high voltage switching

[freeman2]

I need to switch a voltage of 1500v to 0v (ht) at a frequency of 1 to 2khz without loading the supply to much. No current needs to flow into the load just the voltage appears across it. Anyone got a schematic that will do the job?????

 

[em2006]

...No current needs to flow into the load just the voltage appears across it...

If a voltage appears across a load, then inevitably a current will flow into the load.

Added after 12 minutes:

I need to switch a voltage of 1500v to 0v (ht) at a frequency of 1 to 2khz ...

So you want a square wave generator that has 1-2KHz frequency and 1500V amplitude ?

 

[freeman2]

yep the current will flow into the load, but it is not required and a short pulse will reduce this.

The problem is the volt drop from the loading on the ht supply, it seems to drag it down and the frequency of switching does not give it time to recover.

Any idea of a way round this?

 

[FvM]

You didn't tell a word about the load impedance. It's impossible to answer the question without knowing it.
You also should have an idea of the intended switching time.

 

[freeman2]

good point and thanks for the reply, the impedance is low, under 5k and the pulse width can be in the nano seconds to milliseconds
all depending on what i can build and also i don`t want current flow into the load so the pulse if to long has to be cut short to reduce this.
If using a igbt it will have to be microS and nano i would think avalanche transistors,line transistors should have a wider range. But it is
experimental and any ball park design is very helpfull, the best pulse width design to start with is in the microseconds. The load is capacitive.

Working with high voltages sucks when you are used to low voltage design.But it is for a good cause.

 

[FvM]

If the "under 5k" load is resistive, the load current will flow from the first nansecond, consuming > 225 W power. So it's not actually clear to me what you want to achieve.

In addtion, you have to expect a certain switching node capacitance that has to be charged even with no load at the output.

 

[freeman2]

Its EHT not ht supply, i have made a mistake. So the question is can i get high frequency transistor switching without the loading on the supply killing the voltage?

 

[Mr.Cool]

let's say we're talking about an IGBT. you can switch 1500V drain to source with minimal current flow. but to do this in NANOseconds will be rather difficult. look at the datasheet of the switch, it has a certain time delay and rise time (and same for switch OFF). there are very few (any?) high voltage devices that can switch that fast realistically. i have designed 600V switching in 150 nano seconds, and that was with high end equipment. the problem is the parasitic capacitance in the circuit, specifically, between the high voltage switch (IGBT drain-source) and the gate driver circuit which is providing the low voltage control signal (IGBT gate-source). and then the coupling that exists between this gate driver and the digital source (i.e. other side of the opto coupler). you need a killer interface there.

some other things to think about.. are you go up in applied voltage your derating factor should go up too. i.e. you can switch 40V with a 50V rated switch. you can switch 600V with a 1200V rated switch. see the ratio increase? now you want to switch 1500V and you want to do this FAST (nano seconds!?) what switch will do this and what will it's voltage rating be?

if you are accepting of using a GTO as your switch then this is achievable. if you want small current, you are probably better off with two 1200V rated IGBTs in cascade. or 1200V rated mosfet if the current is low. but know that there are more risks with cascade topology causee you must ensure a good switch voltage balance.

good luck

FvM - wow, you seriously have 27,000 points now? helped over 1000 people? nice!

Mr.Cool

 

[freeman2]

Thanks mr cool you are ace......hang on wait a minute, that was nonesense, avoiding the point that was made and padded with waffle....nooooooo you are really mr tony blair. Oh tony with all the wars in the middle east and the wars to come you really don`t have time for messing about, start doing the job your over paid for..PLEASE.

Any sensible answers? even if it is i don`t think it is possible

Added after 10 minutes:

Thanks mr cool you are ace......hang on wait a minute, that was nonesense, avoiding the point that was made and padded with waffle....nooooooo you are really mr tony blair. Oh tony with all the wars in the middle east and the wars to come you really don`t have time for messing about, start doing the job your over paid for..PLEASE.

Any sensible answers? even if it is i don`t think it is possible

 

[FvM]

Unfortunately the last "clarification" is rather confusing the problem in my opinion. A basic schematic including the load and decription of the intended waveform can possibly help.

I'm trying to translate the question in a real world problem, but I don't see a correspondence yet.

 

[karesz]

Hi all,
If 1500V & high-speed switching is needed, is it not a typical/good matter for a Thyratron pls?
K.

 

[uoficowboy]

I have done similar things with high voltage MOSFETs and FET drivers. It sounds like you need to push/pull your capacitive load, so you'll need a pull-up on the drain of your N-FET. This means that you will dissipate some power. Or you can push/pull the load with a HV FET driver. Silicon Labs makes some that I believe would suit your application.

What is your load? Piezo device? Electro-rheological fluid?