Wind oriented buck converter component choice

[BigBreaker]

Looking for advice from someone with power electronics experience to help me find the right parts for some circuits I am sketching out.  The primary circuit is a buck converter for a 200V+ (wild) polyphase wind generator.  The circuit is intended to scale gracefully up to the DanB 5kw+ level where stator heating is problematic (and solved by voltage conversion).  The buck converter would take the retified AC and deliver a steady 24V or 48v to a battery bank.

So I am trying determine (1) the switching frequency for the buck, (2) switching with mosfets or IGBTs, and (3) sizing the needed capacitor bank.  The principle stumbling block has been understanding the spec sheets for the MOSFETs and IGBTs.  They list high current limits with low voltages and seemingly low current limits at higher voltages.  Some sheets also have power limits, but some do not.  What if I want to switch 40 amps at 200 volts? IE 8kw?  I know I need a lot of silicon to switch that much power, but how much has been tough to answer with the sheets.  With that answer in hand, though, it is not too hard to calculate conductance and switching losses as well as the capacitor needs.  My experience with mosfets is on bread boards, not power systems, so I could use some pointers!

Separately I am trying to figure out whether to focus on MOSFETs or IGBTs.  IGBTs seems more durable and efficient at lower switching speeds, which I like.  I get the current tail problem - it's not a show stopper.  MOSFETs allow for faster harder switching and smaller capacitor banks but they seem more voltage limited and sensitive.  There are fewer MOSFETs with acceptable margins over the intended voltage range - IE lots of 200v MOSFETs but fewer 400v+ MOSFETs.  What happens when they go to switch off and the stator inductance kicks in on top of a 200v peak? ouch.  Also multiple MOSFETs mean that if one turns on a bit faster than the others... again cascading failure.  Maybe I'm wrong it is seems like it's tough to kill an IGBT without reverse biasing it and that's not a problem on the DC side of a buck converter.

I have a neat solution to wind MPPT that does not require a timed digital circuit / microcontroller / etc...  The calculation actually happens in a transformer!  But that's a topic for the next diary.

Anyone have suggestions for my switching issue?

[bob g]

what you are contemplating will be the next major step for the diy'er windgenerator

in my opinion.

i asked numberous times on the old forum about the possibility of such a converter, but most folks shy away from anything that adds complexity even if it can be proven to be a good thing.  the old KISS principle at work i guess.

just like all laws/principles there are exceptions, again in my opinion.

now then for your project

in the past many i have talked to got all concerned with the timing of each mosfet/igbt/scr/triac (pick your poison) in a three phase setup. the problem being that popular belief dictates that in order to get control over each of the three phases each switch has to be triggered at specific and different times.

what one has here is battery charging, wherein the 3 phase is going to be converted to dc anyway so who cares about keeping each phase perfectly timed, like you would if you were converting power and needed to run a 3 phase motor.

the idea of a buck converter is interesting to me as well, however

you might look into a switch mode rectifier where the lower set of diodes are replaced with mosfets, and

all of the lower mosfets are triggered at the same time, this reduces the complexity  of the control circuit a bunch.

then the trigger circuit needs a feed back loop from the batteries, and sets a pwm

to the drivers for the mosfets. (basically)

one could also bring in a sense line (rpm/frequency) so that the mosfets remained off until the rotor reached a specified speed, which would make for easier low wind startup.

the switch mode rectifier uses the inductance of the alternator as the inductor

so you basically get a buck converter in the process.

this project will prove to be the best thing since sliced bread for alot of applications in the diy'er world.

this technology is being tested and proven in automotive alternators for the 42 volt upcoming standards.

also the big wind manufactures have moved over to this technology as well or derivitives of it.

the beauty of such technology allows the generator to run at very high voltages (relatively) with lower losses, allowing the converter to stepdown the voltage for charging. alot less heating being the result, less heat more efficiency or more power output, or a bit of both.

what i don't know is whether or not the air core alternator has enough inductance to provide the buck or if more inductors will be needed? i dunno

anyway good luck with the project, i will be watching your progress closely.

a good basic mx60 rated at 150amps and without all the other bells and whistles

specifically designed for an application would be very interesting to me.

bob g

[BigBreaker]

You raise many of the needs of the overall circuit and highlight the benefits.  I have these things in might as well.  That's all moot though, if the FETs and or capacitors cost $2k.  I think the capitors can come in around 100mF and at that size the bullet-proof 60,000 hour electrolytics are reasonably priced (~$100).  Again, though, it's the power FETs that I don't understand.

I was avoiding synchonous rectification for the time being.  High voltage AC doesn't take as big a hit from plain-jane diodes as the low voltage stuff.  Again... KISS at work.  Diodes can be over sized, cheap and will take abuse.  You can put them on the turbine, either up top or at the tower base, and run rippled DC to your convertor on two wires.  Again higher voltages are your friend - that cable doesn't have to be huge.

Here's a sample IGBT I am looking at for the project:

http://www.fairchildsemi.com/ds/FG/FGH50N6S2.pdf

It's a beast and it costs less than $9 a piece at Digikey.  At that price you can buy a lot.

Again, though, the spec sheet says 600 volts at 75 amps, IE a beast but it also says 463W max (at least on digikey).  My guess is switching versus continuous is the difference.  For the convertor, it's basically continuous.  We aren't flash a strobe here.

Also if 463 watts is the limit, then I'd want 10 or 20 in parallel.  That may require a huge gate driving circuit - haven't checked the current pulse yet - and also a big issue with ensuring the gates switch on near simultaneously.  I guess the IGBT take a short term current pulse pretty well but I have read that they can have problems working together.  Perhaps an SCR in series with the IGBT guaranteeing the zero crossing?

[BigBreaker]

You make mention of a cut in mechanism...  Yes!  Otherwise a buck will BADLY stall the turbine at start-up.  That would be a part of the MPPT logic, a topic for another day.

[Flux]

At 200v you are still within the mosfet range but probably you are more within the region of IGBT's. The high voltage mosfets have volt drops similar to the IGBT's. Below 150v mosfets would be better.

Unless you are an expert in high frequency layouts I would keep frequency low ( 30k ish).

These switches have a voltage limit which is absolute and any ringing or spikes must stay within the limit so don't try 200v devices. There is a thermal limit on current usually dictated by the package, but generally you can exceed that for short pulse durations and the safe operating area curves should give you that, but again don't think you can get anywhere near the makers claimed figures especially with TO220 things.

The power dissipation is a continuous dissipation in linear mode and you should not be working other than switched. As long as your losses within the device are within the dissipation figure you needn't worry about that one. In a perfect switch if you have a 200v supply and switch 100A you should dissipate nothing, not 20kW.

Capacitors will mainly be determined by the ripple current, you don't need that much capacitance with a 3 phase rectified supply. Far more important will be low impedance pulse capacitors directly on the devices.

You are very brave to go in at such a high power level as a first attempt. I haven't ventured beyond a 1kW machine at 100v as a buck converter, but I have dealt with bigger alternators with a boost converter, where the converter only handles a few hundred watts in low winds and the machine is matched direct for high winds.

There is a world of difference between playing with things on breadboards and the real power stuff. Ninety % of the issues relate to layout and coping with stray inductance and none of the mickey mouse stuff on most web sites even gives you a clue what to do at power level.

Go to international rectifier site, they have quite a lot of useful stuff on layout and gate drivers.

Flux

[altosack]

Hello BigBreaker,

I'm not a power electronics guru, but I'm an engineer who's been studying this for a while, and one of the things that really confused me at first about the data sheets of MOSFETs and IGBTs is the "power" rating.

This is not the power rating of the circuit, but the power that the device itself (the IGBT) will dissipate (i.e., the lost efficiency of the IGBT in the circuit). So, calculate I^2R and the switching losses (for starters), and then that should be some number less than half of the power rating (at operating temperature, not ambient; do not forget to derate for temperature !).

By the way, I haven't really looked at your IGBT datasheet in detail yet, but the front cover blurb of 40A @ 100kHz and 25A @ 200kHz really looks excellent. Thanks for the heads up on this device; it may be what I've been looking for !

Best of Luck,

Dave

[bob g]

whatever the outcome of your project i hope it grows legs,, long legs!

this subject is something i have been thinking about and researching for a very long time, and i hope folks like Flux and others band together to see one of these things to fruition.

we know in theory it will work, we know that manufactures build these things every day, we know where to get the parts, so now it comes down to learning how to get the bits to play together successfully.

i recently purchased this book

http://cgi.ebay.com/POWER-ELECTRONICS-w-CD-LIST-125-95-Cond-New_W0QQitemZ380033637938QQihZ025QQcateg
oryZ2228QQssPageNameZWDVWQQrdZ1QQcmdZViewItem

for about half what they want for it now, but the point being it is a very good design book laying out circuit design, component selection and most important

circuit board layout (stressing issues with stray capacitance/inductance etc)

if anyone knows of another very  good text on power electronics, i for one would like to hear about it.

i look at some of the igbt's surplus on ebay, and one can get some really high power units pretty reasonably.

not quite to the point of trying my hand at putting together a switch mode recifier myself just yet,, but i am thinking i soon will give it a try and see how much fire and smoke i can get per dollar invested

who knows i might get lucky and get something together that works, aside from the cost of parts the labor involved is far lighter than building towers and lifting generators.

one thing i am thinking is the use of an automotive voltage regulator, they operate on a pwm scheme (and are readily available quite cheaply even new) they can provide several amps of power in a range that might be useful as a trigger for a driver circuit? they have the sense wire for feedback and can control charging voltage as tightly as needed (some being adjustable as well)

so that leaves one with the design of the driver circuit of which some are all on chip from the manufacture reducing the need for alot of design and board layout concerns.

if you get the driver worked out, then using the igbt's should be pretty easy

most of the larger high power units are remote mounted with a twisted pair wires to

provide access to the gate. i see them used that way on other applications so it would appear that the stray capacitance/inductance issues is not much an issue for the igbt if used in this manner.

so i guess at this point it is the careful design of the driver circuit and its board so it can run stable and reliable?

i have the high side rectifiers 275amp @ 300 volts and some very large heat sinks

to bolt them and the low side igbt's to. just haven't bought any igbt's yet.

recently on ebay there was a new set of igbt's set up with the driver boards

everything complete that would have been nice,, actually their were two of them

and they were setup for 3 phase, new with instruction manual etc. they originally were made for a ac to dc converter of some sort. iirc they were 1200volt @ 300amps the two went together for 135 bucks,, i missed the deadline

personally i am not afraid of the power levels required as i work with 440/480 3 phase regularly as it is.

i like this project

bob g

[BigBreaker]

Don't wait for e-bay.  Take a look at this:

http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=FGH50N6S2-ND

$8.24 from digikey in small amounts, cheaper for 10+.  You seem to know more about this than I do.  Please plow ahead!

[BigBreaker]

Find it on digikey here:

http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=FGH50N6S2-ND

$8.24 per unit.

And thank you so much for the power explanation!!!  Flux aludes to the practical limitations of shedding that much heat but I'm not too fussed.  I'm a bit of an PC overclocking enthusiast and I'm used to 100 watts+ with delta Ts lower than 20 degress C.

[BigBreaker]

Gotcha on the power specs.  That's great news.

I'm looking at this IGBT:

http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=FGH50N6S2-ND

It can take 600v, which is about three times the target voltage.  I'm thinking about putting a cap behind the rectifiers but infront of the buck IGBTs to avoid spiking their Vce and killing the devices.  Overkill is good.

Ripple isn't a concern at all on the low voltage side of the buck.  The buck is driving a battery after all and the copper would be short and thick.  My thinking was to go quite low on the switching frequency, perhaps 5k hertz.  If I wanted to get fancy, the switching frequency could be variable based on load or via feedback.  100k hertz seems totally unnecessary.

I'll check the international rectifier site again.  I checked it a while ago and found it useful but I know a lot more now.  I should check back.  One question is of circuit topography.  With IGBTs do I pull the low side of the cap to ground or raise the high side to 200v?  It's the practical stuff that takes all the time.  The symbolic schematics are easy by contrast.

5k watts is WAY too ambitious for a first try, but I'm just interested in the circuit and selecting components that could reach up to those power levels.  The component count isn't very high.  My goal is to post a schematic and parts list from digikey.  The electronics gurus can pick it apart and if no obvious problems are found, perhaps it's worth a try?

[TheCasualTraveler]

     ""this subject is something i have been thinking about and researching for a very long time, and i hope folks like Flux and others band together to see one of these things to fruition.""

