Zero feedback and class A,which is better?

'Lurk' is hardly the term!

For a speaker like this, a zero-feedback amplifier will need a little help. Fortunately there are two things going for you- the first is that the speaker is an ESL and the second is that there is a way to deal with the impedance.

The ZERO is the way to deal with the impedance and will allow any zero-feedback amplifier to operate the speaker so long as the amp is able to make enough power into a benign load.

The fact that the speaker is an ESL means that the speaker is expecting the amp to make constant power regardless of the load impedance of the speaker (which is why transistors for the most part tend to sound bright on ESLs). The ZERO will present the amplifier with the benign load that we are looking for, and a zero-feedback amplifier will have the proper 'constant power' characteristic.

Class A is important here as a zero-feedback amplifier does not have feedback to reduce distortion. Distortion is reduced by other means- class A is certainly one of those ways as it is the lowest distortion class of operation known to man. Feedback oddly enough causes some types of audibly undesirable distortions and so you are seeing a good number of manufacturers, tube and solid state, that eschew the technique. To do so means that component quality and topology suddenly got a lot more important to the overall success of the amplifier!

So the answer to the thread is: both are almost equally important.

T_bone, you got it exactly right. Most SS amps make constant voltage with respect to load; on ESLs that is not the right response.

With respect to the ZERO/OTL thing, we made a device similar to the ZERO about 10-15 years ago. It got to be a bit of a chore explaining why it did not negate the 'OTL character', so eventually we stopped. Paul Speltz picked up the slack.

Here is the explanation: OTLs already have low output impedance and do not make DC, so some of the bigger requirements transformers face have been reduced. To start with, the device does not have to block DC, so its an autoformer with no secondary. This is a lot simpler. 2nd, we're only going from 16 ohms down to 2,3 or 4 ohms, so the turns ratio and distributed capacitance is such that the device has bandwidth out to 1MHz, something that is impossible for a regular transformer. In fact the technology only works if you have a low output impedance and no DC- so its fine for an OTL- particularly in the light of a speaker load that the amp would not work with otherwise! Finally, in the case of our amps, you do not negate the triode-class-A-single-stage-of-gain-fully-differential-zero-feedback aspects. So it is a nice solution. Paul sells a lot to the single-ended crowd as well.