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

Re: SSgap Diagrams



Original poster: "Terry Fritz" <twftesla-at-uswest-dot-net>

Hi Marco,

Many thanks for your thoughts on this!  I was thinking of putting in gate
resistance to reduce the turn on times and possibly help control
transients.  A lot depends on how fast the drive switches to begin with.
Protecting the drive from over current may be a big factor too.  A little
bit of experimenting is needed there.  The "cheap" drive electronics I have
do not talk much about such things ;-))

I don't have cross conduction or power dissipation problems in this case.
The IGBTs should remain very cool which is a big advantage if they suddenly
get hit with a high power dissipating situation.  The RMS current through
them is very low (~3 amps) but the peak current is high (~150A ;-))  

Your current limiting system is very interesting.  I resist adding any more
resistance to the path but perhaps the current limit is more important.  It
is really a matter of how many joules the die surface and bulk can absorb
before the temperature gets too high.  I do have the advantage that the
system is energy (and somewhat current) limited.  Worst case is if the gap
were directly shorting the primary cap (no primary inductance) and the bulk
die had to repeatedly dissipate the systems energy for a long time as giant
current pulses.  The die would slowly heat up and the "pulse temperature"
would get near the destruct temperature.  The die can probably take
enormous currents as long as the instant die surface temperature does not
go over say 175C.  Have to look beyond the spec sheets and into the device
physics there.  I assume the wire bonds and gate structure is heavily over
designed on the die.  I suppose I should sacrifice an IGBT and remove the
plastic to see what is in there.  Probably when I blow one up...  I could
put one on the big high power Tek curve tracer at work that does 3000 volts
and about 300+ amps.  Probably good to know the real breakdown voltage and
just what happens at real high currents.

I am using TVS transient absorbers a lot.  In your system this is not
really on option since you need such great control.  In my case, I can
clamp nasty voltages and let everything else ride.  I don't have the cross
conduction or other dissipation issue to worry about.  I could simply
switch all the devices hard on as a fault protection mode.  The 60mA from
the NST is no concern and that would definitely shut the system down fast.
I also use Kelvin gate voltage connections which help eliminate the stray
inductances.  The gate power is full floating (hopfully, without too much
stray kV on top of that ;-)) However, I really should be switching slowly
enough that it is not a problem.  Having 500 amps at 250 kHz an inch away
from the drive system pretty much rules out too much fancy electronics.
Just the induced currents are a super big concern!!  Have to play tricks to
keep the magnetic fields contained or cancelled out.  

The real fun things are considering if a section fails, how to keep damage
from propagating to the other sections.  Also, how would the system react
to direct streamer hits at say the IGBT gate!  (of course, 400kV can go
right through the plastic IGBT and directly hit the gate structure on the
die too ;-))  I figure it is best just to locally protect the devices as
much as possible and if something blows they will all just have to fend for
themselves to ride out the event.  If I could detect the problem
electronically, I could shut it down really fast.  The real problem is when
(I mean IF ;-)) something goes wrong, it may just appear as the streamers
getting a bit shorter and I will just turn up the variac more ;-)) 

In many ways, this really is a very simple thing to design (a slow ~120Hz
switch).  However, the high voltage, high current, and massive noise is a
real challenge.  I figure it is just best to pull together the parts and
throw in as many local protection devices as possible.  This is sort of a
really odd application ;-)) and there is not a whole lot of info out there
about running IGBTs (or electronics in general) at these levels.  However,
the problems should be pretty straight forward and hopefully not to hard to
solve.

However, I am sure I will have to find a big bean can...  ;-))  Of course,
if it was easy, everyone would do it!

Cheers,
	
	Terry


At 09:51 AM 3/30/2001 +0300, you wrote:
>Terry,
>
>I see you plan to use the IRG4PH50UD IGBT. I used the IRG4PH40UD and
>before having it working I filled a bean can with burned IGBTs. And
>think that my maximum current (besides spikes) was about 30A!
>To avoid you filling a trash can (sensibly bigger than a bean can, at
>least here in Finland!) with burned IGBTs I strongly recommend you to
>read my paper about problems and solutions with the Thor CCPS located at
>
>http://www.saunalahti.fi/dncmrc/papers.html
>
>Second paper from up, chapter "Problems encountered and solutions".
>You'll see you have better to use a gate resistor, etc. I suggest also
>to read as much as you can about gate drive techniques from available
>application notes.
>
>More, I think you would benefit from a simple current limiting system.
>What I have been using successfully is just a 2N2222 and a SMD (thick
>film, low value) resistor for each IGBT: that allowed me to use just ONE
>bean can. See the driver schematics in my paper, which is an UPDATED
>version of the schematics available on my web pages.
>
>Regards
>
>
>Tesla list wrote:
>> 
>> Original poster: "Terry Fritz" <twftesla-at-uswest-dot-net>
>> 
>> Hi All,
>> 
>> Since there is a lot of interest in my little solid state gap project, I
>> posted a few diagrams and schematics at:
>> 
>> http://hot-streamer-dot-com/TeslaCoils/MyCoils/SolidStateCoil/SSgap3-29-01/
>> 
>> I left out some of the plain Jane details of the controller's electronics
>> (lots of LMC555 and 74HCT stuff) since it is messy, but this will give you
>> the idea as to what it is about...  The real key is the Power_block,
>> Sync_sender, and Control_Panel stuff since that shows mostly what it is and
>> how it works.  These are still real rough but show the basics (I just
>> noticed the optical sender's logic is reversed ;-)).  I expect to be
>> putting a lot of time into this next week. ;-)))
>> 
>> Cheers,
>> 
>>         Terry
>
>-- 
>_____________________________________________________________
>
> Marco Denicolai           Senior Design Engineer 
> Tellabs Oy                tel: +358 9 4131 2769
> DSL Products              mobile: +358 50 353 9468
> Sinikalliontie 7          fax: +358 9 4131 2410
> 02630 Espoo  FINLAND      email: marco.denicolai-at-tellabs-dot-com
>_____________________________________________________________
>