[Home][2014 Index] Re: [TCML] IGBT Selection [Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [TCML] IGBT Selection



Hi Jeremy


On Tue, Jul 8, 2014 at 8:42 AM, Jeremy Gassmann <jeremyee78@xxxxxxxxx>
wrote:

> Thank you all for the inputs! I didn't even think about checking out ebay.
> I looked last night and they were about $50 per module. I was hoping to
> reuse my current SG coil and modify it for solid state operation, so I
> guess I should measure the current resonant frequency to make sure it isn't
> too high. Is 100 kHz the upper limit for the CM300 or is it more of a soft
> limit where performance diminishes over 100 kHz?
>

The CM300 is "old" technology for IGBTs so it has significant switch time
compared to the best stuff today.  There are 2 issues to contend with as
the frequency goes up.  1, the IGBT will have to experience higher
switching currents relative the the peak current (more energy dissipation).
 2, the power factor of the driver will suffer as frequency goes up because
of the time wasted during switching OR because the driver is switching the
IGBTs late and there is a phase shift between drive voltage and current.
 Lower power factor just means you arent making full use of the output
volt-amps.  100khz has a halfperiod of 5uS, so the switch time of a CM300
at about 800nS or so is a significant chunk of that time.  The FGH60N60SMD
i recommend is ~100nS switch time (roughly speaking, its the delay time
plus the transition (rise or fall) time) so it would be a far better match
for higher frequency operation, youd just need more of them to handle more
current.  So yes, its a soft limit.


>
> One other question that is kind of related to this topic...how is the main
> power supply for the system designed?

I have seen a lot of variations using
> either 120VAC input or 240VAC input with mentions of using a voltage
> doubler and the like. Are there any design guides out there for this
> subject including what full wave rectifier to use and the size of the
> capacitors that are needed for a given supply output?
>

This would be the easiest approach.  For a doubler you need diodes rated
for at least the total output voltage (2x Vinpeak).  Specing the current
rating is not as straight forward, depends on your heatsinking and desired
power input.  I tend to use diodes with a current rating of at least 2X the
expected mains input current.  This type of supply will draw high current
peaks (rather than a sinusoidal current) from the line, which is harder on
the diodes and gives low power factor from the AC line due to the highly
distorted current shape.  More advanced supplies would use an active power
factor correction type boost converter, but thats possibly more work than
the tesla coil itself :-).  I was attracted to the idea when limited to a
30A circuit, and wanting to most power possible before tripping the circuit
breaker.

The capacitors should store at least 10X the "bang" energy of the tesla
coil.  Say you want a 5kW coil at 120pps, that implies a bang energy of
5k/120 = 41.7J.  If it used a 240V input with a doubler, you would want at
least 420J of storage at ~640VDC, or 2000uF total (2 x 4000uF caps).  More
storage wont really improve performance too noticably, but the caps may run
much cooler because they can handle higher current and have more volume for
soaking up heat (for short term use).