      If someone puts together a kit, like Ghurd did the dump controller, I'll sign up. I only know very basic electronics but I can solder a kit together.

[joestue]

Stay away from mosfets at that voltage.

Frequency is not really an issue here, because efficiency isn't either, unless you want to market this in the future, then it is.

Focus on getting the switching circuitry functioning first, then redesign the inductor.

Active snubbers are necessary above 2-4KW, you don't want 400 watts of recoverable energy slowly destroying the igbt and radiating EMI.

Concerning filters, the incoming dc is subject the same filter requirement as the output, (it must store enough energy) otherwise the inductance of the supply will fry the switch, and/or and cause non desired current waveforms.

Interleaved converters, which will also provide redundancy, will reduce the input and output filter requirement by quite literally, 1/n converters.

I would say start with an IGBT rated for twice the peak current expected at the point of turn off, modeled in spice or whatever,

But to tell you the truth, the way I designed my 2400 watt buck regulator was to run it with one out of the (planned) 4 mosfets, and one out of the 4 diodes as close to the point at which the previous one(s) exploded, then back off 30%, run it for a few days, cycling the power on an automatic timer, with the output feeding an inverter back into the supply transformers, monitoring the waveforms every few hours, waiting for something to change.

After that, put the other switches back in, and do the same again, but this time use non-destructive testing.

[BigBreaker]

I got that sense on mosfets above 200 volts or so (rated) and so I was really leaning towards IGBT.

Poly-phase buck conversion was one of the enhancements I had in mind down the road, but perhaps it's necessary sooner.

I anticipated that the input needed filtering.  A capacitor would be necessary to absorb the inductance generated voltage spike on switch off.  My sense is that those caps would be much smaller than the caps needed for the buck output side.

I need to learn more about snubbers.  I "skipped" over them in my research, thinking that I could forgo that efficiency until later.

Thank you for all your advice and help.

[joestue]

It is not difficult to build active snubbers for either turn off or turn on losses.

both is a challenge.

It is a lot less effort to wind an efficient inductor for a 1-2 kilowatt regulator, than a 5-10KW... besides, unless you want to use copper foil, the wire needs to be 20ga or smaller, and who wants to wind a coil with 50 wires in parallel, then realize it needs to be a little bigger...

[commanda]

http://www.fieldlines.com/story/2007/11/25/22656/782

Read this article again.

Then find an off the shelf unit of enough watts (or several in parallel) and figure out how to control the output voltage with a control signal.

Rectify the output of your wind turbine to dc, and feed it to line input. They work quite happily with dc input. You will need an over-voltage crowbar circuit in case it all goes wrong.

Meanwell make one which is 48 volt and 50 amp output. SCP-2K4

Amanda

[Flux]

Polyphase would reduce the ripple current requirements of the capacitors. For that size I would probably try it 2 phase, more phases would add more complexity with balancing the phase currents.

I think you will do better with the IGBT connected to the 200v dc rail and use high side drivers. These are a bit of a pig but so is messing about chopping the ground line at this sort of level.

The most critical area is the loop on the dc side directly at the switch, you need that to be very low inductance with a ground plane and lots of polypropylene pulse capacitors keeping the loop as short as possible. Ripple is no issue but spikes from the wiring inductance are. Copper bars and batteries might just as well be inductors as far as spikes are concerned, their low dc resistance counts for nothing in this case.

I didn't fancy pcb layouts even with the currents of my modest converter. I used a heat sink as the ground plane and mounted copper bars to this, insulated with silpad, then connected the mosfets and freewheel diodes to the copper bars with the capacitors across the top in the shortest possible route.

I used HCPL3120 opto drivers, which lets me separate the logic ground from the power ground, but the supply for the 15v to these devices needs extremely low capacitance to prevent dv/dt spikes going back into the control board and playing havoc ( especially with a 2917 tacho chip that I use for speed reference).

Flux

[BigBreaker]

Off the shelf?  Where's the fun in that?

I do have a bunch of spare PC power supplies.  Perhaps I should do some exploratory surgery on them to learn more about the practice of putting these together.  I was aware they they are happy with DC.  It all goes through the chopper anyhow.

I was shooting for something user servicerable in my design, with a relatively low part count and reliability over extreme efficiency.  For an expert like you the Weidmuller unit's design and function must be obvious.

