The EV Circlotron used an output transformer with a secondary although they could have used an autoformer. Since it was using 6V6s or the like it could not operate as an OTL, but did have the advantage of being able to operate class B without crossover distortion since the magnetic field in the transformer could not collapse. So it was high power and low distortion for using a pair of 6V6s!
One thing that OTLs can do is have a distortion vs. power curve wherein the distortion linearly decreases to unmeasurable as power is decreased. The only other amps that can do this are single-ended, which tend to have high distortion at high power. Now the ear has a masking principle, wherein the presence of louder sounds masks the presence of quieter sounds. Distortion can thus mask detail, in particular at lower power levels. This is one reason why SETs have that 'magical inner detail', but its not there at higher power levels. Most transistor amps have a distortion character where the distortion increases below a certain percentage of power, which means that at low power levels they are going to mask detail compared to an OTL as their distortion will be considerably higher.
Unlike an SET, an OTL can make power without high distortion. OTLs traditionally over the decades have had distortion figures that rival solid state- much lower than conventional tube amplifiers. The output transformer, or the lack of it, is the main reason why- you don't have the phase margin issues if the transformer is absent (and you are also lacking the distortion contribution from the transformer), since the tube circuit otherwise can have some very impressive bandwidth (we easily get full power to over 100KHz for example- that's pretty hard in a high powered tube amp; we have to limit the bandwidth in the driver circuit).
Lower distortion results in greater transparency on the part of the OTL (note: our amps tend to lack the famous 2nd ordered harmonic common with a lot of tube amps since they are fully differential and balanced; even ordered harmonics are canceled throughout the amp, not just at the output). IOW what I am talking about is both measurable and audible.
Regarding your other comments- OTLs can be as reliable as any other amp. However that was not the case before Atma-Sphere came along- if anything we engineered the world's first truly reliable OTLs. I often pull power tubes at audio shows while the amp is playing to demonstrate just how stable the amp is- the amp plays on as if nothing happened. I short them out with a quarter across the speaker terminals- again the same thing when the quarter is removed. Tubes can fail but the amp does not. We achieved this by the use of the Circlotron output circuit, but configured in a different way that allowed for a very simple circuit- only one stage of gain exists in the entire amp (which is one of the features that allows it to rule the roost on transparency- the more stages of gain, the more places for things to get screwed up).
Our amps are full power to 1 or 2 Hz depending on the model. They play bass wonderfully (I played bass in the orchestra from 7th grade till well past college; I expect the bass to be convincing). You are actually drawing from much of the prior reputation of OTLs, which I've come to call the 'Futterman Legacy', although it was really Harvey Rosenburg, more than anyone else, that convinced the world that OTLs were unreliable and couldn't play bass. We've done a lot to reverse that reputation in the last 30 years; most newer OTLs coming on the market in the last 30 years have been Circlotron designs, not Futterman designs. Despite that, we seem to be the only one that has sorted out out to really control the power tubes so that don't have bias runaway or blocking distortion issues. Clipping recovery is instantaneous, something I can't say about a number of competing designs.
The heat of the amp is a function of its class of operation- not the number of tubes (our amps are Class A2; another way we reduce distortion). If one of our amps is in Standby all day long, at the end of the day you can grab a power tube without getting burned. But if the amp is stone cold, after a one minute warmup the tubes are too hot to touch. In practice, a class A solid state amp of the same power makes about 90% of the same BTUs, although due to its being distributed across a heatsink, at a lower perceived temperature.
We are careful to vet each customer's loudspeaker plans, as due to the fact that the smaller amps we make do have a fairly high output impedance, the results will be less than spectacular
if the amp is put at a disadvantage on the wrong load. Generally speaking, if the speaker is 8 ohms or more in the bass, it will do just fine and easily competes with the best solid state amps on such loads.
Over the years though in looking at amplifier specs in general, one thing that has become abundantly clear is that in high end audio there are only weak arguments for speakers of 4 ohms and less. All amplifiers, tube or solid state, make more audible distortion into lower impedances. If your goal is high quality reproduction, why force your amp to make distortion you can hear (and BTW, the distortion I'm talking about is higher ordered harmonics, to which the ear is keenly sensitive because they are used to sense sound pressure, and so rivaling the best test equipment)?