1KV Adjustable power supply

[SSShocked]

Hi everybody and happy new year !
I need an adjustable power supply which goes from around 100V to 1KV at 500mA.
So, how do i achieve this ? I am going to use it as a laboratory power supply for vacuum tube electronics.
It needs to be reasonably simple (for more reliability) and cheap (I am in Bulgaria).
It must be powerful and won't be bad if it has adjustable current limiting.
I have a few ideas:
Use U-type core.
Secondary on one side, and primary on the other, but there are some other big problems, so look like i will use EE core and wind it with good quality materials (wire and insulation with good planning).
As i know, the U core wound in this way, has very high leakage inductance and capacitance. It is also not effective.

Auxilary winding for feedback with common mode choke.
What about a current transformer ?

Low sw. frequency- around 20KHz to allow low duty cycles.

Maybe SG3524 or TL494 ?

What about very low switching frequency- 400Hz ?
I can find a good enough core and the big size is not a huge problem.
I can switch it with thyristors.

Thank you in advance and sorry for my poor english.

Simeon Petkov

 

[dick_freebird]

If this is for crude bench work then you could be overthinking
it. Maybe just a Variac powering a 1:4 stepup mains xfmr,
bridge rect and filter, and twist the knob to get the desired
power level. 500W at 1kV is not that trivial.

For that matter the Variac could power (4) 1:1 isolation
xfmrs - these being more readily available - with primaries
in parallel and secondaries (w/ rect & local filter?) in
series.

You might also find cheap variable DC-DC bricks with
moderately high output voltage (e.g. 48V telecom) and
galvanic isolation (want > 2X your 1kV), again parallel
the front ends w/ common feedback and series stack
the output. On the surplus market these could be
pretty cheap (though you might search for a while).

 

[SSShocked]

Well, this is a good idea, but is does not suit me.
I need a laboratory power supply.
It must be regulated and it must keep it's max. current at lower output voltage.
Yes. I know it is not simple, but i will try.

My problem is the way how to control it.
The transformer is not a big problem.
I know many people who can wind me a premium quality one.
Just how to design the control ?

Thanks !

 

[FvM]

If you use a voltage-multiplying rectifier, you get rid of the high secondary winding voltage and can treat the design as a standard 500 W switch-mode supply and regular transformer construction. The voltage range suggests a design with good variation capabilities, e.g. a phase shifted bridge. But a standard push-pull inverter should work as well, e.g. using SG3524. Isolated feedback from the secondary is manadatory for a laboratory supply, I think.

 

[SSShocked]

That is a good idea :-D
Then, i can use a ready design and just add the multiplier.
Can i take the feedback from the low voltage ?
Is it going to be a problem when i connect the multiplier ?
I have a nice design for a 0-50V 0-10A lab SMPS, but it is half bridge.

 

[FvM]

For good regulation, there should be feedback from the HV-DC side.

 

[SSShocked]

This is a bit scary :-|
Any ideas how it can be done ?
High value resistors are not an option.
If something screws up, there will be quite a bang and high voltage when i don't need it (Murphy's law :-D).

 

[mtwieg]

If you use a voltage-multiplying rectifier, you get rid of the high secondary winding voltage and can treat the design as a standard 500 W switch-mode supply and regular transformer construction.

I doubt a voltage multiplier is feasible for 500W. The capacitors would have to be huge...

This is a bit scary :-|
Any ideas how it can be done ?
High value resistors are not an option.
If something screws up, there will be quite a bang and high voltage when i don't need it (Murphy's law :-D).

You're building a 1kV supply. There's no escaping some risk, and high value resistors are the least of your concerns.

A forward topology is probably sufficient, a half/full bridge topology would be better but is significantly more difficult to engineer.

 

[SSShocked]

Oh, i forgot about the caps...
Bad, bad, bad...

Yes. It is not easy. The problem with high value resistors is that they very ureliable. It happened to me a few times. It is very ugly when one fails.
Maybe half bridge. The forward has some other problems. I think, that half bridge is better.
I am going to rework the 50V design for high voltage, but how to get the damn feedback ?

 

[FvM]

I doubt a voltage multiplier is feasible for 500W. The capacitors would have to be huge...

We are talking about medium frequency switched mode power supply. Voltage multiplier is pretty standard for HV SMPS. Capacitor size won't be more than some 10 up to 100 nF, it can be put in series resonance with the transformer leakage inductance.

 

[mtwieg]

We are talking about medium frequency switched mode power supply. Voltage multiplier is pretty standard for HV SMPS. Capacitor size won't be more than some 10 up to 100 nF, it can be put in series resonance with the transformer leakage inductance.

I suppose it could work but then the issue is that you have to control output voltage by varying input voltage, rather than duty cycle (which means you would need two conversion stages), correct? I'm having a hard time visualizing a SMPS topology with a multiplier which doesn't immediately rectify the peak AC voltage from the transformer, making duty cycle modulation infeasible.

 

[SSShocked]

Yeah.
So, it will not be with multiplier.
I hope Goldsmith comes and helps about the feedback.

 

[mtwieg]

Voltage feedback should be taken directly from the output, like FvM said. Having high voltage doesn't make that infeasible, it just means you need a high voltage resistor, or several low voltage resistors. Current limiting should also be done on the output side, if you want it to be accurate.