High voltage pulsing circuit

Thread Starter

adit_123

Joined Jul 12, 2010
7
Hi Folks,
Very interesting reading all about the fun n frolic u guys hav had with H.V. Unfortunately i've had to deal with for two years of my career. :) anyways my problem is a little more serious n many ways and techincal. I'm trying to design a H.V pulsing circuit which pulses at a pulse period of 20ms and width 200us. The pulse amplitude varies from 800 to 840V. The source comes from a High voltage DC-DC converter which can produce a maximum of 2kV, and maximum currrent of 0.75mA.
The problem is finding switching circuits/ components which work at HIGH VOLTAGE and at the same time LOW current. Any ideas...?? pls help..
:confused:
 

Dx3

Joined Jun 19, 2010
87
Once upon a time I bought a MOSFET that could stand 900 volts, but after that, the only thing I know of is vacuum tubes. Not much help here!
 

massive

Joined May 7, 2010
20
I do remember silicon chip sep 97 magazine had an article on ign that did some step up trick but might have also been close to the 900 V jobs ,cant remember.
 

timrobbins

Joined Aug 29, 2009
318
It would be good to clarify what your output profile is: you seem to indicate 800V for 19.9ms and 840V for 0.2ms repetitive ?? And what is the loading during that profile?

Can you describe the power source you have available - is it regulated 2kV and 750uA cv and cc characteristic?

Ciao, Tim
 

SgtWookie

Joined Jul 17, 2007
22,230
Digikey has at least a couple of MOSFETs available that will handle that voltage and current.

http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=497-7518-5-ND
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=497-7967-1-ND
I chose those using the parametric search function to have a Vdss of at least 1kv, at least double your drain current requirement, and a low gate charge for fast switching.

The length of wiring between the drain and load should be very short to keep parasitic inductance to a minimum. If the load is reactive (ie: capacitive or inductive) you will likely blow the MOSFET quite quickly.
 

Thread Starter

adit_123

Joined Jul 12, 2010
7
It would be good to clarify what your output profile is: you seem to indicate 800V for 19.9ms and 840V for 0.2ms repetitive ?? And what is the loading during that profile?

Can you describe the power source you have available - is it regulated 2kV and 750uA cv and cc characteristic?

Ciao, Tim
Yes, the o/p profile is repetitive, 800V for 19.9ms and 840 4 0.2ms rectangular pulse. The supply is a EMCO DC-DC converter and the load is resistive, in the order of few 10's Meg ohms.
So now you see my problem, to make a mosfet/ transistor SWITCH with such a small load and low current (Biasing issues). Also there has to be an option for switching the polarity across load..
 

Thread Starter

adit_123

Joined Jul 12, 2010
7
Pulsarjpg.JPG

This is the circuit diagram i'm trying to use for the same. Please note the current flowing through load is not important, only voltage at O/p
 

timrobbins

Joined Aug 29, 2009
318
Can you clarify how you get 800V and 840V. Are those voltage levels generated by the same dc/dc? Maybe if you a simple sketch.
 

Thread Starter

adit_123

Joined Jul 12, 2010
7
Can you clarify how you get 800V and 840V. Are those voltage levels generated by the same dc/dc? Maybe if you a simple sketch.
Yes, tim. If you look at the circuit diagram shown above carefully, there is a 1000V source (DC-DC). Using a potential divider I now have 40V. Using this 40V I now generate a spike, which in-turn triggers the transitor to turn on and turn off.
This pulse is later superimposed on the 800V (cupling capacitors). Thefore I now have a 800 and 840V pulse.
The diodes are purely for rectification.
If my explanation doesnot suffice I shal post a step by step explanation.
 

timrobbins

Joined Aug 29, 2009
318
Maybe if you were to present the circuit in a more modular sequential format, with a simplified equivalent circuit beside it, and some supporting description, then viewers may find it less tortuous to appreciate.

Can you set the dc-dc up to provide 840V - and use that as one of your voltage levels?
 

marshallf3

Joined Jul 26, 2010
2,358
Also, pain in the biasing rear but could you use the MOSFETs in series with resistors &/or caps to equalize the potential across the junctions? That way they could share the voltage and using two 600V ones in series wouldnt press the edge of service as much as a single 1,000V one.
 

Thread Starter

adit_123

Joined Jul 12, 2010
7
Maybe if you were to present the circuit in a more modular sequential format, with a simplified equivalent circuit beside it, and some supporting description, then viewers may find it less tortuous to appreciate.

Can you set the dc-dc up to provide 840V - and use that as one of your voltage levels?
Yeah tim.. maybe this would help you, this block level diagram is self explanatory. So basically the 840 V O/p is open but the resistance connected to 800V is in the order 8Mohm.
Hope this suffices the need to understand. the application is mass spectroscopy, hence the potentials are merely to provide a uniform electric field through a drift region.
BLK_DIAG.JPG
 

timrobbins

Joined Aug 29, 2009
318
Possibly set up two regulated supplies at 800V and 840V. Derive 840V as a 160V zener below 1kV. Derive 800V as a 40V zener below 840V - with resistor to 0V. I'm assuming the load voltage has to be referenced to 0V, and can't be referenced to 1kV level.

Ill think about simplest switch.

Tim
 

timrobbins

Joined Aug 29, 2009
318
Back again.

I suggest you connect your circuit to the 800V point of the zener-zener-resistor load circuit, and then use an optomos to short the 40V zener to change the voltage seen by the load from 800V to 840V.

To apply the voltage just switch the 12V supply to the dc/dc.

But be carefull measuring the voltage - as the meter will load the circuit - unless of course you keep the meter connected at all times.

Your source has 750uA capacity, so you should be able to reasonable zener regulation by tweaking the zeners to get the desired volts and still be within the 750uA limit. The zeners can be made of multiple series parts - but they will have significant tempco (unless you want to connect a large numbers of 5V1 in series!).

Ciao, Tim
 

Thread Starter

adit_123

Joined Jul 12, 2010
7
Back again.

I suggest you connect your circuit to the 800V point of the zener-zener-resistor load circuit, and then use an optomos to short the 40V zener to change the voltage seen by the load from 800V to 840V.

To apply the voltage just switch the 12V supply to the dc/dc.

But be carefull measuring the voltage - as the meter will load the circuit - unless of course you keep the meter connected at all times.

Your source has 750uA capacity, so you should be able to reasonable zener regulation by tweaking the zeners to get the desired volts and still be within the 750uA limit. The zeners can be made of multiple series parts - but they will have significant tempco (unless you want to connect a large numbers of 5V1 in series!).

Ciao, Tim

Thanks Tim, very innovative idea. I appriciate your creativity. Although it wouldn't help me much beacause u need a circuit whose swicthing time is in the order of nano sec not milli sec. So a circuit which is edged triggered not level triggerred. Henc i moved away from my previous design of optocouplers.
But i do appriciate the help. Thanks a ton. FYI i have solved the problem. there was a mistake in the polarity of the transistor used. The circuit is now working completely as sexpected and has a rise time/ falltime of 40ns.

Thanks,
Aditya
 

Thread Starter

adit_123

Joined Jul 12, 2010
7
Back again.

I suggest you connect your circuit to the 800V point of the zener-zener-resistor load circuit, and then use an optomos to short the 40V zener to change the voltage seen by the load from 800V to 840V.

To apply the voltage just switch the 12V supply to the dc/dc.

But be carefull measuring the voltage - as the meter will load the circuit - unless of course you keep the meter connected at all times.

Your source has 750uA capacity, so you should be able to reasonable zener regulation by tweaking the zeners to get the desired volts and still be within the 750uA limit. The zeners can be made of multiple series parts - but they will have significant tempco (unless you want to connect a large numbers of 5V1 in series!).

Ciao, Tim

Thanks Tim, very innovative idea. I appreciate your creativity. Although it wouldn't help me much because u need a circuit whose switching time is in the order of nano sec not milli sec. So a circuit which is edged triggered not level triggered. Hence i moved away from my previous design of optocouplers.
But i do appreciate the help. Thanks a ton. FYI i have solved the problem. there was a mistake in the polarity of the transistor used. The circuit is now working completely as expected and has a rise time/ falltime of 40ns.

Thanks,
Aditya
 

timrobbins

Joined Aug 29, 2009
318
Optomos will give you 40ns off time and ~150ns on time, as measured from the trigger edge - the actual transition times will be less (for the AQW21x series I'm used to).

The limited supply current, circuit resistances, and parasitic capacitances will modify the actual voltage waveforms achieved.
 
Top