Help selecting DC-DC converter topology for capacitor charger / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Help selecting DC-DC converter topology for capacitor charger

Move Thread

LAN_403

ConKbot of Doom

Tue Nov 26 2013, 06:43PM

Registered Member #509 Joined: Sat Feb 10 2007, 07:02AM
Location:
Posts: 329

I'm working on a capacitive discharge welder project using (3) 1/3 farad 50V capacitor banks. I have a 48V 1Kw, 20A power supply that I'd like to use. I'm trying to charge the capacitors to an adjustable setpoint between 5V and 48V, and use a constant power charge profile to minimize waiting between welds. I was also hoping to add a spark erosion front end in the future where I'd hold the thyristors on and use this as a constant voltage power supply, so the 3 DC-DCs would have to be able to parallel easily.

Given that I'm looking at ~ 330-350W per DC-DC converter I figured that I might be out of a buck-converters league, so I was looking at forward, push-pull and full bridge topologies, with push-pull seeming to be a good choice. However I then came across a few controller ICs and papers on current-fed push-pull converter, using basically a buck-converter to adjust the output voltage, with the push-pull running at a constant 50% per transistor (dead time for voltage fed, overlap time for current fed push-pull obviously). These seemed to be favorable to paralleling outputs, However given that I dont require additional isolation, nor a large change in output voltage, I'm not thinking that I would gain anything using a 2(ct):1 push-pull stage.

Should I be looking at a 1:1 (2:1?) flyback converter to try and achieve the constant power output? At "wide open" use cycle-by-cycle current limiting to get a average current limit on the input?

Ive not done stuff of this power level before, just TI/maxim/etc switchers where if you follow the datasheet, it is generally well behaved and you get good results. Thanks for any help!

Electra

Wed Nov 27 2013, 12:11AM

Registered Member #816 Joined: Sun Jun 03 2007, 07:29PM
Location:
Posts: 156

Given a good choice of switching transistor and diode, canâ€™t see why buck convertors canâ€™t be used in the KW range. Perhaps a current mode buck? You can set a hard peak current limit. Then have some high side current sense for the average current.
Trying to think what would work for the constant power function. Using an analogue multiplier and op-ampâ€™s to give a 1/Vout (you want to reduce the current as output voltage increases) seems a complicated way of doing it. Whether just using the average input current to control the PWM would work I donâ€™t know.

Finally work out what your time constants are, can you simplify and just use a current limited supply? Is an extra second or two that important in your applaction.

Dr. Slack

Wed Nov 27 2013, 04:42PM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

What would work for constant power in a buck converter is servoing it so that it always draws the same input current. However, with an input current of 20A at 48v, if you want a higher output current at low output voltage, you will need either some very tasty input capacitors to handle the surge, or a multi-phase buck (look for motherboard CPU supply topologies). At some low enough output voltage, you won't be able to keep increasing the output current, so must revert to constant current output.

If, for instance, you used a 4 phase buck with no input capacitor, then you could get 80A output up to 12v, and then 960W above that.

ConKbot of Doom

Wed Nov 27 2013, 05:54PM

Registered Member #509 Joined: Sat Feb 10 2007, 07:02AM
Location:
Posts: 329

Constant current would also be limiting my power output for a spark erosion front end also if it needs a voltage significantly under the 48v Current mode bucks do look like they would be good, and parallel together well.

The LTC3810-5 looks interesting

The duty cycle would be limited to ~75% at most, but I think I can deal with 36V instead of 48V. The Vrng pin on the 3810-5 sets the maximum current, so feeding that with an inverse of the output voltage and get a constant-ish power output, while still current limiting the output.

edit: Dr slack posted while I typed up most of a response, walked away, and came back 2 hours later :p

I found what I think is a good solution now though

LT3763, high current stepdown converter. Input and output current sensing and regulation and voltage regulation too.

It regulates the input current shunt to 50mV drop max, and the inductor current shunt is set via one of the 2 control inputs, and a feedback pin for voltage regulation. Along with scaled up input and output current monitor pins.

And obviously I cant keep constant power up forever, but I was hoping to keep it up down to the 10/12v region, where I'd have to limit it to 33A/converter, and get inductors which wont saturate at that current.

LT article with example schematics about how awesome their converter is :p

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob