Using tube amp with electrostatic speakers.

Lewm,

Yes, I did ask many questions with Dr. West who was more than gracious with his time and help.  He told me that a 300W SS amp was more than enough to drive the B1s.  I am currently using a Crown K2 (350W @ 8ohms) on the B1s.

I am trying to absorb and understand your modification.  My speakers did not work initially so I had new mylar installed by Soundlab on the A1s and the B1s.  My panels had the Toroid l in it.  A few years ago, I found someone selling upgrade parts that they did not install in their system which I purchased.  The upgrade parts were a large 36uf capacitor (pio) and the Toroid ll transformer (the latest version).  I consulted Dr. West who confirmed the parts.  I removed the brilliance control and replaced it with a four ohm resistor (which was about the setting I had it at).  Dr. West told me I could remove the inductor coupled to ground in the circuit but I have a 12 or 14 ohm low inductance resistor to ground.  I believe the resistor was larger than what was in the circuit but Dr. West said that the Toroid ll can extend lower in frequency and that it would be fine.

A couple things mentioned by Dr. West.  He said the B1S can be used as a full range speaker which I found interesting.  He said that when they used the B1S with the A1 panels they would crossover at 500hz.  Running the A1 at 500hz would essentially be operating the A1 with only the toroid ll.  I tried it but I couldn't get it to integrate correctly (I will likely try it again in the future). It sounded better at an 80hz or less crossover.

Once again, thank you for your knowledge and help.  It sounds that I can drive the A1 with the tube amp with the 2 or 4 ohm tap.  I question whether I will have the voltage swing with a 60w amp for the midbass if the impedance gets high from 100hz to 1khz.

Best Regards,
Gary

One other thing.  Have you ever tried the zero transformer?  It sounds like it would help especially with low power amps that could have trouble driving the difficult load.  The only thing that could be at issue is the lower frequencies that get multiplied up to 40-200 ohms.  I don't know if this would be an issue for some amps.

Gary, Mea culpa.  I forgot that not only were there several different combinations of R and C used by SL over the years, but also at one point they did introduce an inductance in the hi-pass filter, which creates a 12db/octave or second-order slope.  I think that's the inductor you talk about.  I briefly experimented with that part (after purchasing a pair from SL), long before I made the major modification I described in my recent post.  For reasons I no longer can recall, but likely because it sounded bad, I ditched the inductor in the hi-pass filter and reverted to the simpler RC filter, which gives a 6db/octave slope.  I also forgot about the "brilliance" control, which is anything but brilliant.  Ditching the brilliance control was one of the first things I ever did to improve my speakers; I replaced it with nothing, no resistor even.

One must keep in mind that Dr. West is an engineer first and foremost.  Thus it seems he is more concerned with measurements than with how certain modifications affect perceived sound quality.  This means he does not like impedance peaks within the audio band, for example, but he seems less allergic to impedance dips.  At the same time, I regard him as a very fine person in all respects and someone who knows much more than I do about the technical aspects, of course. (Goes without saying, actually.) He communicated with me and Will, when we posted our findings with the AU transformer on SLOG.  Subsequently, he had the open-mindedness to replace the toroid he was using for treble with one that has a lower low end response, in recognition of what we were trying to "fix" in the first place.

For one example of his EE emphasis, Dr. West pointed out to me that my "solution", using two transformers in parallel where one is full range, results in a sharp impedance peak at 250 Hz. (I think he found this by computer modeling, not by actually measuring one of his speakers driven by the Australian 1:90 transformer that I use, because he never purchased one.)  When I made my measurements, I took impedance readings at 200 Hz and 500 Hz , thus completely missing the impedance peak.  (I used a log progression: 20, 50, 200, 500, etc) But the peak (assuming it is there in the real world) does absolutely no harm to the capacity of the Atma-sphere OTL to drive the speaker; there is certainly no discontinuity perceivable.

Just out of curiosity, what value of R does SL use with the latest toroid, if they use 36uF of capacitance?

Lewm,

Congrats to you and Will on the work you have done in bringing forth a significant improvement to the Soundlab speakers.  I can attest that the transformer upgrade was a significant improvement over the prior toroid l transformer.  It is not very often that a user pioneers improvement that results in commercial changes.  Kudos!

