tube amps and electrostatics

Something like 80% of our MA-2 amplifier production has been for Sound Lab owners.

The Quad ESl 63 as been a good match for our M-60 and MA-1 amplifiers. Martin-Logans (which have a lower impedance) usually require a set of ZEROs (http://www.zeroimpedance.com) to work with many tube amps. The CLS-1 is an exception- its impedance was set higher and so is easier for tubes (M-L really wants their stuff to work with transistors, so even though they sound better with tubes, their impedance is usually set lower so transistors will have an easier time with them).

Some of the old Accoustats were an easy load for tubes and some were not, as Accoustat was trying to crack the solid state market too. If one of the higher impedance Accoustats, IME 50-60 tube watts is plenty of power.

The fact of the matter though is that the nature of tubes where they do not always act like a voltage source (which is the ability to double power as impedance is cut in half) is an advantage when trying to drive the higher impedances that an ESL often has. Unlike a box speaker where the high impedance in the bass can be the result of a box resonance, the high impedance of an ESL has nothing to do with resonance and so they still need power to drive them even though the impedance is high. Tube amps do this better than transistors.

As an example, a 600-watt transistor amp can't make any more power into a set of Sound Labs than you might get from a 150 watt tube amp. Put another way, a lower powered tube amp can often make as much or more power than a more powerful transistor amp when driving ESLs.

In short, ESLs and tubes have been one of the better combinations in audio for the last 55 years.

^^ Yup! We shown with Roy's speakers at audio shows in the past. Never had a problem with them.

Regardless, the ability of an amp to double power as impedance is cut in half is not an advantage on an ESL. This is because the speaker really wants to see constant power regardless of its impedance at a particular frequency, as the impedance curve is based on a capacitor rather than the resonance of a driver in a box.

Consequently tube amps have been the preferred choice for ESLs since the 1950s and remains that way today- if you really want to hear what the ESL does, you gotta have tubes, else you are leaving a lot of the speaker's capabilities on the table.

Al, as most here already know, is one of the more knowledgeable contributors on this website. I feel really fortunate that we often agree :)

Al did indeed get it right, as usual. The problem you often run into when trying to use solid state with ESLs is not enough bass and too much treble. When you look at the impedance curve of nearly any ESL, its easy to see why- it is common that the amplifier will be trying to make nearly 10x more power at 20KHz than it might be at 50Hz!

The thing is, as pointed out earlier, the ESL has a low impedance at high frequencies due to its capacitive nature and *not* because it is less efficient at that or any other frequency. It is the latter fact that is often understood by many, (apparently) including Georgelofi; this is why his comments can be safely ignored.

Of course there is not nearly as much energy at 20KHz, but the result is easily heard as brightness. That, coupled with the normal brightness associated with solid state amps (due to odd ordered harmonic distortion) is why such amps do not usually make the best choice for ESLs- even those meant for transistors.

Here is what is real... the feedback used in transistor amps is what keeps them from destroying the speakers. That same feedback is also one of the things that makes a transistor amp bright. It can do that to tube amps too, but tubes can run with less or even zero feedback and have good HF bandwidth even on ESLs.

It appears that George and I are at loggerheads. I have in the past had difficulty taking much George has posted with anything other than a grain of salt, as often (as we see in the post above) the *math gets ignored.*

He is the only one I know of to challenge a technical comment made by Al.

If we can see an ESL efficiency curve that shows that the efficiency of the speaker drops with the impedance curve, I would be willing to concede he had a point. But no such curve exists. What information that does exist suggests that any ESL has the same efficiency at 10KHz that is does at 50 or 100Hz.

There is also personal experience. Something like 80% of our MA-2 production over the last 23 years has been for Sound Lab installations. During that time, we have sold a lot of amps to customers with Quads, Acoustats, Audiostatic and King ESLs. In that time its been really obvious when we encounter an ESL that is designed for transistors- the highs are muted as George suggests. The thing is, that does not happen with *all* ESLs, only some, and there is a fix for that- the ZEROs as I mentioned earlier.

Also in this case I was careful to use the word 'associated' in my comment that he quoted (which excuses it from being a generalization, instead it is a statement taken from the experiences of many people).

Its my opinion that George is grinding an axe.

+1 Mapman

Ralph and Al, the discussion begs the question of how can one know whether an ESL was voiced to be driven by a tube versus SS amp. Obviously Sound Lab ESLs were designed to be tube friendly, ergo Ralph's comment that his amps have been coupled with Sound Lab ESLs for years. Is it the same old answer -- call the manufacturer and ask?? Or try to determine the type of amp that was used by the manufacturer to display the ESLs at shows??

That's easy. Most of the ESLs intended to be driven by transistors tend to have very low maximum impedances. For example MLs tend to be 4 ohms in the bass. When you figure that the typical ESL varies by 10:1 in impedance from the bottom octave to the top, that means that the MLs should be about 0.4 ohms at 20KHz, and a number of them are.

