Well, you need to drive the ESL with differential voltages anyway, so you can make a fully differential MOSFET amplifier - single ended class A at that (you want to avoid referencing the signal to the + power rail if possible). A differential amp with an 800V power rail could swing >700V on each output, and twice that between outputs. You would need >800V MOSFETs but these are actually quite common. The best bet would be to look for fairly low current MOSFETs as these will have low input and reverse transfer capacitance. This is especially important as an ESL directly driven does not have a very large capacitance, and this is the only thing that limits the amps HF response. If that sort of output swing works for you, MOSFETs are a good bet. If not, you will have to go with tubes.
If you do use MOSFETs, the real problem is making current sources from the + rail to the MOSFET drains - the problem is supplying the required voltage reference. P channel MOSFETs capable of withstanding 800V or more are NOT common, so you end up with floating voltage references, which need to swing along with the output signal - the biggest problkem here being that various passive networks tend to consume lots of current and, you get capacitance from the said reference voltage sources to everything else.
A friend of mine that tried this approach back in 1992, and here is what he used:
Each leg of the differential output was a N-channel MOSFET current source that had a well insulated 9V battery as a reference voltage source for the Vgs of the MOSFET. The current source was the top of a MOSFET + BJT cascode, with the BJTs configured as a LTP. The LTP was balanced via a dual OP-amp DC servo (FET OPs and some very large resistive dividers!). This drove a 4" x 6" experimental ESL midtweeter cell with about 1.5mm interelectrode distance (i.e. standard PCB thickness) and a polarising voltage of about 8-12kV - it got incredibly loud! Transistors were BUZ80 if I recall right, +rail was 760V. In 1992 this was the only MOSFET one could find that could work for this application. Unfortunately, I completely lost contact with this guy... I wonder what the standing current was...
If you do use MOSFETs, the real problem is making current sources from the + rail to the MOSFET drains - the problem is supplying the required voltage reference. P channel MOSFETs capable of withstanding 800V or more are NOT common, so you end up with floating voltage references, which need to swing along with the output signal - the biggest problkem here being that various passive networks tend to consume lots of current and, you get capacitance from the said reference voltage sources to everything else.
A friend of mine that tried this approach back in 1992, and here is what he used:
Each leg of the differential output was a N-channel MOSFET current source that had a well insulated 9V battery as a reference voltage source for the Vgs of the MOSFET. The current source was the top of a MOSFET + BJT cascode, with the BJTs configured as a LTP. The LTP was balanced via a dual OP-amp DC servo (FET OPs and some very large resistive dividers!). This drove a 4" x 6" experimental ESL midtweeter cell with about 1.5mm interelectrode distance (i.e. standard PCB thickness) and a polarising voltage of about 8-12kV - it got incredibly loud! Transistors were BUZ80 if I recall right, +rail was 760V. In 1992 this was the only MOSFET one could find that could work for this application. Unfortunately, I completely lost contact with this guy... I wonder what the standing current was...
Audio Amateur had a construction project for a direct coupled ES amplifier using a transmitting triode.
Back issues may be ordered from AudioXpress.com.
An 811 is a currently made $24 transmitting triode that has a cut-off of 2KV and is normally run about 1.2KV or so. I've seen NOS RCA 811s for as little as $10. A lot more rugged than any transistor, these would be my choice.
If you want to play with FETs I would use the 811 as a cascode to handle the power.
Supertex makes high current/high voltage depletion mode FETs in a TO220 case.
http://www.supertex.com/products/selector_guides/102
Back issues may be ordered from AudioXpress.com.
An 811 is a currently made $24 transmitting triode that has a cut-off of 2KV and is normally run about 1.2KV or so. I've seen NOS RCA 811s for as little as $10. A lot more rugged than any transistor, these would be my choice.
If you want to play with FETs I would use the 811 as a cascode to handle the power.
Supertex makes high current/high voltage depletion mode FETs in a TO220 case.
http://www.supertex.com/products/selector_guides/102
Depends on what kind of ESL you are planning to drive. If it's a full range, 1.5kVpp will not be nearly enough. But for a stat working upwards of 300Hz or so, electrode distances and polarisation as well as drive voltage goes down drastically (voltages rise with the square of electrode distance).
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.