|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
1.64.113.188
Why are OTL amps not so popular? What are their major strength and weakness. Any comments are welcome.
Follow Ups:
I found and I hope addressed a number of misconceptions and outright myths that are common to this subject. Its a rare opportunity, I felt like a kid in a candy store :)
I run the most successful OTL company in history, but I can also tell you that I am no marketeer! That is easily the single biggest problem: how to get the message out. Driving speakers, heat, quantity of tubes and output impedance are really not the issue.
You can make OTLs run so cool you can put your hand to the envelope and not get burned. That is a matter of the class of operation! You can easily run an OTL class AB2 or class B, it will run cool and get very long tube life in such conditions. We get long tube life even though we run class A operation (specifically, class A2).
The heat comes more from the class of operation than anything else. If you run one of our amps in Standby all day at the end of the day the heat is negligible- you can put your hand on the tube and hold it there.
Incidentally, OTLs do not suffer crossover distortion problems like other amplifier designs. So operating them with reduced bias is quite practical.
Depending on the tube choice, the number of tubes can be an issue. I like the 6AS7G as it is cheap, linear and readily available in large production quantities. But its only good for about 10 watts so you need a few of them. Our 60-watt amp has 8 per channel and our 220 watt amp has 20. Once you get into that range of power, the number of tubes is not an issue :)
Regardless of the power tube you can get the output impedance down to a fraction of an ohm with 20db of feedback or less. I prefer zero feedback- the output impedance is higher, but I have found that if the speaker needs a lower output impedance (needs the feedback) the chances that it will ever sound like real music is nil, since feedback violates one of the fundamental rules of human hearing/perception: how we detect the sound pressure of a sound.
You might want to take a look at the link for more about this- its not just an OTL thing!
The main thing that I have ever found is that people don't realize that an OTL is something that will work for them! That is the single biggest problem, followed by the fact that there were some OTL manufacturers that made unreliable product, and convinced many (including some on this thread) that OTLs are inherently unreliable. They are quite reliable. It has everything to do with design and execution, like any technology.
"We get long tube life even though we run class A operation (specifically, class A2)."
Here's a type 10 tube biased at 26ma 340 volts plate. That's 8.84 watts plate dissipation. That's within the rated plate dissipation of 12 watts.
If we extend the load line out to the +20 grid volt curve we can see that it's Class A2, running the grid to +20 volts before plate saturation.
We see that we can swing the grid from -20 to +20 and then from -20 to -60 along the load line and the spacing is reasonably linear with the plate voltage moving from 340 down to about 110 and then from 340 up to about 550. A change of 230 downward with a change of 210 upward.
The dynamic curve for this operating point would be reasonably straight and the bias point would be centered in the straightest portion of that curve.
The plate current would be able to decrease (from the idle point) about the same amount as it's able to increase (from the idle point). From 26ma down to 6ma for a change of 20ma. and from 26ma up to 47ma for a change of 21ma.
Here are the plate curves for the 6as7. Will you please place the load line and idle bias point for the 8 parallel triodes of one bank in your M-60 amplifier assuming an 8 ohms speaker load and then do the same current and voltage change analysis as I have done for the 10?
Thank you.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
This is not something that is easy for me to discuss, but you've been on my case about this a number of times over the years and the fact of the matter is there is an aspect about our circuit that is not obvious from the published diagrams.
Its not obvious because we've been careful to make sure none of the diagrams are complete, even in the kit manuals- they don't show everything about this amp. So here is the part that I have not talked about for a long time (and I feel like I am revealing **way** too much but you won't go away without a correct answer): this amplifier employs two very simple sliding bias schemes. The problem is that there are also 2 different patents in this area and we working around both of them to achieve our goal without infringement or disclosure. It is the latter two aspects that have been the gravest concern.
So the correct answer to your question is that there is no one point that satisfies. There are two curves that describe a range- the amplifier increases the bias as the power is increased, in a way not apparent from the diagrams.
Its been a trade secret for some time and if you look back in my posts and our prior (and somewhat terse) conversations on the phone you may be able to find clues that I have inadvertently dropped. I am not willing to file a patent as the information has to be disclosed and I really don't want to do that... there is a reason why so many OTL manufacturers have failed and we are still around.
