Removing output voltage ripple in high voltage outputs

Question

I am looking to reduce the voltage ripple on a DC/DC converter operating at 5kV.

I have looked extensively on the internet and have found multiple ways to reduce output voltage ripple, an interesting reference is by EEVBlog which uses capacitance multipliers with BJT/MOSFET/Darlington transistors.

Multiple other sources state to use multi-stage LC filters, and high value capacitors, etc. However, when the output voltage enters the kV range, the use of inductors is avoided as much as possible, if possible. Furthermore the size of capacitors with large values rated for kV values are very bulky and again, would be avoided if possible. This brings limitations in the amount of voltage ripple that can be attenuated in this range - which makes designing low output voltage ripple DC/DC converters in ripple sensitive applications (such as travelling wave tubes) very difficult. Splitting a transformer secondary into multiple windings reduces the ratings of inductors/capacitors but to what extent does this beat having a single inductor?

Does anyone have any tips on how the voltage ripple can be minimzed in such cases?

I am planning on using HV rated silicon carbide devices in the isolated power supply so possibly something that utilizes a MOSFET in the rectifier can reduce the ripple.

Answer 1

You should consider higher frequency switching operations and a means to minimize filtering needs. Look into interleaving options (running n converters shifted by 360/n electrical degrees for their PWM periods) as well - this will have a similar effect as a single converter running at n*f frequency and can provide a modular/scalable solution. Rob B.'s suggestion was for a 2x interleaved system, but it's extendable to many stages/systems.

Depending on the application... an important point to consider for high-voltage supplies is their performance under arcing conditions. You may particularly want to minimize filter energy for this purpose as much as size/cost/weight reasons.

Answer 2

There are 100 V 5 W Zener diodes on market. You can construct a voltage limiter circuit. All you need to do is building the limiter circuit with less voltage level. For example consider a 5 kV DC with +250 V peak to peak ripples, a voltage limiter with 4.9 kV will kill all the ripples. This solution is likely cost you more because larger diodes that on market is around 15$.

Another solution is dividing the output voltage and filtering them seperately with simple RC networks. But it is a trivial solution. Because this method will likely increase the power loss.

Answer 3

You are concerned about bulky capacitors but can trade a little loss for the benefit of lower ripple, so have you considered using a MOSFET gate follower after the RC filter? A capacitance multiplier?

The capacitance at output will be amplifier gain x C4. You'll need to have a steady load current and good heat dissipation, but you already said you wanted a MOSFET solution.

Answer 4

You did not state the power or current output nor the frequency so it is difficult to understand best approach. However, one approach is to have 2 converter stages operating 180 degrees out of phase to help cancel the ripple.
If you can't use an inductor in your output how did generate the 5kV - without an inductor? A common output filter is an LCL filter with a resistor in series with the capacitor. The capacitors at that voltage are larger but you should not need a very large value. The resistor is for damping. Another approach is to consider a feed forward control.