Solid-State High Voltage Amplifier

[G-Daddy]

Suppose I wanted to build a high-voltage solid-state amplifier, say, for electrostatic speaker direct-drive operation. . .

As in, the voltage out of the amplifier is on the order of 1 kV, and the total power comes out around 100 W (giving 0.1 amps of output current). What would I need to consider that would make it different from a standard solid-state 100 Watt amp? Is it even possible for solid-state output devices to handle this kind of voltage? Can I parallel/series the output devices to share the voltage, and still have a decent sounding amplifier? Is it complete folly to even consider this an option, when direct-drive tube amplifier construction is possible (if not easy)?

I seem to remember seeing at least one solid-state amplifier specifically made for electrostatics, like http://www.innersound.net/amp_whitepaper.htm , though I don't think the speakers these were aimed at were transformerless.

Also, the stats on that amp quote the deliverable power to electrostatics in Volt-amps, and to all normal speakers in Watts. The above whitepaper defines a volt-amp as voltage time current, though "the voltage and current are out-of-phase with each other". I assume they mean you take the voltage and current at their respective maximums. So what does this mean for an amplifier? Will an amplifier which is able to consistently and cleanly deliver current and voltage out of phase with each other be different in design? This will be a necessary feature, since the amplifier will be directly tied to the ESL panels, and these offer a mostly capacitive load.

Thanks everyone for any help you can offer.
- Jonathan

 

[GRollins]

Jonathan,
I'd go tubes, but if you want to go solid state, you'll probably need to use MOSFETs in the output stages. I know Motorola (On Semi, now) has MOSFETs up to 1200V, perhaps higher. I can't see why IRF et. al. wouldn't have similar parts.
I'm not familiar with any bipolars with voltage ratings that high. Geoff is more familiar with bipolars, he may know of something along those lines, as that's more his area of expertise.

Grey

 

[Geoff]

Jonathan

There are BJT devices available with the required ratings (eg ONSemi's MJL16218 with a Vce of 1500V, Ic of 15A and a maximum power of 170W). However, these transistors have been developed for use in high voltage applications such as horizontal deflection circuits and have a very low gain which could make them unsuitable for use in a linear audio amp. From my limited experience of high voltage transistors, only NPN types are available so a complementary output stage could not be used.

This is one of the applications where I think MOSFETs would be more suitable.

Geoff

 

[Nelson Pass]

Tubes cascoding Mosfets make for a very workable
approach. The Mosfet gives you high sensitivity
and can be used Son of Zen style with the tubes
shielding it from the high voltage.

I did this successfully driving the top end of a
set of Hill Plasmatronics, and it worked splendidly.
Of course in this case we had the plasma arc itself
as the load.

For an electrostatic speaker, the best load would
probably be a center tapped inductor, just like the
transformer secondary which came with your standard
electrostatic loudspeaker. If you use one of those,
you can take feedback off what used to be the primary...

 

[GRollins]

Geoff,
The ON Semi site shows that device as "Last Shipments." Are there other suppliers?
Jonathan,
You're not going to need a huge amount of current to drive an electrostat; no need for massive copper buss bars this time. It's volts you'll be wanting in bushel quantities.
I once rigged a direct drive electrostat off an old Voice of Music tube integrated amp (6BQ5 output tubes, I think). The rail was about 250-275V, so the voltage swing wasn't all that large (not in electrostat terms, anyway), but I got a decent amount of volume out of a mystery electrostat panel I had lying about, and a homemade panel that I tossed together. I didn't take notes on what I did, as I was just playing.
My point here being that if I could get so-so volume from a 275V rail, I'd think you could do quite nicely off a higher rail. The final result will depend on the DC bias voltage you feed the electrostat, and the spacing of the elements in the driver.
Acoustat had a direct-drive (tube) model called the X once upon a time. I remember that they had a reputation for blowing up, but I don't remember the failure mode. They used some unusual tube for the output--can't remember which one. I do know that they were capable of some pretty good sound when they were working properly.
As for doing one purely solid state, you could take Nelson's cascode suggestion and stack two 1200V devices with a 1-1.5kV rail, just to have some elbow room. Or, of course, you could strap a tube on top. I kinda like the tube idea (but you knew I'd say that...).

Grey

 

[G-Daddy]

Alright, finals are over, back to the real work . . .

My knowledge of amplifier design is pretty much all theoretical, and most of it I pulled from Randy Slone's 'High Power Audio Amplifier Construction Manual'. But of course, I've made sure i look at other points of view, there's obviously many approaches to this sort of thing. The Pass and Leach topographies I've found particularly instructive, giving a healthy sense of keep-it-simple.

Now if I were to try to build an output stage capable of, say, 1kv swings (my homebuilt ESLs are, sadly, pretty inefficient), and I wanted to stay away from tube stages or hybrid designs, how would this design differ from an averge solid state amplifier? Would the output stage run on a seperate, high-voltage rail? Or would all the stages (say, of a Lin 3-stage schematic) have to sit at +/- 1kv? How could that be workable? Wouldn't ALL my transistors then have to be rated for high voltage? Is it completely irrational to insist on a solid-state scheme here?

On the ESL front, my friend and I've steeped ourself thoroughly in the theory and common knowledge of care and feeding of an electrostatic speaker. So I can see that i really dont need much current (except when the impedance drops to some god awefully low value at 40khz). But of course, for direct drive i need voltage, and lots of it. What I'm looking for now is a general sort of idea of the best way to go about forming a general schematic for this amp. What sort of new concepts and practices should I look up and understand before I can design this thing? Thanks for everyone's attention and help. Merry christmas from the deep South.
- Jonathan

 

[G-Daddy]

Alright, after some reading, I think my previous post is a little simplistic. Since the middle stage (voltage-gain stage) is where all the voltage develops, its pretty obvious that it, too, will have to have 1000 volt rail supply. So any sort of dual rail design from that aspect doesnt make much sense. But can the input stage use a more modest voltage supply? As I see it, no, it too also needs the same supply that the other stages get. Help me out of im not seeing this correctly . . .

So then my previous question stands, aren't i going to need high voltage components for EVERY component in the amp?

Also, since I'm not going to need much in the way of current, doesn't this mean that my output stage is going to be pretty tiny? Am I wrong is saying that in this case, while it will add some current gain, it's basically just there to interface the voltage-gain stage with the load?

I understand the basic building blocks (current mirrors, filters, etc.) but I'm having some trouble seeing the big picture here. Any help anyone can offer on this application would be most appreciated. And of course, thanks to everyone who's already stepped up.
- Jonathan