Terry's DRSSTC - Starting to come together...

["Tesla list" <tesla@xxxxxxxxxx>]

Original poster: Terry Fritz <teslalist@xxxxxxxxxxxxxxxxxxxxxxx>

Hi All,

My second try at the protection card works perfectly ;-))

http://drsstc.com/~terrell/pictures/ProtectionCard.JPG

http://drsstc.com/~terrell/schematics/ProtectionCard-01-sch.gif

http://drsstc.com/~terrell/pcbart/ProtectionCard-01.gif

I had a tense moment when I could not find two stupid 47K 1/4 watt 
resistors in the house to complete it!!  =:O  But in the oldest deepest 
darkest bag of resistors, I finally found to of them!  I will order about 
10,000 extra from DigiKey Monday :o))

This card reads the primary current and has an adjustable trip point from 0 
to 500 peak amps.  If the current gets too high, it drops out the H-drive 
for 500mS.  The card also monitors the IGBT heat sink temperature and is 
adjustable from 0C to 150C to stop the H-bridge if the IGBTs get too hot.

I sort of dry fit the stuff in the box so far:

http://drsstc.com/~terrell/pictures/RoughtFit.JPG

I probably should have tried to squeeze the input filter and cap board down 
in size a little...  The H-bridge almost gets lost since it is very 
small.  I will either have a thinner fan from the top or mount the board on 
it's side.  I have not "tested" the bridge yet, but I really like it!!

http://drsstc.com/~terrell/pictures/H-Bridge-Finished.jpg

The bridge has four IRG4PF50WD IGBTS with built in reverse diodes as my 
OLTC used.  They do about 750 amps and were some of the first high power 
fast semiconductor IGBTs to use copper plated conductors on the die 
surfaces to really get current out (not a secret anymore).  I worked on the 
plating power supplies that make the semiconductor die for them ;-))  (in 
fact, I guess I worked on "all" the plasma supplies that make them...)

The TO-247 packs are mounted directly on the isolated heat sinks without 
any "silly pads".  Electrical insulating pads are super bad for thermal 
transfer!!  The packs are also mounted with spring "Bellville" washers 
which are really super important to "squeeze" the thermal grease down to 
minimal thickness over time.  So the "die" and the "heatsinks" are really 
thermally connected well!!  The thermal protection and nice fan take over 
from there.  The reason for using small IGBTs and very small heatsinks is 
that this whole card has only $75 worth of parts on it!  All the traces are 
super short and the thing is very small which helps speed and reduces all 
that "parasitic" crap.  Sort of a repairable/workable "brick".  It can be 
happily run from GDTs or opto drivers.  For Fo bypass, it has two 4.7uF 
600V CD 940 poly caps right under it across the buss.  Not much will get 
past them to the electrolytics ;-))  I also have a 470V 6500A MOV across 
the output to "catch" the primary loop high voltage incase the IGBTs blow 
themselves away.  I "think" the MOV's capacitance (~400pF) will also help 
"tame" the switching a little.  Big 30V transorbs very directly from 
emitter to gate...  The heat sinks have mounting pins but I tear two of 
them out (got in the way...) and don't solder them incase I have to pull 
off an IGBT/heatsink combo.  There is not too much on the board so "IF" an 
IGBT blows, it should be cheap and easy to repair.  The IGBTs are $7.11 in 
QTY 10 and $5.33 in QTY 100 (sold out from DigiKey tonight >:-))  All the 
connections to the board are big bolt on or screw clamps for easy hook 
up.  The 5.1 ohm resistors may change in value at a whim...  I might have 
added 10K little resistors across the gate just to pull down the IGBTS for 
testing, but that is trivial.  Maybe should put a drain resistor across the 
caps (0.27Joule)...  The board can be "played" with without fear of ESD 
damage...  So while Jimmy, Dan, Steve, and Steve may have already figured 
everything out, maybe there is something a little "new" here ;-))

The controller is looking very much like a digital computer (old style) 
these days...  The big aluminum box is real nice since it houses the mess 
very nicely!!  I worry about propagation errors...  I know a guy that 
builds chess playing CPUs/computers out of plain TTL gates for 
"fun"....  Might have to get his advice...  It would be easy if it did not 
have to work so darn fast...

Cheers,

	Terry