For me, I would need days of study and probably a half dozen near fatal shocks to figure out all the details of a commercial SMPS.  I'm trying to puzzle my way from ground up.  Perhaps once I can design my own, I will come to the same conclusion - buy off the shelf!

[oztules]

Hmmm,

At 1500 dollars (AU), I think 10 old psu supplies would do the same thing, although a small efficiency loss due to power factor with the half bridge psu's would be present.

hmmm..... perhaps with the rectified three phase, (was thinking lumpy single phase) power factor may not be a problem at all......

A change in caps and switch transistors will move the input range window to 50-400v ( full power not available until 140v with voltage doubler switch off, 70v with doubler on) and very cheaply as well. This range should be wide enough to allow for a HV dump scheme. The Meanwell has an embarassingly small window on the input side.

Peter (dinges) did an interesting diary on these here http://www.fieldlines.com/story/2007/10/18/16953/116 and where he uses it as the power source for a spot welder here: :http://www.fieldlines.com/story/2008/2/26/15327/2668

Cheap (free) simple to convert, and by weight of numbers very powerful. Also a good bit of redundancy there as well, It is unlikely for them all to fail at once and have the mill runaway. (in fact once modified and set, they are next to impossible to kill... o/load, short circuit etc.

The noise of the 10 fans may be off putting though.

These things have stopped me (thankfully) from developing larger power inverters, as they are so adaptable and easy to utilise.

Even on this little island, I found six old computer psu's at the tip the other day. They must be everywhere in the real world out there, but you could squeeze a few kilowatts out of these six at a pinch.

Another cheap inverter source would probably be the cheap inverter welders that seem to have come on the market in recent years, I am sure these would be easier to convert than starting from scratch, and with the windmill duty cycle would probably hold up fairly well, and  have power factor correction built in.

In reality, a day or two could be spent on a reliable system as described above, or an indeterminate amount of time developing a stand alone biggy..... which if it fails in operation could spell disaster for the mill, this is not bullet proof like the booster system Flux has described previously.

It is certainly not as sexy as a home brew biggun, but is probably a much simpler way to go, and is scaleable to any size at all, just keep adding like batteries..

.......oztules

[bob g]

Amanda:

well bless your little pea pickin heart!

i have been studying the buck converters for a while now, bought books, researched all i can find on them, etc.

all of which has been mostly in vain, because either they are heavy on theory and light on practical application or heavy on practical application but made for bucking 12volts down to 9 volts or whatever at milliamps for some ultralow power app's.

then i read you post, and it lit up a bulb in my head!

why not modify an existing converter? hmmm? novel concept!

i have no issues with hacking an existing product!

i also have no desire to reinvent the wheel!

so then i remembered a pile of UPS' units that i parted for the power transformers

and remembered a seperate board that was mounted inside.

they are converters but likely boost converters rather than buck, but thats ok isn't it? Basically the components are common just arranged differently?

the converters look to be capable of handling maybe 20-30amps as the inductors are toroid and have very heavy square section copper coils on them (maybe .125" or so)

perhaps i should fetch them and do a reverse engineering project and see how they are laid out, and how they compare to the circuit designs put forth in my power electronic's book.

maybe they can be reconfigured to buck from boost, maybe they are dual purpose buck/boost already?

it sure would be a kick in the butt to find out i have had them in my possesion for a number of years and didn't know it!

maybe i am totally on the wrong track? but you got me to thinking.

thanks

bob g

[bob g]

oztules:

your comment on using multiple psu from old puters is interesting

having a background in ibm's old microchannel puters models 50z thru model 95

the model 60 and 80's had a hurcin' big power supply in them.

they were built to take input voltages from 85 to 260vac from 47 to ?hz (i can't remember the upper limit. and were transient protected for up to 8kvolt as well.

i used to have a bunch of them as well, oh well

but the other makes atx units are widely available around here for maybe a couple bucks each.

i would think your idea of ganging them up would be very interesting, one could disable the fans and use on fan if they were all placed in one case with a little planning.

i could see a rack mounted system with hmmm? (how many do i need for 150amps output?

maybe 30? lol

might need a squirrel cage blower to ventilate them?

would be pretty cool though.

i gotta give that some thought

some serious thought.

redundancy is a good thing as well

then the question becomes

if one psu is around 98% efficient in conversion, how efficient is a bank of 30 in

parallel?  should still be 98%? without inclusion of the blower motor draw

bob g

[joestue]

They aren't 98% efficient, that's why they have a fan.

