Thought on OTL tube amps

Here is the issue with OTLs and speaker loads. There are two major concerns. The first is this:

http://www.atma-sphere.com/papers/paradigm_paper2.html

In short you want a speaker that is designed for tubes. The second is that the smaller you make the OTL, the more it will want higher impedances to really perform right.

A lot has to do with the circuit of the amp and you should check with the manufacturer on this, but in general you will find that the triode-based OTLs can driver lower impedances as their output impedances are lower. Now this brings on a discussion of loop negative feedback, which is a common method of lowering the output impedance on any amp, but here is something everyone should know:

The addition of negative feedback will not allow the amplifier to make more power into lower impedances.

This is because loop feedback really doesn't change output impedance! If it really did, feedback would give you greater power at lower impedances, and it doesn't.

There is an additional factor that comes into play here. Put plainly: There is no argument for 4 ohms or less in high end audio, if **sound quality** is your goal. No matter what kind of amplifier you have, transistor, class D, transformer-coupled tube amp, SET or OTL, the simple fact is that all of them sound better at higher impedances simply due to the fact that they will make less distortion in one way or another. In the case of transistor/class D amps, you get less odd-ordered harmonics, which BTW the ear finds to be very objectionable. In the case of all tube amps, you get less lower-ordered harmonics (2nd, 3rd, 4th) which the ear interprets as 'bloom', 'warmth', 'thick', etc. these lower orders mask detail so if they are gone the system is instantly more transparent without additional brightness (brightness is associated with the 5th, 7th and 9th harmonics).

So the bottom line is you want a higher impedance in the speaker if you can get it, all other things being equal.

Now OTLs, while having an economy of scale, do not suffer the way most other amps do when you make a bigger version. If the designer has done his homework, bigger OTLs will never loose PRat; they will be just as nimble as a smaller amplifier without loosing detail. The other advantage of OTLs is that if built right, they do not suffer crossover or notch distortion issues at the zero crossing point- the only other amps like that are SETs. The point here is that you can get a bigger amp to drive lower impedances if needed.

Now with respect to the impedance curve of the speaker: it is a common myth that OTLs need a *flat* impedance curve. That's not quite true- if the speaker impedance is too low for the amp, then yes, that's a good idea (and it usually is anyway). But otherwise, what an OTL is **really** looking for is that the entire impedance curve be high enough- often that is a lot more important than a flat impedance curve.

A good example is the Sound Lab ESL. This speaker has an impedance curve based on a capacitor and some inductors and has nothing to do with box resonance. This is why transistors have such a hard time with this speaker- its as high as 50 ohms in the bass and only 1.5 to maybe 3 ohms at 20KHz. That's quite a spread. Now a transistor amp can make lots of power in the highs, but hardly any in the bass. So they tend to be bright with no bass or just seem like they can't drive the speaker. You put an OTL on there and suddenly the speaker has bass authority and extension that the transistors could not do, and with an OTL that might be 1/3 the power of the SS amp will make the same volume.

There is an outboard device called the ZERO that is an excellent problem solver for tube amps (OTLs in particular) if you simply *have* to use a 4-ohm speaker. The most common question we get is 'doesn't that obviate the OTL quality?' and the answer is No. The ZERO loads the amp at 16 ohms and has only provisions for 4,3 and 2 ohms. The turns ratio is so low that it has bandwidth from 2Hz to 2MHz, exceeding the bandwidth of nearly any amp its hooked up to. It also does not block DC as its an autoformer. This sort of thing requires that you have an amplifier that already has a fairly low output impedance- you could never do this with a conventional output transformer!

Ralph, thanks very much for the highly informative post.

I have one further question. In this thread, and I believe in another current thread, you alluded to PRaT as something that can be designed for, within the constraints of certain tradeoffs that depend on the type of amplifier. Could you elaborate on what technical factors and design characteristics affect the PRaT of an amplifier? That's something which has never been clear to me. Thanks!

Best regards,
-- Al

I said yes to most, but you are correct the Joule is not lean I could live with this one.

Sure. The issue is this: there are human hearing perceptual rules and there are the specs that we see on paper, and surprisingly they have very little to do with each other. So if you design your amp to look good on paper, it at best will sound like a good hifi.

What we really want is for it to sound like music. For that, we have to get the circuitry/stereo to obey as many of these perceptual rules as we know about. This often results in amps and speakers that don't measure so well. For example, the human ear uses the 5th, 7th and 9th harmonics to determine the volume or sound pressure of a sound. Electronics have the ability to distort these harmonics (enhance them); this results in such electronics sounding artificially louder than they really are. We see this all the time- 90% of the time if an audiophile's wife is screaming them to 'turn that down!!' its likely because these loudness cues exist in un-natural abundance.

You can design to minimize the distortion of these harmonics. The first step is to not use loop feedback, as it is known to **enhance** such harmonics. Of course, then to get rid of distortions you have to use every design trick in the book to try and not make distortion.

IOW what we want to do is engage the human limbic system rather than the cerebral cortex. So in addition to distortion issues, the circuit or system has to be fast enough; if too slow the processing of the sound moves to the cerebral cortex. IOW the experience of the sound becomes intellectual instead of toe-tapping.

This is a sort of basic introduction but I think is also the easiest to understand. Obviously without feedback the THD of the amplifier tends to go up, but THD is a description of a settled-out sine wave and really says nothing about the behavior of the amplifier with a constantly-changing non-repetitive waveform, which is what nearly all music is. The evidence right now is that negative feedback actually *increases* distortion with non-repetitive waveforms- quite the opposite of its intentions.

If you look at this from the field of Chaos Theory, what we find is that the formula that describes feedback in an amplifier is nearly the same as we see in Chaotic systems- and not what we want for reproduction. Chaos Theory suggests that amps with feedback exhibit a chaotic behavior which will include harmonics of the fundamental waveforms up to and exceeding the 81st harmonic! I find it interesting that as the science of math evolves, it is opening doorways to improvements in audio that many people have thought were far too cut and dried for there to be any serious evolution.

JWM

The Atma-sphere MA1 amp on my Soundlab A3 speakers is anything BUT lean sounding. I find it to be full of bloom and weight.