It is this sliding bias scheme that is why the tubes do not cut off when the amp is at full power, despite your simulations indicating they should. In your simulations you have one set of conditions at idle, but to simulate full power you would need at least a 2nd set of conditions. That would approximate what is going on.
All of the amps are set up to do this so at idle they won't run as hot, but will still make full power without excessive distortion. But its more than that; to get the amplifier to have instantaneous overload recovery some of the design aspects are essential. They have the added plus of decreasing IM distortion and increasing reliability, which IMO is why we are still here after 36 years.
"this amplifier employs two very simple sliding bias schemes."
I understand.
See the linked post below.
With that understood, linearity because of sliding bias schemes still does not make an amplifier Class A in anything other than the simplest of definitions (current never cuts off).
Ralph I have always thought your amps sounded great and are of a very good design and very well made.
You should be congratulated for your achievements.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
in simplest definitions, yes, of course there is grid current during part of the waveform cycle with our amps, thus the Class A2 moniker. Seems to me I first heard about A2 with one of the Fisher amps from the '50s, the A-50 if I got it right.
And now there is also an A3 designation which, if I understand correctly, is the subject of an patent application. Not sorted out yet if it can apply to an OTL, nor am I at liberty to discuss the conditions of the A3 operation.
Krell has had a sliding bias system for some time. The patent they employ is one of the patents we have been careful to avoid. Their amplifier is also defined as class A. Since transistors operate with base current all the time, I think their amp is just simply termed 'class A'.
I really don't need to go through this again but Class A is much more than just whether or not the tubes cutoff.
Class A is about loading, load lines, linearity, bias points WRT the dynamic curve, the dynamic curve being the straightest possible for the chosen tube and natural symmetry of each tube's output waveform, etc.....not just cutoff.
There is always some leakage current so if cutoff is the only criteria for Class A then all amplifiers are Class A.
So cutoff is really just a red herring in the discussion.
Let's just leave it at that. You look at it your way and I look at it my way.
A lot of us listen to amplifiers that are biased in the middle of a very straight dynamic curve and have symmetry without the use of any feedback at all. They adhere to the classic, full meaning of the term "Class A" (1 or 2).
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
We are both on the same page with that. I don't care for the rancor either.
If you have a fixed point then that is all there is to it.
When the bias is allowed to move in response the music it does allow for additional flexibility. The question is whether or not doing such a thing will mess up the sound. But what we found was that without it, the amplifier was much more restricted; that class A (a goal of the design) was going to be *very* inefficient and didn't sound nearly as good unless all the same design aspects of the sliding bias point were incorporated anyway.
So it appears that in an OTL there are more important fish to fry than the specific class of operation. The most important thing I have found is to be able to effectively control the power tubes. You might think that that is elementary but its a fact that most manufacturers skimp in this area, which places a tension between bass response, overload recovery and bias stability. Essentially, its a reliability issue and an overall stability issue. When we came along we were essentially making the first OTLs that were unconditionally stable- something that had not been seen before. The reliability and stability is why we are still in business where others have failed. To that end, its not like its a huge business either- its a fact that OTLs are so far seen as a niche market, and most of the reasons are misconceptions seen elsewhere on this thread.
We have been looking at a much more active form of sliding bias for nearly 20 years. But IMO to implement it correctly means to use a tube that is not in production, even though there are 10s of thousands of them out there (some tube vendors have offered to send them to me for free- they don't have a lot of use right now) but our 20-year rule prohibits the use of tubes that are not in current production. So we have gone with the passive system instead.
.
Have Fun and Enjoy the Music
"Still Working the Problem"
Good article, a cogent defense of zero feedback, with which I agree.