The output diode rectifier alone predicts <90%

The older power supplies will be a lot easier to parallel, but you will still have to rebuild the control loops. Quite a few people have tried to parallel ATX types, without rebuilding the control loops, but not many have got more than a few working at one time.

[oztules]

Well Bob,

This is not armchair talk, I have built about 35 of the 36v traction chargers. These charged 6x T105 batteries in floor scrubbing machines 7 days a week.

These used three AT power supplies, modified like the ones Dinges did in each charger box. I used the original psu cases and fans and they bolted into a custom made case (with meter and control card). From what I understand, they are still in use (I'm retired now). (We bought 200 computer psu's including the computer cases when the AT was superceded by the ATX for $5.00 each.

Here I have found a box designed for three of those psu. They bolt straight in after conversion and away we go. This was for 36v 15A but could just as easily been 12v 45A with a quick rewire.

Also built some using 6 psu units for 30A@43v.. or 1.2KW. for the bigger Trojans (L16's).

 I have designed and built my own units for this kind of power, but for the time and effort and in my case... reliability, ( I must have used up a weeks supply of International Rectifiers output capacity of FETS just testing the dammed things. The psu is a better answer....it takes longer to build a decent box to put it in than the conversion.

And if you do blow one up, toss it and dumpster dive for another (or fix it which is easy as well when you get the hang of them.)

When you have modified them as described, they are difficult to kill, can be put in series or paralled  it makes no difference to them. BUT you must isolate the circuit board from the case in order to use them in multiples. This can be as simple as a rubber "O" ring under the tracks where the screws hold them to the psu case.

150 amps would be easily achieved with 10 units in parallel. window of 140- 260v ac or using voltage doubler (built in) from 75v-140 ac There is usually a MOV to help protect against overvoltage as well... if feeding dc then X 1.414

Max DC at 10A is probably about 25v. At 12v (13.7 etc) should get 20A if you want to push them, better at 15A.

Remember to change the caps to a higher voltage on the secondary if using over 15v on the output.

The output diodes get warmer than the switchers. Efficiency should be same as a normal psu. I don't know what that is, but I assume 85-90%. Below 150w, you can probably design passive cooling for the little critters.

they really do work.

[BigBreaker]

Both buck and boost use an inductor - capacitor pair, but I think boost uses the inductor as the main power storage element and buck uses the capacitor.  That would mean swapping out some high power components representing a material fraction of the device's cost.

Also the switching element, probably MOSFETs, are definitely in a different part of the circuit: parallel with the main loop versus series.  The voltage regulation, if any, would also need to be changed.

[BigBreaker]

Ok.  I have the basic circuit mocked up in SPICE.  I haven't used a schematic level model before, only individual device models.  It's definitely a learning curve and I haven't been able to run any simulations yet.  I used the "discrete" libraries, which seem to be idealized components, but the part parameters like capacitance, resistance and inductance are not showing up.  My software is a professional version designed to do a lot more than I am asking of it which makes this all tougher to do.

I have settled on a two phase IGBT system for switching the high voltage side into the low voltage battery charging stage.  This will guarantee each IGBT a significant recovery time which greatly increases the average current it can handle and reduces device stress.

I read up on snubbers, which are actually pretty simple.  I had capacitors in many of the right places for snubbing but my intuition wasn't perfect.  I'll probably need some more work in this area.

For the spice model, I think I'm going to use simple clockless conversion logic - switching on the IGBT when the buck capacitor voltage drops and switching it off when the voltage hits a certain limit.  A series inductor moderates the current into the battery.

Once the voltage conversion is simulating nicely I'll layout traces for a PCB and post my parts list from digikey.  As is the convertor will stall a mill badly.  The next step is to let the mill voltage (RPMs) run up with additional logic but one step at a time...

[BigBreaker]

Ok, most spice models I have found top out at 10 volts Vce which is a fair bit away from the 120 volts+ I am trying to simulate.  I thought my spice package was broken or my models were wrong but it was a capped out Ve that made everything look wrong.

Anyone have a high voltage spice model handy for an IGBT?