High voltage/low current Power supply

[Red_Micro]

What could be cost-effective approaches for high voltage/low current power supply? Let's say you have 600V AC three phase input, and you want 5V/50mA, or 12V/100mA for a control circuitry? Parts at high voltages start to become expensive. Off the top of my head I would use a flyback converter with some 1500V rated transistor. Has anyone had this kind of experience?

[David Hess]

What is wrong with using a 50/60 Hz transformer?

[Red_Micro]

What is wrong with using a 50/60 Hz transformer?

The size. Usually these kind of high voltage/low current supplies are used in small metering accessories.

[ChristofferB]

I'm sorry but I'm not sure I get what you're asking. Do you want to make 3-phase 600V or do you want to make a higher voltage from the 600V?

What voltage do you actually need? For most HVDC applications I'd go with inverter+voltage multiplier.

[schmitt trigger]

My understanding is:
High voltage input , with a low voltage, low current output.

Am I correct?

[Red_Micro]

My understanding is:
High voltage input , with a low voltage, low current output.

Am I correct?

Yes. Sorry for the confusion. And when I say high voltage input, I refer to three phase systems 600V.

[schmitt trigger]

Back in the early 2000s I saw a SMPS designed to run from such a high voltage. It was complex, because one couldn't find many components with the required voltage ratings.

I don't think a flyback would be the best option, the reflected voltage and the leakage inductance spike could task even the highest rated Mosfets and diodes. Most likely a better solution will be a two-switch forward.

But I could be wrong, there may be newer devices and topologies to address specifically these scenarios.

Personally, since the power ratings are so low, I wouldn't immediately dismiss a custom 50/60 Hz transformer solution. I would at least investigate it.

[Zero999]

It's not something I'd build myself, if I could avoid it, but I couldn't find many PSUs rated to 600V input. 500V seems to be more common though.

I found this on Digi-Key which is rated to 528V.
https://www.digikey.com/product-detail/en/recom-power/RAC05-05SK-480/945-3185-ND/9356898

Do your really need 600V? 600VAC 3 phase is 346V phase to neutral, which makes things easier.

Here's a DIN rail mount power supply which will run off 600V 3 phase. It gives 24V out, which should be easy to convert to 12V or 5V.
https://www.digikey.com/product-detail/en/delta-electronics/DRP024V060W3BN/1145-1086-ND/4833495

[Kleinstein]

One could consider a capacitive dropper circuit to some 300 V and a more standard  SMPS from there. 12 V 100 mA is a little over 1 W, so some 4 mA at 300 V. This does not need an excessively large capacitor from the dropper circuit. It gets even better at lower power.

[David Hess]

Another option if you do not want to build anything is to drop the voltage with a transformer and apply it to a common 240VAC input power supply.

There are lots of good options now for high voltage switching transistors.  I agree with Schmitt Trigger about avoiding the flyback configuration; the added isolation requirements of the transformer will do nothing good for leakage inductance.  I might implement a high frequency inverter and then regulation in a second stage but it is certainly feasible now to do it all now with a standard topology adapted to higher input voltages.  Pick one which has the most relaxed requirements for switching transistor voltage rating.

[H713]

Capacitive dropping would also allow you to use a relatively small 60Hz transformer designed for a 240V input. Such transformers can be quite small.

[Zero999]

Capacitive dropping would also allow you to use a relatively small 60Hz transformer designed for a 240V input. Such transformers can be quite small.

Be careful. A mains transformer is inductive, especially when lightly loaded and connecting a capacitor in series with it won't necessarily drop the expected voltage. The wrong value will form a resonant circuit with the transformer, causing a high current to flow, a high voltage across the transformer, some sparks and smoke.

[Red_Micro]

I was also looking into off-line-switchers like this approach: ti.com/tool/PMP11302

[jbb]

So 600V L-L...
+20% for high line
*sqrt(2) for rectifier
= 1018V DC

This might be doable with a 1700V SiC (Silicon Carbide) MOFET flyback converter. Several companies make parts exactly for low power flyback work.  Despite the high peak voltage stress, flyback is a strong contender because it can be designed for really wide input voltage range.

I expect that someone, somewhere makes a suitable product already.

Options:

• Make your own: you'll learn a lot but it will take ages
• Can you go single phase to neutral?
• Maybe you can find a wide input range DC DC converter (they make them for PV equipment) and deploy your own rectifier?

[Red_Micro]

So 600V L-L...
+20% for high line
*sqrt(2) for rectifier
= 1018V DC

This might be doable with a 1700V SiC (Silicon Carbide) MOFET flyback converter. Several companies make parts exactly for low power flyback work.  Despite the high peak voltage stress, flyback is a strong contender because it can be designed for really wide input voltage range.

I expect that someone, somewhere makes a suitable product already.

Options:

• Make your own: you'll learn a lot but it will take ages
• Can you go single phase to neutral?
• Maybe you can find a wide input range DC DC converter (they make them for PV equipment) and deploy your own rectifier?

Flyback is a good approach for handling wide input voltage range. But the fact that I may not strictly need isolation, and considering the high price of SiC FETs, made me think about a buck converter like this TI App Note  (ti.com/tool/PMP11302). Unfortunately, I can't go single phase to neutral. Just Line to Line in this application.

And of course, this will be a good a experience where I expect to learn a lot.

[Neomys Sapiens]

Ok, 450V are not 600V and 20mA is just a fifth of 100mA, but those might have been exemplary values, so:

There is even a chip for that. Microchip LR84K will take up to 450V and provide you with a 20mA LV supply.

[Red_Micro]

Ok, 450V are not 600V and 20mA is just a fifth of 100mA, but those might have been exemplary values, so:

There is even a chip for that. Microchip LR84K will take up to 450V and provide you with a 20mA LV supply.

This one seems a good idea. I will have a look at it. Thanks!

[NiHaoMike]

The size. Usually these kind of high voltage/low current supplies are used in small metering accessories.

Since the power needed is small, use a capacitor in series with a SIDAC to get fast rise time pulses that can be converted using a much smaller transformer.

[jbb]

If looking at capacitive droppers in an industrial setting, beware large spikes passing straight through your dropper cap into the clamp circuitry.

On SiC MOSFET flyback:
- flyback converters work just fine without isolation
- SiC MOSFET is on 0 Volt potential, unlike buck converter
- SiC MOSFETs do cost more money, but may lower total system cost

[Zero999]

I was also looking into off-line-switchers like this approach: ti.com/tool/PMP11302

The general idea looks fine, as long as it doesn't need to be isolated.
www.ti.com/lit/df/tidrk45/tidrk45.pdf

The schematic I found on TI's website does contain an error. D3 the BY255GP-E3/54, is only rated for 1.3kV, yet it will see double the reverse peak mains voltage across it.  This will limit the absolute maximum RMS AC input voltage to 1300/(2√2) = 460V. Another diode could be added, but I'd just run the whole thing off a suitable bridge rectifier and it should be fine up to 600VAC.
http://www.farnell.com/datasheets/2046942.pdf