I was given a lecture once on the challenges of those hockey puck thyristors used in loony power systems such as converting 1E6VDC into AC supergrid. It seems that things could go very badly wrong very quickly and not leave enough behind to tell you what had failed first. An interesting combination of brute force and very accurate switching control. All of which could be stopped by a stray weasel or dropped sandwich!
its a silicon fluid/gel... clear, sticky, messy to touch or work with. yep. Thats what it is. Also used in vibration control and damping apps.
-RNM
Nobody is arguing that.
It is not, however, used for heat transfer as you previously mentioned (in this specific application). Aside from the fact that there is nothing above it to transfer heat to, it is several orders of magnitude lower in thermal conductivity than the alumina the chip is bonded to. (remember, the substrate is DBC, with solder attach above and below.)
It is used for vibration damping of the aluminum bonding wires as I previously mentioned, and it is used as a cover so that the hard clear epoxy added next does not touch the dice or bonds, but rigidly supports the copper straps that were soldered in to support the current.
John
Makes me wonder if an lpcvd oxide layer (something low temp) could be designed to keep the boundary layer tripped. There are certainly ways to make local turbulence, or at least flow focusing (we do it a lot in microfluidics) to mitigate cavitation. That'd certainly be a unique application, I'll admit. Taking it that a phase change cooling setup can't be done directly on the die? Generally the best ready to move heat away from a local source.
And, oh the places this thread goes.
No fat? Where's the fun in that..
Hey Scott (or anyone else), I'm looking for a dicing saw. Know who I can talk to?
The object to be diced is a 67mm dia diode brazed to a molybdenum carrier. I need to saw down to just above the braze joint to turn the one diode into a diode array. I need to map the forward voltage across the diode at room, and map the turn on voltage at 4.5K using a bed of nails.
I suspect we'll have to buy one used, as I don't think I can ship it offsite. The diode is kinda, shall we say, "warm". I think it would contaminate the saw permanently, so I'm not sure if any testing labs could do it.
John
Makes me wonder if an lpcvd oxide layer (something low temp) could be designed to keep the boundary layer tripped. There are certainly ways to make local turbulence, or at least flow focusing (we do it a lot in microfluidics) to mitigate cavitation. That'd certainly be a unique application, I'll admit. Taking it that a phase change cooling setup can't be done directly on the die? Generally the best ready to move heat away from a local source.
And, oh the places this thread goes.
I suspect that phase change might cause pockets of varying temperature. At the macro scale, it's great. But since the silicon is so large but with very very small individual cells, I can see thermal variations across the die causing runaway.
I'd have to review my IGBT parametrics, as I think they are neg coeff like bipolars, as opposed to positive coeff as mosfets are. So I suspect IGBT's will runaway under phase change cooling.
John
You need to find something from the old 6" and less days. Modern used ones still fetch $20,000 or more. These guys might have something. http://www.fabsurplus.com/sdi_catalog/salesItemList.do?listTypeId=19
Last edited:
Thanks. I'll forward the link to the guy looking into it.
It's weird seeing 200 mm surplus saws, that's some huge disks.
I wonder how many 1N4148's can fit on a 200mm disc??
John
Surely the question is, after auditioning, how many sonically acceptable 1N4148s fit on the disk... 1? 2?
Surely the question is, after auditioning, how many sonically acceptable 1N4148s fit on the disk... 1? 2?
None.
It was funny, one of the girls was probing a disc of them, the final good dice count was weird, I think it was a negative number.
The auto probe system computer was 16 bit. The total was over 32K.
John
- Status
- Not open for further replies.