However by use of a set of ZEROs you solve this problem, as I have mentioned before. Let's take the example of the ML and its 0.4 ohm impedance at 20KHz. The ZERO has taps that set up the amplifier load at 16 ohms, assuming that the speaker load is 4,3 or 2 ohms. If you are using the 4 ohm tap that 0.4 ohm load will look like 1.6 ohms. If the tube amp employs negative feedback it will have no trouble making this work with flat (linear) bandwidth. If the tube amp lacks feedback, you might have to use the 2 or 3 ohm tap, and it will still work quite well.

So really what it comes down to is whether the tube amp makes enough raw power that might be suitable to drive the speaker. IME, the Accoustat is a good example. Some versions of the Accoustat, as mentioned earlier in this thread are high impedance, meaning a little OTL like our M-60s can drive them fine across the entire band. Other Accoustats have that dreaded low impedance and require the use of the ZEROs. A good friend of mine had a set of Accoustats like that, and he used our M-60s with a set of the autoformers. The amps made plenty of power- and the setup was in a room 17' by 24', and we never clipped the amps.

The point here is that one must not confuse efficiency with the impedance. The ZERO is a problem-solver here- it allows almost any tube amp to effectively drive almost any ESL, provided the amp makes enough power in the first place.

Rodman99999, Thanks, I think I see what you mean.

Bruce, I would not loose heart!

The issue that makes ESLs hard to sort out is that their impedance curve is related to a capacitor rather than drivers in a box.

But the allure is of course that ESLs are very fast and revealing, and if set up right, not painful for all that extra detail.

I agree that some things about ESLs can be counter-intuitive. For example many magazine articles have been written about how hard they are to drive, difficult load yada yada, but that really isn't the case.

OTLs, as you know, have a reputation about being load sensitive, but oddly have also been the preferred combination with ESLs going back to the 1950s. We have made that even more difficult for ourselves (since we make OTLs) by removing the feedback from our amplifiers, yet still they are able to drive ESLs quite well. So the 'difficult load' thing is really a common myth not supported in the field.

What *can* make an ESL hard to drive is when the manufacturer of the ESL decides he wants the speaker to be more transistor-friendly. MLs are a good example- 4 ohms in the bass, 0.5 ohms at 20KHz. Now there are many who would say that is a difficult load, but there are no crazy phase angles, so with proper matching (a set of ZEROs) such speakers become quite easy to drive with tubes since its more about impedance than anything else.

If you are contemplating going from a box speaker to an ESL, you may have to deal with things like having the speaker further into the room to make it play right. However the energy that comes off the back of the speaker is used by the human ear/brain system to get a better idea of where the sound is coming from; IOW to improve the soundstage imaging.

So I would not write off ESLs just because some things about them are counter-intuitive. If you have questions about whether a certain ESL will work with your ARCs, all you have to do is ask... I can tell you though that the combo will be good with Quad ESL57s and 63s, you will want the ZEROs with their later models, similarly with nearly all MLs. You will be able to drive Sound Labs directly no worries (however you will want the latest version of the backplate as it corrects a crossover error, making the speaker a lot better sounding and easier to drive), also King, Audiostatic and Accoustat. You will need a set of ZEROs with the Sanders ESLs as they are low impedance.

Also, your amp will be able to keep up with solid state amps, often with the ability to play with as much power as solid state amps of considerably higher power rating. This assumes the match is good (meaning that even if a set of ZEROs is needed, this will still be true).

^^ yes. But then you are dealing with 20 ohms or more in the bass. A 600-watt transistor amp just became a 250-watt amp.

This is why tubes work so well- the higher impedances are not as challenging. In the case of the Sound Labs, this means a 150-watt tube amp can keep up with a 600-watt transistor amp no worries.

+1 Lacee

Bifwynne, speakers get a reputation for being 'hard to drive' for a reason. What you might want to think about is the impact on the amp. It does not matter tube or transistor, if the amp is driving a hard to drive load the result will be more distortion from the amp.

For this reason you are usually better off, if sound quality is your goal, to get a speaker with higher impedance (as they are often easier to drive), regardless of the type of amp. Now if **sound pressure** is your goal, then a lower impedance speaker can help if you have a transistor amp.

But usually in high end audio we are more concerned with finesse. So this is something to consider! FWIW, most loudspeaker designers have not sorted out this simple fact that I have posted above. So you will continue to see lower impedances regardless of the speaker technology.

IOW if you were simply able to raise the impedance of a speaker without changing anything else, it would sound better as the amp driving it will sound smoother and more detailed due to a reduction of distortion. ESLs are no different and this is why the ESLs with higher impedance curves also have the most loyal following (they sound better). That is why the *ancient* Quad ESL57 is still around and still has loyal fans. Its easy to drive.

Bifwynne, another speaker you might look at is Audiokinesis. His speakers are not only more efficient, they are also very easy loads (some of them are 16 ohms) without crazy phase angles, yet also quite musical and good bass extension.

I think you are correct that speaker manufacturers ought to be taken to task. Back in the old days when tubes were the only game in town, speakers were a lot more efficient and higher impedance. When the less expensive power of solid state came along, we saw both impedance and efficiency go down.

When transistors came along, amp manufactures realized they could build an amp with same power as a tube amp, but it might only cost 1/10th as much, yet they could still charge 90% of the tube amp retail cost. A similar thing was going on with speakers- its costs about 1/10th as much to build a lower efficiency driver (as opposed to a high efficiency driver). To get back some of the perceived loss of efficiency, the impedances headed south. IOW, its all about the dollars.

IMO its telling that you can have two speakers, one 10X more efficient, and the more efficient one can have the same bandwidth and be just as revealing, IOW not giving up anything for being more efficient.

I can't think of a good reason for a high end audio loudspeaker to be less than 8 ohms. I've mentioned this plenty of times before- the lower the impedance, the higher the distortion, making any amplifier harsher and less detailed due to the types of distortion and our human hearing/perceptual rules. IOW if you want to make a speaker that seems smoother and more revealing, all you have to do is increase its impedance (all other things being equal- same box characteristics, same crossover points). It will not be that the speaker itself is all that different, but the sound of the amp driving it will be!

The B&W 802D is, FWIW, an excellent example of Voltage Paradigm technology.

The woofers are each 8 ohms, but also 3db less efficient than the 8 ohm mid range and tweeter. They are in parallel, which requires the amplifier to double its power into the woofer load. This brings their output up by 3db, causing them to match with the rest of the speaker.

Yet the spec on the speaker says they are 'nominally 8 ohms'!

The requirement of the speaker is such that almost any amplifier driving it will have to have a fair amount of feedback to do so. This will cause the amp to be un-naturally bright. IMO this makes the speaker a poor choice as you will not be able to find an amplifier that will actually cause the combination to sound like real music.

This is not true of ESLs.

Impedance can be measured with a potentiometer of about 50 ohms or so, a signal generator and a DVM. It takes some time, as with this technique you have to put points on a graph.

You put the pot in series with the speaker and drive the combo with the generator. The pot is wired as a rheostat, and adjusted until you see the same voltage drop across the resistor as you see across the speaker terminals. You then measure the value of the pot and plot it on the graph for that frequency. Then you move the frequency and repeat. This takes time to do 20-20KHz, but it works well.

However there are computer programs now that can make the job a bit easier.

The impedance of the speaker does not tell you if it is Voltage or Power paradigm- its easier to find that out by asking the manufacturer what sort of amp they use. But this can be handy to sort out if you have a difficult load in the impedance curve- as we all know, some amps might overheat or the like if the impedance is too low. So it is very useful for that sort of thing.

It was using a technique like this that some Sound Lab customers discovered a few years ago what the Sound Lab impedance curve really was (and it was a lot different then Sound Lab had said at the time). Apparently Sound Lab had used a simulation that had a bug in it. This has resulted in Sound Lab making some changes that not only made the speaker a lot easier to drive for any amplifier, but also made for a better sounding speaker.

Bombaywalla, yes, Roger sets the impedance of his ESLs quite low, to make them more compatible with solid state.

...the Moscode 600 is a hybrid employing Mosfet output stage, even though it has a lot of watts!!, mosfets don't do current like a BJT (bipolar)output can
This is why it wouldn't drive the ESL's, as with ESL's the impedance curve can dive to below 1ohms at places and very capacitive as well, and for this a solid state amp should have a BJT (bipolar) output stage that is high current and be very stable into capacitive loads.

Georgelofi, I'm pretty sure that is not the right explanation. Keep in mind that a much smaller amp with tubes, the Dynaco ST-70, was able to play the Stax, with only 35 watts. Those older Moscodes had problems (IMO) unrelated to the fact that they used MOSFETs (IMO the topology itself was an issue).

We had customers play the Stax with our original MA-1s (which back in those days made about 100 watts) and they seemed to play well together. The old MA-1 was an OTL with no feedback and a fairly high output impedance, yet sounded find with the Stax. So 'current' is not the answer. Stable into capacitive loads- that makes more sense as that is something true of both the ST-70 and the MA-1.

And what you would have also heard was a more defined and extended treble because those quads and these (ML's Acustats and other els's) go below 1ohm in the treble and start that impedance dip at 10khz.
Tubes I aggree with these speakers still sound good but they are definately subdued in the treble compared to a GOOD s/s amp that is BJT output and that can do current into low impedances. Like you found with the ML2 which can almost keep doubling it's current down to 1ohm for each halving of impedance.

This is not entirely true. Quads and Soundlabs have a higher impedance at 20KHz. In the case of the ESL57 and ESL63, the 20KHz impedance is high enough that even our small M-60 (60-watt OTL) can drive the speaker easily with no loss of HF performance.

In the case of ML, later Quads and Sanders ESLs, because the manufacturer is wanting to open up their market to more transistor amps, the impedance is quite low at 20KHz (some MLs are only 0.5 ohm).

In cases like these, you do as Tomcy6 mentions- you get a set of ZEROs (http://www.zeroimpedance .com) which allow you to easily drive ML or whatever! We have several customers doing just that with our older MA-1s (12-tube version) that only make 100 watts/channel, and they have no problem whatsoever making high frequencies on the speaker, with proper response to 20KHz and beyond.