Posted by CV on November 11, 2003 at 04:34:12:
In Reply to: Re: ESL Direct Drive Amplifiers......? posted by Brian Walsh on November 09, 2003 at 09:33:55:
Allo fellas,
Subject close to my heart this one. I just typed a long reply out and the browser lost it, so please excuse the terseness...
The problem ain't so much finding tubes that are happy with the voltages required (6HS5 beam triodes can be found for a few bucks each), it's getting the current required.
Sums: take a panel capacitance of say 2nF. This is 4000 ohms at 20kHz.
Which means that if you want to swing 4kV peak at this frequency cleanly, you need one amp of current.
If this was an SE amp (or indeed class A), you'd need a dissipation of 4kW *minimum* in the tube, and plenty of feedback to reduce distortion as the elliptical loadline gets you at low currents.
Now, obviously real music doesn't consist of full power sines at 20k, but if you look at transients on a scope, it can be quite sobering. I suspect this is why people claim ESLs soften leading edges and lack dynamics - they're not hearing the driven properly.
Despite this, I've had promising results driving a martin logan CLS panel with a SE 4212 triode (1500V) via an amorphous-core phase splitter. At low levels, sounded great, any volume or complexity to the music, forget it.
I plan to up the volts to 3.5kV using a different tube and run the panels as mids only, crossing over to a highly capable ribbon tweeter at 1.5kHz.
This may still not be clean enough. I suspect the way to go is differential, like this:
http://www.ultranalog.com/9912/p25/
By running the panels as mids only (treble issues include slewing and roll-off, and cutting the bass reduces overall power demands), it may be feasibe to use an air-core choke at the output.
Anyway, as for swinging 10kV into Soundlabs full-range *cleanly*, I suspect you'll be S-O-O-L [hint: O-O = "out of"]. Unless of course, you fancy using those air-cooled jumbo ceramic valves, feedback and enormous power supplies and kilowatts of power just to power the valve filaments...
Hint: amorphous core transformers can sound extremely transparent...