to try pursuing at this point.
Yes, I think I have hit the end point of this game. If I had a spare $5K to spend, I would improve the drivers of my B-D Design Oris 150's to their top of the line units. Unfortunately, we have a near zero output manufacturing country that causes the Euro to inflate to 1.33 USD from 0.87 USD in less than 5 years. So I no longer look forward to that upgrade.
Kurt
45s don't make a lot of power- I have to imagine that the Oris horns are essential. I've heard very good things about them- we have a few customers running them.I for one am not done with the American economy or production capabilities yet; obviously I think we still have a lot to offer.
It sounds like you have built up a pretty serious custom amp- not too many people use HF to run heaters. Had you thought about building more of them for fun and profit?
"It sounds like you have built up a pretty serious custom amp- not too many people use HF to run heaters. Had you thought about building more of them for fun and profit?"Yes, but then I think about how hard that would be and how much money it would drain me for a very small niche market. I really don't know why manufacturers haven't caught on to the ultrasonic heating for DHT's in SETs because A) it sounds better with a blacker background, and B) nobody wants any hum on even 110 dB sensitive speakers like I have. Also, it only costs about $100 additionally in parts to have this feature. Not bad for high-end SET amps of $5K or more out there. It's just lucky that 45's don't hum all that badly in the first place, but why have any hum?
(You know what's the coolest thing about ultrasonic heating? You use small ferrite toroid cores with 20 or so windings per side and you have an isolation transformer capable of delivering enough power to fully heat a 300B and have it floating properly. Nothing to it! The magic comes from the high frequency for making cheap tiny transformers to do the job. One power supply, one oscillator, one small class AB power amp (a power op amp might do) and two or more small isolation transformers.)
But also, one of the best attributes of my amp is that it has 01A DHT drivers which sound so exceptional there, and there's just no way of putting that tube into production because those delicate thoriated globe DHT's are only found used today. And it's not a good choice for a broad application anyway because the input sensitivity of my amp is now too low for most people. Thankfully, I designed my own phono stage with so much gain it more than makes up for that. But this is an all analog system.
For digital I go to my HT system. That's good enough for digital to me.
Kurt
Its base is a great-sounding 2-channel system that uses ASL SETs and DIY high-sensitivity speakers...
I've added full-range, diffuse-source surround speakers and am still using the Eminent Tech. 12 centerchannel speaker. Preamp is the fabulous-sounding 6-channel, tubed, c-j MET1. Sources are all digital, and they sound plenty good to me.
It's VERY satisfying--BIG, spacious, warm, well-extended and -defined in the bass, and surprisingly so for such inexpensive main drivers, highly transparent--for all the music to which I listen, which is mostly large-scale Classical stuff but also including small-group jazz and even smaller-scaled stuff.
-------------------------------------------------------
Tin-eared audiofool and obsessed landscape fotografer.
http://community.webshots.com/user/jeffreybehr
This post is made possible by the generous support of people like you and our sponsors: