High voltage SMPS with ultra-low ripple

[Enceladus]

Hello everyone! I would like to design a <100W switch mode power supply with ultra-low voltage ripple (2-5mV peak to peak). Usually it's not a big deal, there are plenty of circuits around, but this power supply also has to be a high voltage one (up to 400V). The input voltage is 25-28V. The output voltage will be fixed, the output currents are not high (<160mA) and the isolation of the output is not necessary (the ground will be the same). Ideally I would like to make it as efficient as possible so I can use passive cooling (heatsink).

So I wonder, what would be the best topology in this case? Flyback + LC filter? Boost + linear regulator? What PWM controller would you recommend?
Thank you!

[T3sl4co1l]

Flyback is the way to go, yes.

Start with a 240:24V flyback transformer, either pulled from equipment or purchased new.  Laptop power bricks are probably a good reference.  Just run it in reverse, driving the secondary with a higher current transistor and current shunt, and the same controller if it's not directly offline but can be aux-powered (e.g., UC3842).  Probably use a regulator to supply the controller, since 20V+ is kind of high otherwise.  A dumb zener + transistor regulator is fine, or a 78xx or whatever.  The aux winding on the transformer (if applicable) can be left open.

Keeping common ground will help greatly with keeping noise low.  Having the ground return path right there makes filtering much easier.  Don't neglect filtering the input or the output.

Tim

[hli]

4mV at 400V output voltage is 0.001%. It will be a stretch just to find a voltage reference accurate (and stable) enough for that. What are you building this for?

[exe]

4mV at 400V output voltage is 0.001%. It will be a stretch just to find a voltage reference accurate (and stable) enough for that. What are you building this for?

He was refering to ripple, not to DC accuracy.

[T3sl4co1l]

What is DC accuracy but lower frequency ripple?

Presumably, the intent is to keep high frequency noise within those limits (which will be beyond average, but not unreasonable to solve), and hopefully mains frequency ripple (assuming the 28V source has any, which it might not -- that makes the error amp less challenging to design).

But without a spec, we cannot make a meaningful judgement, no.

Tim

[Enceladus]

Thanks everyone!

I would like to make a HV power supply for a tube amplifier. For this application low voltage ripple is usually essential to get rid of "hum" (not sure if this term is applicable here as it is usually associated with 50 Hz noise). I've built a power supply using LT3751 https://www.analog.com/en/products/lt3751.html#product-overview controller and a GA3459 https://www.coilcraft.com/ga3459.cfm transformer. The schematic is attached. Initially I didn't put an LC filter at the output so I had to add it after that externally (L = 33 uH, C = 22 uF). It works OK, but still there are 80 mV ripple as well as some high frequency noise. I played with filters parameters a bit but it didn't help much.

PS
Also it looks like the transformer is not powerful enough for 60W output as it gets pretty hot (~75-80 C) on full load.

[Enceladus]

Similar to this one? http://www.clivemaxfield.com/area51/do-not-delete/max-4-john-01.jpg
I wonder if charge multipliers are capable of providing low ripple at 100W output power. Can you recommend any application notes on designing SMPS with a charge pump? Thank you!