Electronics > Projects, Designs, and Technical Stuff

Achieving very low output voltage ripple in HV DC/DC converter

(1/2) > >>

JimmyCad:
I am designing an SMPS, operating at high voltage with 2 outputs of 5kV and 2.5kV, output currents are 8mA and 160mA respectively. I need to achieve an attenuation on the final output for maximum output ripple of 50mV.

I have came across the following by Jim Williams, and I would like to know whether it could be extended for use in higher voltage outputs. The push-pull converter has an output voltage of 300V only, but achieves an massive attenuation, with only 100uV noise. However I don't think the chip is appropriate since my primary side voltage is quite high at a nominal 270V.

https://www.edn.com/design/analog/4326853/High-voltage-low-noise-dc-dc-converters

I have a few further questions about the document:

It mentions using "controlled transition techniques" which I assume is the same as the "controlled switching edge times" that he refers to. What is this? Is it only relating to hard-switched converters?

Why does the "symmetrical transformer drive" of a push-pull converter make it much better than an alternative half bridge would perform, for example?

Any additional suggestions to research will be appreciated.  :-+

T3sl4co1l:
1. You can cascode transistors to get higher voltage ratings.  Mind that this gives voltage gain, so may tend to defeat the transition rate control, or introduce more chance for oscillation.  So allow for the possibility of needing little RCs or ferrite beads around the active transistors, to control this.

You still need an aux supply to run the controller, but that can be a tiny little AC-DC module, or maybe you already have it from what's generating your 270V source.

2. These are aimed at hard switching, yeah; I don't see why they wouldn't be applicable to resonant as well, but those are quieter to begin with so it may just not be as important.

Resonant converters also tend to have a harder time avoiding diode noise (thinking of the LLC network for example), because the diodes conduct somewhat less than a half cycle each, that is, they kind of just do their own thing at whatever rate of current and voltage they happen to.  (I have seen synchronous converters where the rect switches are operated at fixed duty cycle, which seemed to work out fine; controlled-transition could be used there to similar benefit.)  And with the primary side switching noise being somewhat isolated by the series inductance, this can be a relatively large concern.

3. Probably just because making a controlled-transition driver and floating it up in the air, isn't easy to do.

Would be nice to have though.  They make a version that's a controlled-rate driver with general purpose logic inputs I think, which would give an interesting opportunity to try.  Both for higher voltages and for bootstrap drive if you're so inclined.

Tim

Kleinstein:
One can use a capacitance multiplier with a emitter follower even with a lower voltage transistor. The transistor only sees the voltage drop used to regulated, so something like the worst case ripple at the input + 1 V or so. It still needs some protection for turn on transients, but this is mot so hard. So just for ripply reduction one can get way without cascaded transistors.

The controlled slow switching is kind of going away from hard switching a little, mainly to avoid the really high frequency noise part.
So intentionally switch slow and trade less EMI for high loss.

Marco:

--- Quote from: JimmyCad on August 16, 2019, 11:12:26 am ---The push-pull converter has an output voltage of 300V only, but achieves an massive attenuation
--- End quote ---

The converter causes the ripple, the C-R-C output filter causes the attenuation.

If you want to keep the final output cap at say 1 Joule for safety (160 mA will still kill you eventually of course, but at 1 joule at least the discharge won't kill you immediately) the capacitance for 2.5 kV is around 330 nF, while with a 10 Watt resistor you could only get about 330 Ohm of resistance in the filter. So passive filtering won't be quite as successful for you, nor will the ripple of the converter before filtering be as low given they are measuring at 60 mW and you want 400W.

You probably want an active filter/regulator after the switching one.

T3sl4co1l:
Incidentally, LC filtering isn't very practical, because the inductor needs to be very large, and then won't perform well at high frequencies.  The problem is the high impedance, say for 4kV and 100mA that's 40kohms average load, and so maybe you want the filter to have a characteristic impedance of a couple kohms.  Which means if you have 0.1uF in there, you need Z^2*C = 100mH too, which is, good luck with that.  It also needs to be rated for the full voltage, in case of short circuit causing full voltage to drop across it.

So, RC filtering is about what you can do, and small LC filtering which is effective for EMI/RFI along the wires (which themselves won't be anywhere near 40kohms to free space, more like 100s ohms).

Tim

Navigation

[0] Message Index

[#] Next page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod