This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Development of AC/DC 100kw 30kv step up converter

Other Parts Discussed in Thread: UCC3895, UCC28950, UCC27321, UCC37321

HI  everyone!

I am going to develop  DC/ DC  600v  to 30Kv  100kw step up converter for some science project.

Actually I am newbie in ZVS/ZCS technology, what kind of controller you would suggest for this not easy project. I liked UCC28950, but in datasheet i found only step down application scheme, is there any way to use this controller for step up converter for 30kv if yes then how?   or better to take UCC3895 without driving the synchronous rectifier MOSFET?

In UCC3895 datasheet scheme threre are gate transfomers is it better solution than using for example UCC27321 and UCC37321 drivers?


  • Hello Arman,

    UCC28950 is a controller for phase-shift-full bridge converter, which is basically a isolated buck (step-down) converter. However, it surely can be used for step-up with adjusted transformer turns-ratio.

    Before selecting controller, it is better to discuss the topologies that you select. Could you please advise which topologies do you prefer? single stage two-level phase-shift-full-brdge wouldn't work perfect. You may have to consider multi-level converters, like the Cascaded H-Bridge and etc. You need to breakdown the 30kV to fit proper power semicondcutor.

    Then we can talk about the controller and driver. Most likely, it will be good to design it with Multi-level converter with DSP control. and gate driver will be the interface between the DSP and power transistor.

  • Thank you for reply Wei

    It would be preferable to use single stage phase-shift-full-brdge because we have IGBT modules (two in one module) 300 A rms 1200 V and they surely enough for input 111 KW power, considering that in worth case 90% eff. of system and 100 KW exit power.

    In case of Multi-level Cascaded H-Bridge converter output wave is quite sinusoidal and it's harder to filter it at the output, where each surplus capacitor costs enormously, in respect that system requirements of ripple coeff must be as less as possible.

    The main attention we attached to character of load which sometimes may happen arcing, and then in the time of as fast as possible our computer will apply a diminished D factor. However, whilst it happens, the igbt may not be smoked out, therefore is extremally essential to keep into transformer inductance and core as small power in terms of single cycle as possible. That is the reason why we see the multilevel bridge un-impropriate, as it keeps in trafo much more energy as at one cycle used, to contrast to ZVS/ZVT where at end of cycle trafo is drained just empty by definition of ZVT.

    Thus the architecture would be EMI filter in network side, three phase rectifier with large reduced capacitors which laid Vcc swinging within 470-535 Volt in network freq, what is compensated by proper D adjusting in phase with. Just capacitors there takes absolutely too much a cubature. Of course the high freq capacitors will be installed in proper count, and sendwitch feeder line will be applied to avoid the inductance of high current wires. After that stays a 4 igbt bridge, between halves is installed a 7,5 cm2 core transformer with proper small inductance in series, without of any capacitors (which for peak current 1000 Amp and reactive power 120 000 VAR would be unimagineable. Thus, after secondary winding comes the Gretz 4 diode rectifier, actually it consists on 120 diodes, ferrite choke and 40 kV capacitor battery, where we have bounded hands to apply only existing megacapacitors. Centertapped winding is wrong option due insulation problem in trafo and proper anti-arcing distances.

    So, the core of question where we have no clear mind: DOES the any controller designed for ZVT AND exit MOSFET-style rectifier controlling may be applied in our situation where 60 kV mosfets are unexistent, just ignoring those outputs and use the Gretz bridge? Or rectifier controller section is important part of algorithm, and Gretz rectifier will destroy this algorithm?