High Voltage Waveform Controller

Dr. Bull, PhD (Press here Dummy)

Hi,

I'm trying to design a voltage controller to impress an arbitrary, predetermined and programmed high voltage waveform, ~3000volts with, say, 50 to 100 nanosecond resolution, into an aqueous sample with a MOSFET switch. I was hoping that a microprocessor controlled PWM could provide me with the switching profiles, frequency and duty cycle, necessary. I'm currently considering a controlled feedback loop to increase the PWM / MOSFET duty cycle when the voltage is low and reduce the switching when the voltage starts to overshoot.

I had been looking at the BASICStamp, but now I'm seeing that it implements the PWM function differently than what I would have anticipated. Perhaps the answer could be BASIC Stamp with digital PWM circuitry added on, etc.


Is anyone here familiar with an existing solution? Suggestions?


Appreciate it.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
The sum total of my electrical engineering is a passing familiarity with the fire extinguisher codes.

Carl Hayes

Me, I'd probably design something a lot like a voltage regulator, or perhaps a modulator, for a broadcast transmitter, at those voltages. Do FETs really go so high these days?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
· -- Carl, nn5i@arrl.net

Dr. Bull, PhD (Press here Dummy)

While I haven't tried it yet, I'm told that I can put two 1500V units in a series arrangement to switch 3kv.


Is this not possible?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
The sum total of my electrical engineering is a passing familiarity with the fire extinguisher codes.

Carl Hayes

Should be OK.· Use parallel resistors (parallel to the FETs, that is, but in series with each other) of appropriate values and ratings to make sure the drops across the FETs stay equal.· Else the one with the greater voltage may fail.· Lessee -- say 5 megohms each, 2 watts, rated for high voltage.

It'll be a little hard to feed the gates, since they too will be 1500 volts apart.· And you'd better be very sure they switch simultaneously, or one will blow.· This problem is magnified greatly if it's an analog amplifier instead of a switch.

I wish I could watch, from a safe distance of course.· Things like that never work the first time.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
· -- Carl, nn5i@arrl.net

Phil Pilgrim (PhiPi)

I wonder if a vacuum tube, like a 4CX1000A, might be a better choice here than MOSFETs. You'd still have to drive the screen at about 325V, but that's more approachable with solid state switches than 1500V. Also, I think I'd filter any PWM to analog before the final stage. Otherwise, you'll be generating splatter on every radio receiver in the county.

Finally, electrical isolation between your PWM logic and the HV side is also a good idea. Optos will be too slow, but there are digital isolators capable of 90Mbs and 5000V isolation. Analog Devices' ADUM2400CRWZ is an example.

-Phil

Carl Hayes

Phil Pilgrim (PhiPi) said...
I wonder if a vacuum tube, like a 4CX1000A, might be a better choice here than MOSFETs. You'd still have to drive the screen at about 325V, but that's more approachable with solid state switches than 1500V. Also, I think I'd filter any PWM to analog before the final stage. Otherwise, you'll be generating splatter on every radio receiver in the county.

Finally, electrical isolation between your PWM logic and the HV side is also a good idea. Optos will be too slow, but there are digital isolators capable of 90Mbs and 5000V isolation. Analog Devices' ADUM2400CRWZ is an example.

-Phil

Phil, you left the R out of country.

A large transmitting tube was my first thought too.· 3Kv is the absolute maximum rating for a 4CX1000A, so I'd probably use something with a little bit of margin.· The 4CX3000A has a maximum rated plate voltage of 7KV in Class AB₁ so it would probably be OK.· So would a 4-400A, which can handle 4Kv.

Somehow I doubt that kind of power-handling ability is required, though.· There may be other types, perhaps used in oscilloscope deflection amplifiers, that can handle high voltages at modest power.· I have no data sheets for those, though.

I note the requirement for 50 to 100 nanosecond response -- that is, 10 to 20 MHz response.· Will any microcontroller handle that requirement?· Certainly not any I've worked with, which however constitute a limited set.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
· -- Carl, nn5i@arrl.net

Phil Pilgrim (PhiPi)

Carl Hayes said...
Phil, you left the R out of country.

LOL! True enough!

-Phil

Geekgirl

How about just trying to control the lower voltage input side of a very serious pulse transformer?

Parts cost would be in the hundreds of dollars instead of the thousands of dollars for the exotic power amp tubes and coolers.