Hi Steve, I had a look at Semikron but the parts seem a bit slow, the inverters need to generate a waveform up to 10 KHz so I need to be switching around 100 KHz at least. Regards Luis -----Original Message----- From: piclist-bounces@mit.edu [mailto:piclist-bounces@mit.edu] On Behalf Of Steve Smith Sent: 21 March 2006 18:46 To: 'Microcontroller discussion list - Public.' Subject: RE: [EE] design advice wanted Take a look at UPS design they have the same circuits in them any manufacture is very similar + and - 15 ish for the upper and lower driver (isolated from each other) and optos for the signals and Desat detection. The 3Kv is a problem as this is the inter winding voltage so the transformer needs to be well designed and off the shelf bits just don't cut it. Piped LVAC (HF) is noisy and may lead to EMC issues but when it arrives at the driver board its simpler than piped DC you can use two separate transformers or design a transformer with good inter winding voltages and drive either single transformer or local switching supply. Check out semikron they make an upper / lower driver with built in psu works from 24vdc provides all the igbt drive and desat detection I think they are rated to 1200v dc bus so they sound like an answer to the problem I think they are called something like an SKHI21 long time since I used them Regards, Steve -----Original Message----- From: piclist-bounces@mit.edu [mailto:piclist-bounces@mit.edu] On Behalf Of Luis Moreira Sent: 21 March 2006 15:37 To: Microcontroller discussion list - Public. Subject: RE: [EE] design advice wanted Hi Guys, Thanks for all the input. I do favour the DC-DC converter solution but: - As you say the issue is how to design it well into the board as to not be a problem to the driver circuits. Also there will be 24 of this dc-dc psu. Suggestions and tips welcomed. - With the IGBT switching with a system that will operate at 1200V +-3KA peak max. Will the DC-DC converters be affected? This is to upgrade an existing system which as far as I can tell is not the best design in the world. Any experiences welcomed. Probably there are other issues that I need to look at such as capacitive coupling, reliability, PSU fault monitoring and so on... Best regards Luis -----Original Message----- From: piclist-bounces@mit.edu [mailto:piclist-bounces@mit.edu] On Behalf Of Dave Tweed Sent: 21 March 2006 15:03 To: piclist@mit.edu Subject: Re: [EE] design advice wanted Luis.Moreira@jet.uk (Luis Moreira) wrote: > I am in the process of upgrading one of our systems and the bit I am > looking at the moment has 24 IGBT H bridge modules. Each of these > modules as a driver board which I need to replace. This board will need > + - 15V at steady current of around 300mA ( when switching IGBTs current > will be sourced from local storage caps).The IGBTs will be switched at > frequencies up to 350KHz. My biggest problem is how to supply power to > the boards as they need to be isolated from the rest of the system > (sitting at HV). The solutions I found are: > > - Fit an of the shelf isolated switching power supply which seems good > but probably the price and possible problems with switching noise makes > this solution not a very good one. > > -The other which was suggested by a friend is to supply the board with a > 40KHz at around 20V peak sinusoidal generated by a resonant circuit. > this would reduce the size of the transformers on the boards. Voltage > would then be rectified and regulated. If I understand you correctly, that's about 9 W per board, or 216 W overall, ignoring efficiency issues. Designing resonant circuits at these frequencies and power levels is definitely a non-trivial task. The sinewave approach has its own problems with noise if you are planning to put an ordinary bridge rectifier on the board to get the DC -- the current will be drawn at the peaks of the waveform, producing nasty current spikes that will be hard on the driver circuits and radiating all kinds of noise. Square-wave drive would be much better in terms of switching efficiency and ease of filtering, but of course the fast edges create their own issues with radiated noise. I used square-wave drive once when I needed to get about 100 W across a rotating interface, but the whole thing was fairly compact and enclosed in a metal shell anyway, so electrical noise wasn't much of an issue. Perhaps some sort of modified squarewave would be appropriate. Keep the tops as flat as possible to spread the current flow throughout the cycle and reduce ripple at the output of the rectifier, but round the corners off to reduce noise as much as you can, consistent with keeping the switching losses in the driver low. Going back to the first option, an isolating DC-DC converter at the 10 W level should be fairly compact and not very noisy at all if it's properly laid out. -- Dave Tweed -- http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist -- http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist -- http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist -- http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist