Improving the standard flyback converter for high voltage applications

Thread Starter

SiCEngineer

Joined May 22, 2019
238
Hello all,

I often see the flyback is used in high voltage, low current applications such as for CRT monitors and the like. I was wondering, what are the general improvements that can be made to a flyback converter to make it better suitable for high voltages, but at more significant currents? As a reference, my application requires around 200mA.

I see some variations such as the two-switch flyback converter (I like this one!), the active clamp flyback converter, etc, all of which can achieve very high power densities, especially if using new semiconductor devices.

One of the main drawbacks of flyback is that energy is stored in the transformer, and therefore must be sized for this. Whereas in other topologies since there is theoretically no actual storage within the transformer, it can be smaller and more compact. I have came across this patent https://patents.google.com/patent/US5796595 which is an interleaved, active clamped flyback converter. I assume (?) that the reason for this is that now energy transfer to the secondary side occurs at twice the rate it did originally, due to the 180 degree phase shift between the two flyback switches. But I don't see this being appropriate for HV, since the outputs are connected in parallel and typically we see HV outputs generated by multiple stacks of multipliers connected in series.

Maybe I do not have a "question", but I am really interested in hearing people's thoughts on how to improve the flyback converter to generate high voltages. Is there a way to deal with the high parasitic capacitances, in the hardware? What are the best ways to wind a transformer for high voltage to minimize these capacitances? Literally anything you can think of to contribute to this thread will surely help me with my endeavour to produce a highly efficient, compact flyback.

Specification: 5kV and 2kV outputs @ 20mA, 180mA. I have tried very hard to design a resonant LCC converter but have found it too difficult so I am attempting to make a start on a more simple, preferably fixed frequency flyback that I can get a working prototype of ASAP.

Any and all advice appreciated!
 

Janis59

Joined Aug 21, 2017
1,227
2kV*200mA=400W. Its not so small at all. Thus, for such power the option is bridge circuit with C-W multiplier.
For example, excitator side similar than here https://i.pinimg.com/originals/5d/c4/de/5dc4de9883460dc6b11a371b4dc3511c.png (let correct the frequency to the 20-40 kHz). and secondary side lke this https://www.researchgate.net/profile/Shiho_Kim/publication/31240985/figure/fig2/AS:668938052575235@1536498668265/Cockcroft-Walton-full-wave-series-voltage-multiplier.png
Or use one of ideas shown at http://www.danyk.cz/ upper section (note the language switcher bar).
 

Orson_Cart

Joined Jan 1, 2020
18
400W = 2 flybacks interleaved, 4 outputs rectified in series to get the 2kV reliably at this o/p current, 50kHz, max, we have built them to 6kV this way for insulation testers, kind regards ...
 

Thread Starter

SiCEngineer

Joined May 22, 2019
238
400W = 2 flybacks interleaved, 4 outputs rectified in series to get the 2kV reliably at this o/p current, 50kHz, max, we have built them to 6kV this way for insulation testers, kind regards ...
What topology would you suggest for a higher frequency high voltage supply?
 

Thread Starter

SiCEngineer

Joined May 22, 2019
238
2kV*200mA=400W. Its not so small at all. Thus, for such power the option is bridge circuit with C-W multiplier.
For example, excitator side similar than here https://i.pinimg.com/originals/5d/c4/de/5dc4de9883460dc6b11a371b4dc3511c.png (let correct the frequency to the 20-40 kHz). and secondary side lke this https://www.researchgate.net/profile/Shiho_Kim/publication/31240985/figure/fig2/AS:668938052575235@1536498668265/Cockcroft-Walton-full-wave-series-voltage-multiplier.png
Or use one of ideas shown at http://www.danyk.cz/ upper section (note the language switcher bar).
I notice one of these seems to be a push-pull type configuration. Looks nice. What frequenccy do you think a push-pull can be pushed to in a kV level application? Is 300-500kHz achievable at all?
 

Thread Starter

SiCEngineer

Joined May 22, 2019
238
that depends on how high the desired freq, and voltage ....
Voltage is high (relatively, imo). Highest voltage is 6kV. I aim to eventually do some kind of trade-off study about how as freq. icnreases, the size of magnetics, etc, changes and is then limited. So to answer the Q, as high as possible before it becomes detrimental.
 

Thread Starter

SiCEngineer

Joined May 22, 2019
238
What is the input voltage? 110ac or 220ac or 12Vdc? Do you need isolation from the power line? Do you need regulation?
Input voltage is rectified aircraft AC, 400Hz, but I am to assume a 270VDC input voltage plus minus 10 perc. Isolation from power line, I am not sure. I assume this is already done by the time I receive the DC voltage. Yes I need regulation, it is very sensitive equipment and good regulation/ minimal ripple is essential.
 

Thread Starter

SiCEngineer

Joined May 22, 2019
238
I have tried the LCC converter but the transient response with frequency control is too sluggish. I tried to regulate with a buck current front end but having fixed frequency LCC was not working out well for me.
 
I notice one of these seems to be a push-pull type configuration. Looks nice. What frequenccy do you think a push-pull can be pushed to in a kV level application? Is 300-500kHz achievable at all?
Realistically, 50kHz is about as high as you can go, assuming a QR flyback with 6 o/p wdgs, rectified and then put in series,

the losses at higher frequencies from the diodes and the wdg capacitance cause excessive heat and diminishing returns above this freq.
 

ronsimpson

Joined Oct 7, 2019
935
Many of my HV power supplies were in the 50 to 120khz range. I started out at 15750 for a reason. Most were QR single transistor flyback.
The HV winding can have high loses at high frequencies. This is a piece of a supply from a couple years ago. Here I broke the secondary up into 3 pieces to keep the self resonant frequency of the transformer as high as possible. Also fast HV diodes were not available. If you split the secondary into 3 or 6 sections you get the 6kv and 2kv. In this option there is only 1kv across each secondary.
1587330955113.png
About 40 years ago I started splitting up the secondaries to keep the ac losses down.
 
SiC makes a huge difference to HV rectifiers, only very small snubbers needed - the voltage doubler o/p seen above from Mr Simpson is fine but the diodes must handle 2kV + any over-volts due to internal capacitance and L1 leakage, that ckt would not work at 120kHz, dividing into sections overcomes corona discharge too - which can easily and quickly eat thru insulation ...
 

ronsimpson

Joined Oct 7, 2019
935
that ckt would not work at 120kHz
I have used it way above 120khz.

Years ago we made "spare tire" flyback transformers. It was hard to drive the transformer when it self resonated at 3, 5 or 7x the switching frequency. Also the stack of diodes had to with stand full voltage. Using the "TV horizontal" approach is good because the transistor and diodes see a half sine wave with no sharp edges. Another reason to think QR or some approach with out sharp edges.
1587339736866.png

Next we broke up the secondary in 5 to 10 pieces and hid the diodes inside the transformer.
For monitors we made the turns on each winding slightly different so each resonated differently. (staggered tuned)
For industrial power I kept the turns the same.
1587340190469.png
 
Nice photo's of the 5 & 20kHz stuff - I should like to see some 120kHz stuff - any one who has worked on 6kV + knows that 120kHz is not a good starting point unless you have access to the good insulation systems and the best of SiC diodes ...
 
Top