Yes the inductor was in parallel with the resistor.  I only have the resistor now (inductor removed) per Dr. West (another thing you probably had your hand in).  I do not know what they are using now for a resistor value but I believe it is the same (36uf and 10 ohms (4-2.5ohm potted resistors)).  As mentioned, Dr. West stated that the resistor value could be increased with the Toroid ll but not by a significant amount (I have 12 or 14 ohms).

What are the consequences of completely removing the brilliance control that is in series with the transformer?  Does it make the speaker much brighter?  Does it have some effect on the transformer at high frequencies or is it reliability issue? 

I would also like your final thoughts on the topic of the thread which is using tube amps with electrostats and more specifically Soundlab electrostats.   The Soundlabs obviously have wild impedance swings from HF to LF.  From the thread I assume that a tube amp should be selected to be able to handle the HF low impedance (with the zero transformer being a nice option to aid the tube amp).  Are there any issues you know of on why a tube amp cannot drive a high impedance load such as 30-200 ohms?

Once again, thank you for all your help.  Your knowledge has aided me tremendously.  I really like the fact that even old Soundlab speakers (like mine) can be upgraded and made to sound great.

Regards,
Gary

4 of those 2.5-ohm "coffin"-type resistors?  They are probably wired in series/parallel, but do you know the net resistance of the bundle?  Could be 10 ohms or could be 2.5 ohms.  Neither value is gonna make your OTL any happier than mine was before I chucked the crossover. (Really, give all credit to Will; all I did was follow his lead because I was so unhappy with the sound I was getting from my 845PXs.)  You are definitely better off with 12-14 ohms than 10 ohms.  However, I am a little surprised at the values; they are identical to what I was using with my "old" toroid (10 ohms/36uF), before the big modification (removing the crossover entirely).  The -3db point of the high pass filter is inversely related to the product of R times C, where C is in Farads and R in ohms.  For a 10-ohm resistance with 36uF, I calculate a crossover point of 440 Hz.  I would have thought they could use a much lower crossover point with the new toroid, but I bear in mind that Dr. West is very conservative in his design so as to favor reliability.  From my experience making measurements, the intrinsic impedance of the speaker is much higher than 10 ohms at all frequencies up to somewhere between 2kHz and 5kHz. At 5kHz, it's about 8 ohms for my 845PXs.  (Bear in mind, measurements were made with NO crossover, driving the SL factory bass transformer and the full-range Australian transformer in parallel with each other, directly with an audio frequency generator.) So, as you can imagine, when you interpose the RC network, at low and mid frequencies, much more of the energy is going through the resistor than through the speaker (R is smaller in value than speaker Z at those frequencies, so R dominates). This may be why some guys on SLOG discovered long ago that very high wattage resistors seemed to sound better.  There is almost no limit to the wattage ratings they tried, more than 100W, even.  Before my epiphany, I was using an enormous 10-ohm resistor hand-made by one of the British contributors to SLOG.  It could tolerate at least 500W. (One benefit of bi-amplification is that only the treble amplifier then "see"s that R in parallel with its output; the bass amplifier drives the bass transformer only through the inductor.  Thus paradoxically, your SS amplifier might work best on the treble (because it can handle the power demand and the low impedance created by the R, with your tube amp on the bass).

If you go up to 20 ohms, the crossover point would be ~220 Hz. (The beauty of the equation is that once you know the values for one set of parameters, you can calculate in your head what will happen if you change R or C.  If 10 ohms and 36uF give you 440Hz, then 20 ohms and 36uF give you 220Hz, etc.)  I would have thought that the new toroid would be quite happy at around 300 Hz, not too far from where you are at if you are using 14 ohms now.

There is no difference between bypassing or removing the brilliance control and setting it to zero. But I preferred to remove it; it's a cheap part that muddies up the signal path. You will hear an improvement. No, the speaker is not too bright without it, in my opinion.  The easiest thing to do is to re-solder the input and output wires to the brilliance control together on one binding post of the control.

As far as impedance variations of the speaker per se, the ESL can be thought of as a giant capacitor.  It has very high impedance at very low frequencies and then gradually falls off in a fairly linear manner to very low impedance at very high frequencies.  I wrote here earlier that in my opinion, very low impedance at 10kHz or 20kHz is not a real problem, because the energy requirements are very low at those frequencies.  What makes for problems is the circuitry that comes before the panel, e.g., that 2-ohm impedance dip at about 1kHz that came with the old toroid due to its crossover.  Nearly every company does something to make ESLs more friendly to SS amplifiers, which in turn makes it hard on tube amplifiers.