What amp to build for a ESL speaker?

[SM7UYJ]

I have a pair of ESL-speakers that are a quite nasty load for my amplifier.
Impedance is 2 Ohms over most of the spectrum, but drops to ~0.5 Ohm at worst.
Design goal would be a amplifier that can handle speaker loads as low as 1 Ohm.
Any suggestions of what designs I should look at?

/R

[SM7UYJ]

No one - or is it so obvious and I've missed it?

/R

[audiobard]

Hi,
People use Hafler DH500. You need something with high power for big voltage swings to get high spl from electrostatics,but also that can deliver high current because of low impedance, and high stability because of capatience of the ESL.Amplifier with mosfets is better for low impedance because there is no beta droop with high current operation, but they are more prone to oscilation in capative loads. So high power and exellent stability.

[Calvin]

Hi,

a low ohmic impedance value is not the main problem for most amlifiers.
The true problem is the imaginary part of the impedance and it´s influence on amp behaviour and power needs.
The more complex the impedance, i.e the larger the value of phase shift the bigger the problem. Most amplifiers feature a feedback loop, sensing the output of the amp and feeding it back to the input. This loop determines the behaviour of the complete amp in first case and it works sufficiently well only within limits. The load impedance alters this behaviour and can in extreme cases push the system over its limits. Magazines test amps usually with less than 60° phase shifting loads. A panel fed via an very lowloss transformer can easily reach phaseshift values >>60° (I measured values of up to 86°). Such large phase shifts push a lot of amps over their limits. They start to oscillate and can eventually blow up, if the safety circuits don´t cut the line in time.
The second effect of a very complex load is that it demands lot of reactive power. The power the amp has to deliver consists of two components. The real power that is used to drive the load and reactive power that -to say so-
oscillates between powersupply and output transistors, thereby generating losses in form of heat. The phase angles describes the relationship between the usable real power and the wasted reactive power. If the load shows i.e a phaseshift of 30° this means that for every 1W of total power the amp has to deal with there are only 0,86W of real and 0,5W of reactive power. For 45° the relationship is 0,7Wreal/0,7Wreactive. For 60° its 0,5Wreal/0,87Wreactive. For 80° its 0,17Wreal/0,98Wreactive. This means that the amps power supply, output transistors and heatsinks must be able to handle the additional reactive power that can be much larger than the real power. Amplifier power is specced with a real load, hence 0° of phase shift. Having to work into large complex loads reduces the real power the amp can deliver considerably.
So its two effects. The direct influence of the load on the feedback loop and hence the amps behaviour and second the increased power demand.
Luckily with ESLs low ohmic impedance values go together with lower values of phaseshift.
Amplifiers that work stable need a lot of ´phase reserve´ to cope with the additional phasshift introduced by the load. Amplifiers that don´t work with global (overall) feedback, hence that take the feedback not from the output itself but e.g from a point ahead of the output transistors, or that omit with overall feedback completely (often to find amongst tube amps, e.g. SE Triodes; too the output transformers of tube amps help fighting against large phaseshifts).
Apart from a stable working circuit such amps will feature ´oversized´ power supplies, very power capable transistor arrays and large heatsinks.

Class-D amplifiers are much less affected by complex loads and can easily be built with loads of power. But those that take feedback from their outputs are affected similar to class-A, class-AB amps. Those which take the feedback from ahead their output filter react strongly with a peaking amplitude response at the upper bandwidth limit that has to be equalized (output coil and complex load form a resonant circuit). Besides the sonic fingerprint of any class-D I tested was rather ´uninspired, technical noted´ but not ´involving, or emotions evoking´. Sounded like reproduction, not life. Still though they seemed to have the strongest ´grip´ at the panels and the advantage of a stable drive at relatively low cost may be the decisive point for a class-d amp.
If its for pure sonics a powerful(!) SE Triode without global feedback is probabely the holy grail.....but this comes with a terrible price penalty

jauu
Calvin

[SM7UYJ]

Tnx!

I actualy had my former panels driven by a 15W SE no feed back MosFet amp!
It lacked a bit in bass control but everything else was top notch.
Perhaps that is the way to go - build a beefed up version with more power?

/R