Steve


>
> Again, thank you all for the help!
>
> Jeremy
>
> On Monday, July 7, 2014, Steve Ward <steve.ward@xxxxxxxxx> wrote:
>
> > Jeremy,
> >
> > Most DRSSTC builders go to ebay to find used IGBT modules for TC use,
> this
> > is why the CM300DY-24H became so popular, they have been on the surplus
> > market steadily for years.  When they first hit ebay i was buying them
> for
> > just $10 each, but now days expect to pay about $50 per module.  They are
> > considerably slower switching than the APT module so you would likely
> want
> > to keep the TC resonant frequency <100khz.  I should also note that the
> > CM300 appears to be one of the most rugged "300A" modules ever, they just
> > dont make them like that anymore.  Nowdays, IGBT manufacturers are
> claiming
> > higher current ratings on less Si, and that means we cant abuse the parts
> > as much.
> >
> > The only drawback to using the compact H-bridge module is cooling it.
>  The
> > higher density of components makes it more difficult to remove the heat.
> >
> > Half-bridge does work fine for driving a TC, of course you only get half
> > the voltage out.  Some people prefer having less parts and like
> half-bridge
> > for that reason.  I think the advantage of higher drive voltage is the
> main
> > reason to go H-bridge, twice the power at the same current.
> >
> > If you want to start small and relatively cheap, i really like the
> > FGH60N60SMD.  Its a very fast part, i use it up to 400khz.  I would
> suggest
> > keeping the peak current to just 200A, and if you need more you can
> > parallel them.  Id also suggest a seperate heatsink for each transistor
> so
> > that you dont have to use electrical isolation between the transistor and
> > the heatsink (heatsinks would be live).
> >
> > I like to use EEM.com and octopart.com to source IGBTs, i often find
> > distributors that i've never heard of that sell these parts in larger
> > quantities, but less $/part than digikey and mouser.
> >
> > Steve
> >
> >
> > On Mon, Jul 7, 2014 at 9:01 AM, Jeremy Gassmann <jeremyee78@xxxxxxxxx
> > <javascript:;>>
> > wrote:
> >
> > > Yeah, I definitely noticed the price tag. But comparing that to the
> > > CMM300 half bridge, it is on par with the price and it is all
> integrated
> > > into one package which I would think would help reduce parasicitcs. I
> > guess
> > > this begs the question: what is the advantage of doing a half bridge
> > versus
> > > a full bridge? I would assume that it would be higher power but I am
> > still
> > > learning power electronics (I'm an RF engineer).
> > >
> > > I was also thinking that this part would kind of be one to grow
> > > into...basically I could start by running it way backed off (lower
> > voltage
> > > and current) and reduce the chance of destroying it. That way I could
> > learn
> > > with reduced risk of a costly mistake. Then once I am comfortable I can
> > > upgrade the power supply around it. Are these sound strategies to use
> or
> > do
> > > IGBTs like to be driven more toward their limits? Thanks for the
> response
> > > and insight!
> > >
> > > Jeremy
> > >
> > > On Monday, July 7, 2014, Steve Ward <steve.ward@xxxxxxxxx
> <javascript:;>>
> > wrote:
> > >
> > > > Jeremy,
> > > >
> > > > The APTGLQ200H120G part looks outstanding, and the price tag matches
> > > that.
> > > >
> > > > Assuming you use a good driver that switches the IGBTs right near the
> > > zero
> > > > crossing in primary current, they should be capable of driving a
> fairly
> > > > large tesla coil.  Id suggest operating within the 770A spec on the
> > > parts,
> > > > but at 600V thats quite powerful.  Provided they have good cooling,
> one
> > > of
> > > > these parts could probably drive a 5-10kW coil.
> > > >
> > > > There really isnt an "Fmax" for a part, its a continuous curve of
> > > > "operating current vs frequency".  Id suggest running at <100khz if
> you
> > > > really want to push the peak current level to the max, but they would
> > > still
> > > > work fine at less peak current and a higher frequency.
> > > >
> > > > Steve
> > > >
> > > >
> > > > On Sun, Jul 6, 2014 at 9:52 AM, Jeremy Gassmann <
> jeremyee78@xxxxxxxxx
> > <javascript:;>
> > > > <javascript:;>>
> > > > wrote:
> > > >
> > > > > Hello everyone. I am starting my journey on the DRSSTC road and
> have
> > > some
> > > > > questions regarding the heart of the system...the IGBT. There are a
> > lot
> > > > of
> > > > > options out there including half bridge packages and even full
> bridge
> > > > > packages. Looking at the data sheets shows a lot of characteristics
> > > such
> > > > as
> > > > > input and output capacitance, rise and fall times, and switching
> > > energy.
> > > > My
> > > > > question is: what are the most important factors to look at when
> > > > selecting
> > > > > a device? Obviously current and voltage ratings are very important
> > but
> > > > what
> > > > > about keeping capacitance as low as possible, etc? Are there any
> good
> > > > > documents out there the correlate these parameters to an fmax of
> the
> > > > > device?
> > > > >
> > > > > I did find a device that looks like it would be a good candidate:
> the
> > > > > Microsemi APTGLQ200H120G. It is a full bridge power module. Anyone
> > work
> > > > > with this device or know if there any reasons not to use it? Thank
> > you
> > > > all
> > > > > very much for the help!
> > > > >
> > > > > Jeremy
> > > > > _______________________________________________
> > > > > Tesla mailing list
> > > > > Tesla@xxxxxxxxxx <javascript:;> <javascript:;>
> > > > > http://www.pupman.com/mailman/listinfo/tesla
> > > > >
> > > > _______________________________________________
> > > > Tesla mailing list
> > > > Tesla@xxxxxxxxxx <javascript:;> <javascript:;>
> > > > http://www.pupman.com/mailman/listinfo/tesla
> > > >
> > > _______________________________________________
> > > Tesla mailing list
> > > Tesla@xxxxxxxxxx <javascript:;>
> > > http://www.pupman.com/mailman/listinfo/tesla
> > >
> > _______________________________________________
> > Tesla mailing list
> > Tesla@xxxxxxxxxx <javascript:;>
> > 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