Like some of the others said, OTL amps tend to run hot, because you need to parallel output tubes to get the impedance down to a manageable level without using NFB, or even with NFB in the case of the Futterman-type amplifiers. At the same time, paralleling tubes gets you more power. The few companies that still make OTL amplifiers as commercial products, like Transcendent and Atma-sphere (and perhaps a few companies in Europe), do an excellent job. The point is that, much like SET amplifiers, OTL amplifiers have their limitations. However, mate an OTL with a fairly efficient, high impedance electromagnetic or electro-static speaker, and you have a match made in heaven, IMO. It is odd for me to read the OTL-critical posts here by guys who espouse SETs, with their flea power, high output Z, and very high measured THD, yet cannot abide the somewhat less limiting properties of an OTL. No single technology in audio is perfect.
One specific point: Not all OTL amplifiers are capacitor-coupled. Circlotron type amps, like those made by Atma-sphere, are direct-coupled.
You must not be following Audio Asylum for the last few years. Dennis's Serious Stereo amp is perfect - just ask him (and Jeff).
The first Watt, of course.
I think mainly because of their relatively low output power and because there are basically only two tube types still available for OTL amps.
Other known drawbacks of OTL amps are waste heat and a somewhat higher residual hum.
So far I have designed and built several OTL and many transformer coupled amps (PP).
I tend to use OTL for critical listening at mid/low level and the transformer coupled amps for general purpose listening at mid/high level.
Both OTL and transformer coupled amps sounds good to me (even though it seems that sometimes OTL sounds a bit more 'accurate').
Best Regards
Luca
ecc230
We can make them so quiet you have to put your ear to the horn to even hear them at all. IOW the noise floor is as good as any tube amp made.
IMHO it takes more effort (i.e. money) to make the OTL hum as low as in best conventional PP designs.
Several DIY people have privately written to me that they have not succeeded in making their DIY OTL as quiet as they wanted it to be, mainly because they had not planned in advance to build a quiet enough (i.e. more expensive) power supply.
When delivering several amperes of instantaneous current into the speaker load, the HV power supply of OTL amps is much more stressed than in conventional transformer coupled PP.
Hence you have to conservatively dimension the PS transformers in the 'above 500 VA' range even though the average audio power is only a few watts.
If I wanted to use my own OTL (non circlotron design) with very sensitive horns, I know that I would have to either modify the power supply or to reduce the sensitivity by increasing GFB above 8-10 dB.
We all know that you succeeded in making ultra quiet commercial OTL amps, but it is very very hard for me to say that they are affordable either.
DIY is by its very nature an affordable method to get the tube (and OTL) advantages.
I think that it would not be wise to say that OTL hum can be as low as in best conventional designs without adding that, in order to get that result, a more expensive power supply is necessary!
ecc230
You just need to understand where the hum is coming from, and its a simple fact that many people working with DIY OTLs have not sorted that out!
Here are some fundamentals:
- You don't need DC filaments but you do need to watch out for electrostatic noise, but that is true of any amplifier
- Its OK to have a sawtooth waveform in the power supply, so long as it is canceled out in some way. I've built amps with no filter caps in the output section that hardly hum at all!
- Grounding the amplifier to the AC line *correctly* is important
- the most important thing is snub your rectifiers!!
For sake of clarity this is the basic output totem pole topology used by most non circlotron OTL tube amps (I used a similar one too).
I do not want to bother too much the DIY inmates with these technical details and so this is the basic explanation given by Mr. Broskie (of tube cad journal) of the reason why such a topology is more prone to hum than conventional PP (I would say simply because of its output asimmetry).
I subscribe this simple explanation of Mr. Broskie just because I had found it very applicable to my OTL design and so, in order to reduce the residual 50/100 Hz hum, I knew in advance that I needed a lot PS capacitance (4000-5000 microF or more).
This large capacity requires either the use of some device to limit the current inrush at power on or some retarded relais to exclude some series resistor after full charge of the PS filter capacitors.
IMHO, mandatory use of:
- larger filter capacitors,
- PS power transformers in excess of 500 VA,
- current limiters and/or the like to limit current inrush
will likely result in a more expensive power supply w.r.t conventional PP.
In the case of circlotron OTL I suppose that, as you must use two independent HV power supplies, the inherent better hum rejection of circlotron, which would lead to a more economic PS, is likely offset by the need of doubling the number of HV PS.
This is why I have built only a few OTL amps and many PP amps.
Best Regards
Luca
ecc230
-seems to me that is one reason many of them use feedback as that will reduce hum.
I've never been a fan of totem poles as they are often higher distortion and need the feedback just to work (though it seems to me there was someone on the OTL forum who had a zero feedback totem pole amp that he built as a guitar amp). I prefer instead designs that don't need feedback in order to work right, so the feedback is optional.
So the bottom line seems to be: '*Some* OTLs might be more hum-prone.'
That's what I said.
Turning back to the topic of the original post I think that we can agree that OTL amps are not so popular because:
1) they exhibit a larger parts count (i.e. they have a more expensive bill of materials) w.r.t. conventional PP; this limits their interest for DIY people too.
2) in order to get good results (in terms of not only absolute audio performance but also in terms of reliability and durability) there are only a few proven designs, i.e. circlotron/Atma Sphere (non totem pole), Futterman/Trascendent (totem pole).
3) the only nowadays readily available tubes for OTL designs are 6AS7/6080 and 6C33; all other OTL tubes of the past are very difficult to find and/or too expensive.
Best Regards
Luca
ecc230
I think some are using the EL519 too. Neither are particularly expensive.
IME, OTLs are cheaper to build on a DIY basis than a transformer-coupled amp of the same power, once you make the OTL big enough (60 watts or so).
A lot depends on how you build it, where you get your parts, etc.
But I don't think that is what makes them 'unpopular'. IME it has more to do with misconceptions in the field, for example one Futterman variant was so unreliable that it convinced most of the marketplace that if an amp was an OTL it would be unstable and would blow up (I have come to call that the 'Futterman Legacy' as it has been the single biggest marketing problem that we have faced in the last 30 years, and its something that all other OTL manufacturers have had to deal with as well). The 'Futterman Legacy' was so successful at this that in the marketplace there was no debate about it at all, while at the same time the tubes/transistor and analog/digital debates raged on!
So IOW I think it is misconceptions and marketing that are at the roots of the problem.
Perhaps it would be more fair to call it the "Counterpoint SA4 Legacy" or the "KSS Legacy". Not to mention the Sans Pareil products. Futterman's own amps and the conventional derivatives thereof (NYAL, Prodigy, etc) were quite reliable IME.
Still, your basic point is well taken.
I was already dealing with this issue prior to KSS, although they certainly did not help matters...
I always have thought that Prodigy did the best job of any Futterman I've seen.
Never had a single problem with two pairs of monoblocks built for me by Julius, plus one pair of knock-offs built in DC and sanctioned by Julius, over a period of at least 20 years. Then I had Prodigys for a few years. No problems there, either.
Counterpoint SA4, KSS, and Fourier are primarily responsible for the bad rap, IMO.
My OTLs are basically Atma-sphere amplifiers with a lot of tweaks. I put them together as kits back in the late 90s, with the help of a knowledgable friend. I then decided to educate myself about tube electronics so I could understand this and some other circuits better than zero understanding. As I recall, the first iteration of the amplifier did have a very minor issue with "noise", but it was never true "hum", in the sense of 60Hz or 120Hz noise. Later iterations have consistently been dead silent, as in I don't know sometimes whether the amps are turned on or not, if there is no program material. You may fairly argue that I am using relatively inefficient ESLs, but lately I have been able to enhance greatly the efficiency of the ESLs. Still dead silent. Over 35 years, I have also owned several Futterman-type amplifiers; I never heard any hum from any of those, either.
They typically have to use high perveance, low mu tubes as the output and many of them. I played around with a Futterman HT3 and it used six 6LF6 and two 6LU8 as screen regulators per amp and a funky input hexode tube that had very short life expectancy. These tubes are harder to match. The amp also has to have an array of large capacitance (the Futterman used Rubycons about 18 of them) coupling the output to protect the speaker from DC offsets and to insure good low frequency response. It also imployed considerably negative feedback (about 60db). They also don't sound any better to my ears than does a good transformer coupled amplifier. In fact many conventional amps sounded better to my ears. Ray
"The gift of imagination is a gift of the Gods imparted to a few who receive innumerable kicks in the a$$ their entire life." Le Corbusier (Charles-Édouard Jeanneret)
I have heard these several times and liked them quite a bit. However, they are large, hot, low power and have high output impedances which means that they require a flat impedance speaker and high impedance speaker to sound good and that type of speaker is rare.
I agree that OTLs run hot, and are rather inefficient. But I don't think think it is really true to say they are particularly low in power (i.e. audio power) in comparison to tube amplifiers with transformers. Nor is it reallly the case that they have high output impedance compared to tube amplifiers with transformers; quite the opposite, in fact.
For example, the Tim Mellow OTL amp, with just two 6C33C output tubes, is 25 watts rms and an output impedance of about 0.25 ohms. The Alan Kimmel circlotron, with eight EL509 output tubes, is about 210 watts rms, with an output impedance of about 0.5 ohms. Even the "lite" version, with only four output tubes, claims to be about 100 watts rms output, with an impedance of 0.8 ohms.
They can be pretty simple to build too, as for example the Tim Mellow design.
Chris
I have not heard those. The OTL that I have heard used just seemed to have a lot of trouble with real life speakers due to impedance problems. However, they were 6AS7 amps and the 6AS7 has a plate impedance of 200+
keep in mind that each power tube is only good for about 10 watts. We run them conservatively.
I like the sound of your amps. I have heard OTL with some low efficiency and low impedance speakers before and they sounded not so great in the bass although still nice on the ear.
Low efficiency and in particular low impedance speakers are a problem for any amplifier, not just an OTL. The fact of the matter is that even a transistor amp that easily doubles power from 8 to 4 ohms will not sound its best on a 4 ohm load, for the distortion will increase with the 4 ohm speaker, and it will be both audible and measurable.
Four ohms does help if you are trying to get raw sound pressure, but if you are trying to get the most realistic sound it represents a compromise at best.
Paul Speltz, who markets a product known as the ZERO has an interesting letter from Steve McCormick, stating that while his amps drive 4 ohms easily, they sound much better driving the 4 ohm speaker trough the ZERO, meaning that the amp is actually loaded at 16 ohms. In speaking with other designers of transistor gear that I know from being at CES and the like, they all concur, to paraphrase: "just because it can drive 4 ohms or less does not mean its sounding its best."
I have always built SET and PP tube amps and building proper speaker loads for them has always been a high priority for me.
generally speaking and IMO, the need for globs of NFB to help with linearity and get the impedance down to acceptable levels without adding even more parallel tubes.
Naz
The Atma-Sphere line is an example. How about a 500 watt zero feedback OTL?
Where do you see "zero feedback"? I couldn't find the word feedback on the page linked, the other pages linked from there, or the manual. I'm not disagreeing, I just haven't found anything yet.
Yes, the MA-3 (and the MA-2 and MA-1, as well as the preamps) are zero feedback. The M-60 and S-30 employ 1 and 2 db respectively, but we have customers that have had us build them without any or switchable.
We probably ought to add the words 'zero feedback' to that page...
Is current feedback (degenerative feedback), feedback?
I think you meant no loop feedback.
A circlotron amplifier is two cathode followers, driven in antiphase----bridged.
A cathode follower has lots of current feedback.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
I've been vociferously reading audiophile stuff for more than a decade and I can say this: Nobody I know of, be they designers, reviewers, or hobbyists, ever refers to or consider a cathode follower to be "feedback" of any sort. Indeed, when the word is used, it almost always refers to voltage feedback, and global at that.
I don't mean to be pedantic, but you started it. :)
Seriously, I kind of appreciate you keeping Ralph on his toes here, but, really, you're holding him to a "standard" that is really nothing other than your own. For what it's worth.
It is loop feedback that causes the issues I have described elsewhere, due to propagation delays in the circuit, regardless of the amp.
Degenerative feedback is different as it occurs in real time.
Thanks - that makes sense now.
Agree that any amp can be built with no global NFB but in practical terms that means more tubes with OTL and most use quite a bit.
BTW, I'm not an OTL basher, I think any topology can be made to work. Ulimately, it's a question of personal priorities and again, that synergy word.
Cheers,
Naz
Agree that any amp can be built with no global NFB but in practical terms that means more tubes with OTL and most use quite a bit.
Actually this statement is not correct.
There is confusion in the industry about the term 'output impedance'! Take a look at the link below- under the Voltage Paradigm, the term 'output impedance' does not refer to the actual output impedance of the amplifier, it refers to the servo gain of the feedback circuit.
If adding feedback really did lower the output impedance, we could easily drive lower impedance speakers with greater power simply by adding feedback. This would of course violate Kirchhoff's Law , one of the most fundamental laws of electricity. Obviously you can't do that :) The reality is that to get more power, you need more tubes, bigger transformers (or more transistors, bigger heatsinks, etc.) and the like.
So it is obvious that the term 'output impedance' has become conflated with the use of negative feedback but in the real world the two are quite different. Under the Power paradigm, output impedance is the actual impedance of the output circuit, which will be exactly the same whether feedback is applied or not. IOW, this has a lot to do with how much power the amplifier will make and into what load.
To give you a practical example, we have our M-60 which makes 60 watts into 8 ohms using eight (10-watt) power triodes to do it. The amp makes about 40-45 watts into 4 ohms. Now if you add 30db of negative feedback to it, those power levels don't change at all. If the output impedance were indeed lowered, the 4 ohm figure would be seen to rise. But is doesn't, so we know that the output impedance is unchanged.
This is true of all amplifiers and what we can take from this is that there are special charged terms in the Voltage Paradigm that are used in a way that does not exist elsewhere in the electronics industry (and so it should not be surprising that audiophiles and engineers alike are confused as a result); 'output impedance' is one of them ('sensitivity' in loudspeakers is another). However in the world of audio, the Voltage Paradigm is the predominate means used to do test and measurement- that is what you see in the pages of Stereophile and other magazines when amps and speakers are tested.
I am very much convinced that the use of the Voltage Paradigm has set the audio industry back as it seeks to create numbers that look good on paper rather than create specs that relate directly to human hearing/perceptual rules. In seeking to 'look good' (one of the most powerful human motivators BTW) it literally throws the baby out with the bathwater; anyone who has looked at amplifier specs knows what I mean- you can't tell much about how the amp will sound from the specs.
Great explanation of GNFB and I totally agree. In the real world, GNFB tends to suck the life out an amp the more you use and all who have tested this would probably agree.
My concern was REAL linearity and as you correctly put it GNFB only serves to improve the apparent OP impedance and distortion. So on an apples for apples basis with zero or minimal fixed GNFB, the greater number of tubes in parallel the lower the REAL output impedance and the better the load line.
The tubes used in most OTLs do not exhibit great linearity over very wide voltage swing so if you are managing to minimise GNFB whilst maintaining reasonable power and efficiency your techniques must be commended.
I think this in essence was what Tre was saying and I won't dispute your numbers because you don't stay in business for 30yrs plus without doing something right:-)
Naz
The tubes used in most OTLs do not exhibit great linearity over very wide voltage swing so if you are managing to minimise GNFB whilst maintaining reasonable power and efficiency your techniques must be commended.
I first saw the 6AS7G used in a transformer-coupled amplifier project published by RCA back in the 50s. It didn't use any feedback so I figured it must not be too bad. Some of Futterman's early amps also used them- I think he went away from them due to the number being used to make power. But they are decent in the linearity department, not as good as a DHT, but they are still good, plus they are cheap and in production.
I've also used 6H33s with good results; my favorite would be the 7241 if they were being made! Four of them are good for about 150 watts :)
Ralph, I remember you once saying that you don't want the mu of your output tubes to be too high. And I don't quite know why. The 7242 is kind of a higher mu version of the 7241. (I think it gives up something in terms of Gm to the 7241, however.) But it always tempted me, because with its higher mu, and since the output Z of the circlotron is inversely related to mu, I could imagine it would give low output Z for only a couple of tubes per channel. The 7242 is just as unobtainable as the 7241, but what about the principle involved? Wouldn't a higher mu output tube be a good choice, assuming low Rp and decent Gm?
