OTL Asylum

Posted by Lew (A) on July 06, 2021 at 09:05:40

In Reply to: RE: older Sound Labs posted by Simon Templar on June 08, 2021 at 01:04:05

I bought a pair of M1s around the late 1990s. I drove them with a pair of Atmasphere MA240s. That model was built on MA2 monoblock chassis and used six 6C33C Russian triode output tubes in lieu of a bunch of 6AS7s. In power it is roughly equivalent to a 12- or 14-tube MA1. With the M1s, I never felt short of adequate power to drive the speakers full range. About 8-10 years later, I bought a pair of 845PXs, and to save money, the factory was willing to sell me the panels as a separate entity; they noted that the backplate for the 845PX was identical to that formerly used for my M1s. Note that my M1 backplates had the first or early toroidal transformer which was fed by a first-order high-pass filter consisting of an RC network, where to the best of my recollection, R = 6 or 8 ohms and C = 36uF. The resistance was made up of four of those rectangular "coffin"-shaped resistors wired in series-parallel, so as to increase power handling. So this is the "old" set-up that Ralph mentions that both eats amplifier power and dramatically reduced speaker efficiency, if you use tubes. Moreover, some other guys who posted on Sound Lab Owners Group (SLOG) noted that it resulted in a dramatic impedance dip to ~2 ohms at 1-2kHz. For an OTL, that is NOT where you want to see such a phenomenon. However, the scuttlebutt was that the factory used solid state amplifiers to voice the speaker, and the impedance dip was not seen as a problem by them. Around that time, I was growing more and more dissatisfied with my 845PXs, mainly because I had expected a big upgrade compared to the M1s, and I wasn't hearing it. There was a problem in the midrange that I now believe was due to that impedance dip. Around then, a guy from Australia was posting on SLOG about a cure for these ailments. His name was "Will", and sadly he is no longer with us. Nice guy. I started up a conversation with Will. This led me to determine an impedance v frequency curve for my own 845PXs. That confirmed the low Z at around 1-2kHz, but it also showed that the intrinsic impedance of the speaker without any RC network is much higher than 6-8 ohms, more like 20-25 ohms from about 200Hz to 5kHz. Thus one can understand if you put a 8 ohm resistor in parallel with a 25 ohms impedance most of the energy put into the system is going to flow through the resistor, where it does nothing but heat up the resistor. In addition, that R puts a ceiling on the impedance; it can never be higher than the value of R in parallel with speaker impedance, in the range where it operates on the treble tranformer. I have already made a short story long, but the upshot was that Will advocated dumping all the crossover components, both R AND C, and replacing the treble toroidal transformer with a massive EI type full-range audio step-up transformer that he had sourced in Australia. I immediately bought a pair and that is what I use to this day in lieu of the toroidal treble transformer. No RC network at all. You still need the bass transformer in parallel with the new full-range transformer, in order to derive enough energy to get good bass response. I kept the inductor that precedes the bass transformer, along with the OEM bass transformer, and I have been happy as a clam ever since. Most of this was published on SLOG. You can look it up in the archives I think. All of this conversation on SLOG led Roger West, who was never less than a fine gentleman throughout the process, to revise his design. He did not agree with the idea of no RC network, but he did agree that perhaps the crossover point was in the wrong part of the spectrum, and he introduced a heftier toroid that can accommodate lower frequencies, thereby permitting him to lower the crossover point. I do not know what values of R and C are now used. I absolutely love my 845PXs, ever since I went naked. My speakers present an impedance in the range of about 20-25 ohms from 200Hz to ~5kHz, falling to 2 ohms at 10kHz, where the low Z does no detectable harm, and that's unavoidable since an ESL is a giant capacitor. Below 200Hz, Z goes up. I can drive them to chest-thumping levels at way below full power from my amps. I am guessing 50W would suffice. But I am not advertising for anyone to follow that path; the Australian step-up transformer is now long out of production. I would go with what SL now recommends as their latest set-up.

Edits: 07/06/21   07/06/21   07/06/21