Adjustable High voltage power supply (2KV, Iout<2 mA)

Hi there,

In my new project, I need to design and built a digitally programmable power supply. The specs are as below:

* Vout max = 2 KV
* I out max < 2 mA (1 mA is OK)
* Resolution: 10 Volt
* The output should be fast enough to change within few hundred milliseconds.

The linear approach (Bulky high voltage transformers etc) seems to be not applicable here.

Any suggestion is appreciated.

Resonant flyback on an ETD29 transformer 5 o/p windings with 400VDc per winding in series, 800V diodes, run from a 24V primary with a current limit to limit power to the output, simple PWM control of mosfet gate at 8kHz to give audio indication, and low diode losses...

Orson Cart said:

Resonant flyback on an ETD29 transformer 5 o/p windings with 400VDc per winding in series, 800V diodes, run from a 24V primary with a current limit to limit power to the output, simple PWM control of mosfet gate at 8kHz to give audio indication, and low diode losses...

Click to expand...

Thanks Orson for your prompt response.

Could you please explain a little bit more about what you've suggested? I'm new in SMPS and all my previous experiences were in linear power supplies. Do you know useful articles/pages (including schematics) on the internet for better understanding? I googled "EDT29" and didn't find high voltage ones. Also, Digi-key doesn't show any result for this transformer.

Thanks.

The "ETD29" is just a reference to the type of transformer that he was suggesting. "ETD" is a reference for the style or shape of the ferrite core of the transformer. The "29" part is a reference to the size of the "ETD" style core as these cores are available in many standard sizes. I think what Orson was suggesting was that you start with an ETD29 ferrite core, and the appropriate former, and wind a custom transformer that you could then use in a resonant flyback topology. He was also suggesting that the transformer be wound with 5 secondary windings, each producing 400V, which could be wired in series to reach your 2000V goal. With this 5-secondary, series-wired setup, you could have a separate voltage tap for different out voltage ranges. I've attached a simple drawing of what Orson was describing. Another plus to the multiple-secondary design is that the diodes for each secondary output only needs to carry the voltage of that winding, if I'm understanding the operation properly.

The multi-winding transformer has to be custom made in any case. Producing the 5 secondary windings with proper insulation is a bit demanding.

Alternatively you can go for a voltage mutiplying rectifier, but it involves energy storing capacitors and has more problems to achieve fast voltage variations. As an advantage, you can use an off-the-shelf SMPS transformer in reversed configuration.

Even for the suggested multi-winding "diode split" configuration, you have a trade-off between fast voltage variation capability and output ripple. You didn't yet specify output voltage quality, but any output filter capacitors will be only discharged with a load current. So you either need to place a permanent bleeder resistor or an active discharge on demand.

Perhaps the high voltage could be generated with a TV flyback transformer modified to give a 2kV output.

Here's an example circuit for powering such a transformer to get a high voltage output. Additional primary turns would reduce the output voltage (keeping the two primary turns ratio the same). You would need to experimentally determine the optimum number of turns.

Changing the transistor supply voltage would allow regulation of the output voltage.

What is the minimum output voltage needed?

In the multi-stage flyback above, you need to have a cap to soak up the pulse current at every 400v o/p, even 47nF 630V would do.... else the effective leakage gets a bit high and increases the spike volts on the pri side switch...