Lew, the only reason I've stuck to lower mu is tube matching (plus of course that is all that is practically available...). Of course mu is not the whole story by any means, but if you have a higher mu (voltage gain in essence) then you will have to be more careful about matching power tubes. Beyond that I see no reason not to try it- in your amps and in the later Novacrons they are equipped with an independent driver and bias control for each power tube, so at least you can set the base bias point correctly. Its still a good idea to match when you are dealing with such high transconductance as these tubes have! (the exclamation point is a bit of an understatement, BTW...)
I recall being shown a very old issue of Audio mad years ago with a very interesting OTL system. The amp used a rare triode 6337 with 50 ohm output impedance and a special driver was made with a 50 ohm voice coil. Damping factor, of course was low, but power transfer is wonderful.
Ask this question in OTL asylum. Your going to get a whole different perspective.
Bottom line is that most designers can't or won't design a good OTL simply because of the complexities involved. Listen to an Atmasphere OTL amp sometime and then decide if you can live with the tradeoff's of heat and limited drive capability. They have a sound all their own.
As stated below many designs run the tubes hot. In addition, the more tubes, the higher the odds of a failure.
Another issue is speaker damage if DC finds it's way to the to the speaker terminals. Now every DC coupled solid state amp has this same issue but most have protection circuits. This can't happen in a transformer coupled output stage. Primary to secondary shorts are extremely rare and even more so in an audio transformer.
P.S. There is no reason the same protection circuits found in SS amps could not be added to an OTL amp. It's just a DC sense circuit that opens a relay in series with the speaker jacks.
Yes there can be more tubes but we seem to get long life out of them which is why we warranty them for a year.
So far in the last 36 years I have yet to see legitimate damage to a speaker. I have seen people install incorrect fuses or even incorrect power tubes (!) which has resulted in damage, I have also seen amps that make several times the power that the speaker can handle get damaged. But those don't count for obvious reasons. 36 years is a long time and a lot of amps...
Tubes, even 6AS7/6080, 6C33Pi, and other low gain, high current tubes, have output impedances of hundreds or thousands of ohms, even when paralleled. This impedance mismatch causes severe loss of efficiency. Unless corrected it also causes other problems such as asymetric distortion (but feedback and signal conditioning can address that). Output transformers have their own woes.
The root problem of using tubes as output devices for speakers is that tubes are inherently high impedance devices and nearly all speakers are low impedance devices. The two standard ways of addressing this are transformers and paralleling a bunch of output tubes (OTL). Transformers inherently do not like to handle many octaves linearly, and are big, heavy and expensive.
All amplifiers and amplifier topologies are compromises. The design engineer balances one thing against another. OTL and transformers are simply a matter of picking your poisin, or put another way, "Which parameters would you like to compromise and how much?"
I think it's Mazda that's running a commercial right now that says their designs are uncompromised. That's patent bullshit. Every complex device is a compromise of many factors. If their's was a truly uncompromised car, it would (roughly) offer every luxury, never break down, haul cargo like a semi, haul ass like a Maserati, get 100 MPG, and cost $10.
But back to your core question, I think the reason OTLs are not very popular is that they require a bunch of output tubes that generally wear pretty quickly because they are overloaded and don't put out much power for their number of tubes and power consumption. I think of them generally as winter amps. They certainly should be able of sounding better, being flatter, and having less phase distortion than a design that uses transformers.
Mazda's current television advertisements also show a Tektronix 541 (large vacuum-tube CRO, for the younger set) on a large 'scope cart on the tarmac during their testing.
Except for the last sentence :)
1st, the output impedance of an OTL is in line with SETs, even if the OTL has no feedback. It can be much lower if feedback is employed! The tubes involved have much lower plate resistances than'hundreds or thousands of ohms'
as mentioned above.
Also, the lifespan of the tubes is long- in our case we have the longest tube warranty of almost any tube equipment manufacturer, one year.
Nice discussion.
A transformer on the output also prevents backwards EMF, from the drivers from kicking back into the Finals' stage.
It "buffers" the Final's stage, and drops the tubes' inherently high Z to where it does good powering the low Z voice coil.
Modern McIntosh solid state amps use output transformers, advantageously.
Jeff Medwin
"Modern McIntosh solid state amps use output transformers", because it is in keeping with the McIntosh tradition, and I believe those are autoformers, not transformers.
A transformer on the output also prevents backwards EMF, from the drivers from kicking back into the Finals' stage.
I would think that an output transformer would simply step up the back EMF and present it to the source. (output tubes)
dave
the back EMF voltage to signal voltage ratio will stay the same but I agree, it will not be "blocked" or "prevented" from reaching the output tubes.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Yes, indeed. That's similar to the action of an ignition coil in a car (without the coil, there would be no spark).
I'm afraid that statement by Jeff shows a fundamental lack of understanding of transformer principles. However, we shouldn't really blame him; he's been sadly misled by others in a number of areas.
Stick it Ray,
I was taking my information from a discussion I had in the 1980s from none other than Robert W. Fulton. And if I was so incorrect and misled, how come Ralph Karsten chose not to address my post ?
Jeff Medwin
you have been sadly misled by others in a number of areas. Or maybe you just didn't understand?
A transformer (and this coming from Robert Fulton too, I also knew him) does exactly what its name implies. It *transforms* impedance, and this goes both ways.
So if you load the 8 ohm tap with a 4 ohm load, the load impedance on the power tubes will be drastically cut! Don't think it will stay the same- that is a common myth.
Hiah Ralph,
Decades ago, Robert used to talk to me about this fellow (you) developing a nice OTL amp. He was always neutral to positive in referencing you. You guys were both in the Twin Cities, no ??
Never had the opportunity to meet you, but Robert was like a Dad to me. Some are working on a web site in his memory.
Cheers,
Jeff Medwin
When I met Bob it was 1977. It was about the same time I got the idea for the amplifier.
We used to meet at one of his employee's house, that of Bob Fredere. It was there that I also first met Bill Johnson of ARC- they were good friends. It was really through Bob and his association of audiophiles that got me in to high end audio. Some of those audiophile friends are now working at ARC.
I had obtained a set of 80s from a friend, but soon had a set of J's and then a set of Premieres. After Bob F moved away to Utah, I continued to meet up with Bob Fulton at the house of a Fred Devere in St. Paul. That is where I heard Bob's newest work with the Oval Window bass cabinets (a modified Helmholtz resonator). At the time it was the deepest and most effortless bass I had ever heard and still one of the better presentations.
Fred bought a set of our early MA-1s on Bob's recommendations. A few months later Bob died quite unexpectedly due to pneumonia after back surgery. Many people don't realize the contributions he made to the sport; more than any other single person he is responsible for the high end cable industry, which has had a very broad influence- even Radio Shack sells upscale cables now. There are of course his speakers. I think they have been surpassed but they still qualify as high end, and he proved that there was an upscale market for such things - when he was making the J's there was not a lot out there that was competition. Now the speaker manufacturing is one the largest sectors of the high end market.
What a lovely post. Robert was a very special person in audio, because he knew not only the technical aspects, but also he knew what music and real instruments sounded like. I too was friends with Bob and Carol Fredere, Fred DeVir, Warren Gehl, John Tuttle, Tom Moore, etc.We may ask you to write a little piece in the up-coming web site.
Regards and thanks,
Jeff Medwin
Edits: 07/26/12
But, they are "cool" as hell.
OTL does not sound any better with an amplifier using a quality audio transformer.
I have designed and built two very similar amplifiers (same tube count & same tube types) using four 6080/6AS7 output tubes (per channel).
One OTL and the other with output transformer.
Apart from the obvious difference of output power (i.e. maximum sound level), there is a distinctive feature of the OTL one.
It is able to subjectively reproduce deeper and cleaner bass frequencies w.r.t. the transformer coupled one.
This is why I recommend to DIY people to try both amp types.
Best Regards
Luca
ecc230
Interesting statement, that an OTL sounds similar to an amp with quality transformers. Joe Curcio said that his MQ-100 tube amp, using Magnaquest iron throughout, is "equaling the performance of OTL but without the complexity and power requirement disadvantages of OTL topologies".
I've not had an opportunity to compare the two side by side but is this relatively true?
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: