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

Re: [TCML] IGBT paralleling



I'm still fuzzy, is it, or is it not technically/theoretically possible to parallel IGBTs (we're rebuilding Wardenclyffe and for giggles going solid state and no single IGBT will work.) Obviously not really, I just didn't get a completely clear answer yet, yes or no and why or why not. One cannot parallel BJTs without equalizing resistors and even then it is dicey (because current in parallel divides) but one can Parallel MOSFETs for like a PWM or something with little or no trouble (because voltage in parallel is constant, and FETs have infinite gate resistance once on?) With IGBTs in parallel the power dissipation of each one will be half, so I still think there would be potential applications for this. Also I'm hearing about IGBT modules, what are they, what do they do, can they be used for coiling purposes... Thanks for your patience guys, interesting good info even if my original question is still opaque to me.

Scott Bogard.

On 7/14/2010 1:36 PM, Steve Ward wrote:
The DC current rating for an IGBT is basically just the power dissipated
(Vce sat @ Ic) times the thermal resistance from die to case, keeping the
peak temperature<  ~150C.  At least several of the IGBTs i looked at
appeared to arrive at the current spec in this way.

I typically ignore the current rating and instead look at conduction losses,
switching losses and thermal impedance, because honestly thats the bottom
line.  The amp rating is just some number that is almost never applicable to
the application.

As far as voltage headroom goes, operating at 50% of the part rating is
quite good for most cases.  Ive pushed some of my bridges all the way up to
their voltage rating (zero headroom), and that might work under controlled
conditions.  Ive also tested some older IGBTs for their ruggedness to
transients caused by hard switching of the IGBT or diode.  It seemed like it
took very big and high energy transients before my CM300DY failed, and that
normal "switching noise" kind of stuff would not really bother it much.
This suggests that running a higher bus voltage may be worth it.
Reliability data to settle disputes like this seems scarce.

I think a lot of coilers tend to neglect the thermal aspects of whats going
on in the IGBT (among other things).  Admittedly, its not a simple task, it
requires knowledge of some sort of SPICE program, and to extract the thermal
time constants from the data-sheets plots of Zth vs time.  Ive done it
before to check the temperature peaks that i expect my IGBTs to see, and
also to figure out the average die temperature after a minute or so of
operation.  I suspect thermal stress is a leading factor in the failure of
most DRSSTCs, particularly when someone tries to push 500A through a TO-247
device... the thermal mass is much smaller than a brick, and you dont see
people pushing bricks proportionally as hard.

To give the real numbers on the coil DC is talking about, it was CM300s
operating at ~1800Apk playing 4-note polyphony.  With no power factor
correction on the AC line, we were maxing the 100A meter installed on the
variac.  The RMS primary current is estimated to be (via pspice simulation)
in the 200A range.

Steve

On Wed, Jul 14, 2010 at 9:10 AM, Michael Twieg<mdt24@xxxxxxxx>  wrote:

Literally no amount of circuitry can eliminate ringing and spikes due to a
device's internal inductance.  The actual collector of the device is not
accessible to protection circuitry (unless you plan on tearing off the case
and soldering directly to the die).

And "pushing" current is very different from pushing voltage.  The current
ratings for IGBTs are given under the assumption that they are being hard
switched, which causes far more power dissipation than if they were soft
switched.  That's why the current ratings can be pushed so far for tesla
coil bridges.  It's not that the manufacturers are being conservative with
their ratings (they have no incentive to do so), it's that we're not
operating them in a typical fashion.

Unlike the current ratings, the voltage ratings can't be pushed.  It's a
pretty hard limit which is independant of what load you're driving.  If you
try to operate your bus voltage above those ratings, they will fail very
quickly.  And any sane engineer will give at least 25% headroom on the bus
voltage (much more if they don't have very good snubbers).

-Mike

On Tue, Jul 13, 2010 at 10:23 PM, DC Cox<resonance@xxxxxxxxxxxx>  wrote:

Proper circuit design eliminates the spikes and ringing problems.

Most IGBT circuits are "pushed" a bit.  Steve Ward, at my open house, was
pushing his CM600s with up to 80 Amps RMS at 220 V input.

Dr. Resonance





On Mon, Jul 12, 2010 at 4:48 PM, John Forcina<forcijo10@xxxxxxxxx>
wrote:

Apparently so.  I would like to see how DC will be able to get away
with
this with these mysterious IGBT's because if he claims they are rated
at
2400V and 1650*1.414=2333.1v that gives almost zero headroom for
voltage
spikes ringing etc...

It is pretty much impossible to make a bus layout that can accommodate
these
igbt's.  Even with a very low inductance laminated layout the sheer
slow
switching speed will create large switching spikes and will lead to a
certain death to the igbt's.  Not to mention, any decent power engineer
will
know to de-rate the igbt's by several hundred volts to accommodate for
this
and also because it's just good practice.

On Mon, Jul 12, 2010 at 1:12 PM, Drake Schutt<drake89@xxxxxxxxx>
wrote:

So DC you're saying that you raise AC voltage to 1.7kV before
rectification?

Sent from my iPhone


On Jul 12, 2010, at 9:08 AM, "Brian"<brianv@xxxxxxxx>  wrote:

  Hmmm Im a little confused, 2400 vdc IGBT is the handling voltage of
the
device , once an IGBT is on...it is on, I am not sure what this
wasted
head
room is all about. Once the miller capacitance is overcome the IGBT
is
considered on and now connects the rail voltages. Whether it is big
voltage
or little voltage it don't matter. If you wish to drive them hard at
full
rated with 2400Vdc then drive them hard if you wish not too then
don't.
I
am
not sure where the idea came in that there is a bunch of wasted head
room
that has to be filled...maybe I am missing something in this
dialogue
somewhere...

-----Original Message-----
From: DC Cox [mailto:resonance@xxxxxxxxxxxx]
Sent: Sunday, July 11, 2010 6:27 PM
To: Tesla Coil Mailing List
Subject: Re: [TCML] IGBT paralleling

With a 2,400 VDC IGBT, running it at a line doubled 220 VAC gives
2400
VDC
-
616 VDC --- over 1,784 Volts of wasted headroom that needs to be
filled,
hence the use of a power transformer to boost the AC input from 220
to
around 1700 VAC.  The headroom I'm referring to is similar to your
audio
reference only in this case wasting AC power headroom on a large
IGBT
that
should be driven at higher potential to maximize coil output.

This is, of course, all not relevant with standard medium size coils
using
rectified line drive (such as CM300 circuits) or line rectified
drive
with
a
voltage doubler circuit common with CM600 IGBTs.

D.C. Cox




On Sun, Jul 11, 2010 at 4:13 PM, Drake Schutt<drake89@xxxxxxxxx>
wrote:

  Dc- what do you mean when you refer to headroom in this post?  I'm
used to the term only in music production referring to dB.

Regards
Drake


On Jul 11, 2010, at 3:22 PM, DC Cox<resonance@xxxxxxxxxxxx>
wrote:With
CM300s or CM600s you can just double the 220 VAC line to get 642

VDC

for a good match.

If running, perhaps, a CM2400, you want to go up to near 2,400 VDC
on
the drive, so you end up using a 220/480 Volt 3 phase to get up to
at
or near the 2,400 VDC rectified.  This gives you better output
because you go from
642 VDC to 2400 VDC that is being switched into the primary
inductor.

Typically, with really large systems the only way to get from
220/440
VAC to produce the 2400 VDC drive max is to use a small 25 to 50
kVA
xmfr (surplus pole units).

Not a dual pole pig unit, just a 220/440 VAC 3 phase xmfr
delivering
around 1650 VAC before rectification.

The main point with large DRSSTC type coils is efficiency ---
eliminating all those losses in the heat&  UV light production in
the
spark gap, and obtaining quicker dI/dt rates.

As you pointed out small and medium size coils run just fine
without
using any pole xmfrs.  Usually above 15-18 ft long sparks the pole
xmfr boost helps out get to the higher potential of the larger
IGBTs
without wasting a lot of headroom.

Dr. Resonance




On Sun, Jul 11, 2010 at 2:14 PM, Gary Lau<glau1024@xxxxxxxxx>
wrote:

I'm going to bare my ignorance here.  I thought that the whole
point
of

solid state TC's was that you don't need a multi-kilovolt power
supply.
Are
pole pigs really used to power these?  A _dual_ pig powered
magnifier???

Regards, Gary Lau
MA, USA

On Sun, Jul 11, 2010 at 12:16 PM, Scott Bogard<
sdbogard@xxxxxxxxx

wrote:

Hi John,

    By chance is there a compiled list somewhere of "good IGBTs"
that are used and those that aren't.  Anyway back to the
original
question, is

  it

  possible to parallel them?  Lets say for kicks I'm building a
dual
pig powered 30kVa magnifier with a LTR cap (I'm clearly not, we
are
talking theoretical here.)  Clearly the peak currents will be
beyond any

  reasonably

  priced IGBT, is it possible to parallel lesser current units to
handle

  the

  load, and what would that entail?  Thanks.

Scott Bogard.

On 7/10/2010 5:09 PM, John Forcina wrote:

Those IGBT's seem far from ideal.  The TO220 package is a very
poor


  choice


  as far as thermal conductivity and the datasheet says it all
0.75C/W.


  You


  will not be able to remove enough heat from the surface of the
IGBT
die

quick enough between current pulses and the device will fail.
They do

  not


  have a internal anti-parallel diode so adding that externally
will
add


  to


  the final cost also.  Not to mention doing that will add
additional
loop

area and stray inductance between units.  One more thing is the

  switching


  times are surprisingly slow for that small of a unit.  td(OFF)
96ns.


  It's


  not that slow however it does seem slow for that small of a
device.
I

have
seen much better overall performance from larger IGBT's.  My
suggestion

  is


  to just spend more money and use IGBT's that have been used and
proven


  to


  work in Tesla Coils time and time again.  There must be some
reason
that

we
all use them ;)

On Sat, Jul 10, 2010 at 4:38 PM, Scott Bogard<
sdbogard@xxxxxxxxx


  wrote:



  Interesting,

     I am in the wee beginning stages of building my first SISG,
and

  as


  such am in the market for IGBTs.  I found these, which look very

inexpensive
and have decent ratings.






http://www.newark.com/fairchild-semiconductor/hgtp12n60a4/single-igb
t-600v-54a/dp/90B5642

  My thinking is if heat is a problem or peak current, can I just


  parallel


  them?  at $1.50 a pop it seems infinitely better than 1 $18 IGBT
of

nearly
the same ratings...  I didn't look at temperature or package

  information


  yet, so maybe there is a problem there.


Scott Bogard.


On 7/10/2010 8:26 AM, McCauley, Daniel H wrote:

Scott,


The electric ratings may be the same or similar, but you also
have to compare the mechanical ratings -  in particular the
thermal

ratings.

This
would be junction-to-case thermal impedances etc...  The
expensive

  IGBTs


  that are commonly used in DRSSTCs are usually ISOBLOC type
packages

which
excellent thermal impedances.  Compare this vs. a TO-247
package
of

  the


  same

die.

And when comparing a TO-247 package to an ISOBLOC, keep in
mind
that

  you


  STILL NEED to add a thermal insulator between the TO-247 and
heatsink,

which
just makes the thermal impedance even worse.  The ISOBLOC (or
SOT-227)
doesn't require a thermal interface other a small smidgeon of
thermal grease or a graphite pad.

Dan
http://www.easternvoltageresearch.com
DRSSTC, SSTC, Flyback, Plasma Speaker Kits





-----Original Message-----
From: tesla-bounces@xxxxxxxxxx [mailto:
tesla-bounces@xxxxxxxxxx]
On Behalf Of Scott Bogard
Sent: Friday, July 09, 2010 9:52 PM
To: Tesla Coil Mailing List
Subject: EXTERNAL: [TCML] IGBT paralleling

Greetings all,
    So, after a bit of researching I've noticed there are
IGBTs
on Newark with exactly the same ratings as some of the SSTC
approved

  IGBTs,


  but

at 1/10th the price.  What makes these others so special that
they are better, and if it is just a matter of peak current
ratings, since

  IGBTs


  are

gate driven, can we just parallel a few to get the required
pulse current rating?  I ask because I've not heard of
anybody
doing this, then

  again


  I've

only begun to research SSTC a little bit ago.  Just musing.

Scott Bogard.
  >>>>>>>>>
_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla
_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla

_______________________________________________

Tesla mailing list

Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla

_______________________________________________

  Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla

_______________________________________________

  Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla

_______________________________________________

Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla

_______________________________________________

Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla

  _______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla



_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla

_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla

_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla

_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla

_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla

_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla

_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla