[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Sold state IGBT disruptive coil spark gap idea (SISG)

Original poster: "Gerry  Reynolds" <gerryreynolds@xxxxxxxxxxxxx>

Hi Terry,

I've got an old 2oz jar of Wakefield thermal compound (white thick paste) that is non conductive and 0.05 degree per watt for a 1 mil thick 1 square inch film. Suspose to be 1/4 the impedance of silicon grease. Dont know how this compares to the silver stuff nor do I know what the price is.

Gerry R.

Original poster: Vardan <vardan01@xxxxxxxxxxxxxxxxxxxxxxx>


I am waiting on orders again so things are slow :-p

I have been studying if SCRs could be used instead of IGBTs, a much cheaper option, but the high dI/dt punch through problem seems devastating.... I ordered the "magic" SCR though just to "try"... They are, again, only available in the "SUPER 247" box...

Two things are now an issue.

1 - I am firing 12000V IGBTs at 900V. If I add just another SIDAC, I can fire at 1000, 1100, 1150, 1199.9999... I wonder if I am just wasting headroom on the IGBTs? They are "darn" expensive things and getting a little more voltage out of them does save serious beer money... You need some margin for "stuff" and lead inductances... Maybe a short circuit case is a "bad load"... Not at all sure... But I don't want to just have "waisted margin".... Unlike MMCs the 1200V limit is extremely real and unforgiving... But I got a lot of OVP across it...

Maybe I would be better off using 900V IGBTs closer to the limit since they are so much cheaper... But more support electronics and PC board space...

Probably too early to worry about that stuff since it is "just get ONE working you dummy" time right now :-)))

2 - The "SUPER 247" package from IR is "unique"... A great idea since eliminating the hole allows room for a lot more die and the massive copper slab give great thermal transfer!!! I used to write company internal papers 15 years ago about such things ;-))) They did everything I told "other" people to do :o))) The super 247 is also significantly cheaper for the power and this project needs all the "cheaper" it can get ;-))

However.... There is not a darn "real available commercial heatsink" out there form them. Nobody mounts a 600W device with a "clip"... All the TO-220 clip heat sinks are not wide enough for the 247...

So I am thinking of thermal glue like this:



But our case is only like 10W, so maybe just JBweld or "common" epoxy would work just fine.... I worry about air bubbles in the layer but with the copper slab that is not a big deal... I used to work with the latest and greatest things there, but that was stuff I was glad "I did not have to pay for" (the above silver epoxy is two orders of magnitude cheaper though...) Non-conductive is real cool since it can easily shoot arcs over the surface of silver filled non-conductive epoxies... But I would not trust that property over time. Best not to be messy with the epoxy ;-)) Isopropyl alcohol wipes it up real good before it cures... Good Ol' JBweld is very good at thermal shock and all kinds of real world "not failing" issues... Probably the best...

It would be wonderful if NO heat sink was needed anyway... But since this seems nice for lower voltage higher current systems... I^2R heating will probably need more thermal than just the SUPER 247 box alone (40C/W)...

Still thinking ;-))  If this all works out, it will be SO cool!!! :-)))