Patrick Turner wrote:

[snip]

I looked at ESL transformers very seriously earlier this year
and the problem is winding one with low self shunt capacitances.

If you have a typical 1:200 step up transformer with a miserly 100pF
across the HV sec,
it appears as 200 x 200 x 100 pF at the primary, ie, 6uF.

Quad's ESL57 had 35pF but their SUT was a wonder to behold, and full of
tricks to overcome the capacitance, mainly by means of using closer
gapped panels for
treble and with less drive voltage, so the capacitances were not
transformed upwards so badly.
The ESL57 had high bass panel C but with deliberately high leakage L it
forms a
low pass filter.

200 x 200 x 100 pF = 4uF, not 6. Never mind :-)

More to the point, isn't a large part of the problem (with ESL
transformers) ours to remove? Meaning that most ESL transformers are
designed to be driven from a 4/8 ohm source. Logical enough, if you're
making transformers to plug on to an existing amplifier.
For example, the ones on Sowter's website, with ratios around 100:1.

But if we have a tube output stage, the B+ is (say) 400V, and we can
have a 600V swing quite easily. Stepping that up to 3kV is only 5:1,
which should be quite manageable.
The drawback is that this is definitely a special transformer. (While
we're winding it, let's put a reduced-power 3rd winding on, to run that
bass speaker as well.)