Modern Class D amplifier recommendation for ML electostats?

[DonH56]

In @SIY 's review, he mentions/notes the SR (sample rate) of the AP's A/D in various tests, but not the actual measurement bandwidth he selected. So who knows what's going on with the HF THD. I reckon the steps are just the harmonics falling off the top of the measurement bandwidth. And why 12.5kHz? So the first harmonic is outside the measurement BW?

No, the "steps" are the same as we've seen before, nonmonotonic THD at middle power levels for the higher test frequencies. Stuart's are a little different than Amir's, perhaps partly difference in analyzer and frequency sweep resolution? This is the plot from Amir's data here on ASR (to avoid any copyright issues pasting Stuart's picture):

I'd love to know the cause, don't remember if I have heard it. Rising at high frequency is typical of all amplifiers due to reduced feedback (loop gain falling), but the GBW of these class D amps is stinkin' high, and falling loop gain does not explain the falling THD after the peak as power continues to increase. Probably obvious, just beyond the grasp of my little pea brain, but cannot say I have thought about it much. @SIY, do you know?

 

[SIY]

to avoid any copyright issues pasting Stuart's picture

Attribute it and I'd absolutely call that Fair Use.

 

[Oniiz86]

Why did you exclude Nilai? That's been the best I've tested to far with unreasonably capacitive loads.

@SIY I was just curious does Purifi's 1ET400A perform poorly in this regard, I believe you reviewed it favourably here https://audioxpress.com/article/fre...d-audio-bosc-and-purifi-audio-eigentakt-eval1

Is Hypex's Nilai & NCx500 OEM amps significantly better in this situation?

 

[SIY]

@SIY I was just curious does Purifi's 1ET400A perform poorly in this regard, I believe you reviewed it favourably here https://audioxpress.com/article/fre...d-audio-bosc-and-purifi-audio-eigentakt-eval1

Is Hypex's Nilai & NCx500 OEM amps significantly better in this situation?

The Nilai was better with a capacitive load on the test bench than the NCx or 1ET400A. How that translates to the MLs is not something I can answer since I've never had a pair of those, but all of them worked absolutely fine with my Quad 988s. The Nilai might be the safest choice for this particular oddball load.

 

[pogo]

The Nilai was better with a capacitive load on the test bench than the NCx or 1ET400A.

Does the PSU design used not also play a major role

 

[RayDunzl]

The capacitive load of the electrostatic panel is (at least in my case) fed by a step-up transformer (maybe 100:1 winding) to give the signal the high voltage necessary to drive the panel.

Is that considered in the above conversations?

 

[DonH56]

The impedance of every ESL I've seen or measured gets very low at HF due to the capacitive load of the panels. BUT, the transformer coupling means the phase is often actually inductive and not capacitive. I have not looked recently, but a few years back looked at impedance plots for a handful of ESLs and most but not all were inductive over ~1 kHz. Some were capacitive, however.

I can see the PSU limiting power into low impedances but I do not see how it would affect stability unless decoupling is inadequate.

 

[Sokel]

I have tortured ESL 57's and Mangepan 3.71 with 1200as2 and with purifi 1ET400A (and Audio Research monos too but that is not for this thread)
Both class D's could drive them loud enough in a big-ish more than 100m² room .Ice seemed to go more if needed,Purifi was really good but close to its limits.
Go for the big Purifis since you need full range.

 

[Blumlein 88]

Soundlabs and Acoustats are very much like a capacitor at the upper end. Both reach very low impedance and the phase is lagging by 75 degrees at 20 khz. The phase lags less above 20 khz, but never becomes inductive.

IcePower amps are known to work fine on them in my experience.

I don't know the impedance curve, but Stax electrostatic speakers had a reputation for burning up amps by causing an oscillating condition.

The change to inductance just above 20 khz on my Quad ESL 63s caused tube amps connected to them to have a mild ultrasonic resonance. With some it was enough to just lift the response a bit below 20 khz.