Failing MOSFETs

[NoSmoke]

My PWM/MOSFET circuit for windmill load control was progressing nicely. It was driving fairly substantial test loads (about 6A at 120V fullwave rectified, but unfiltered, mains AC). The PWM signal to the MOSFET gate was clean as a bean (at 500Hz and a bit over 4V) and the MOSFET got barely warm (over at a range of duty factors). The previous PIcaxe startup issues I was having b/f were resolved and all was looking good until the MOSFET(s) starting failing (Drain shorted internally to Source). The fails occured (twice so far) immediately when the load power was applied as evidenced by the load (high wattage light bulbs) going immediately to full brightness. Both the Picaxe and driver were powered up at the time.

Heeding previous advice, I have attempted a schematic diagram (using DesignSpark PCB). On attempting to attach the file however, I find that the max allowable .bmp file is 19.6KB whereas my schematic is 3.5 MB! I suppose it could be trimmed down somewhat but at 19.6KB I doubt there would be much left (& .bmp seems to be the only file type DesignSpark will create that is acceptable for upload).

Anyhow, the MOSFET is a IRL640 (200V, 17A, logic level gate) and the driver is a MCP1407. The Picaxe/driver 5v supply is transformer isolated from the 120VAC supply.

Can anyone offer any suggestions as to why the MOSFET is failing? Maybe feeding the MOSFET unfiltered rectified AC is causing it grief (?).

Also, can anyone please tell me if a schematic pin can be attached to a shape (in DesignSpark PCB)??

Also, also, why does the forum have to go off line every day around 0300(?) GMT?

 

[MPep]

I guess you need to change your name now??!!

Download CutePDF from here. This installs as a printer. From DS-PCB you should be able to select another printer ("CutePDF") which then lets you save the drawing anywhere you want.

Forum goews off-line as it goes through back-up procedures and maintenance.

One 'general' design rule is to use a MOSFET with a Vds rating of at least twice the anticipated applied voltage.


Can't help much more, sorry.

 

[John West]

As I recall, your goal was to experiment and determine the minimum possible configuration of components. It looks like you've learned you might need to add a few more. Good luck, and don't forget the safety-glasses.

As a wild guess, I'd suggest the problem may be inductive spikes. If you can sort out the schematic and provide some good quality photos of your circuit, I'll try to provide a more complete response. You might want to try converting the BMP file to a JPEG. BMP's are a notoriously large file format.

 

[pilko]

Download an App. called Snagit. It alows you to capture anything on your screen, either whole screen,
part of a screen, or even a scrolling window.It saves the file as a JPG or you can open it in another
format.

 

[inglewoodpete]

...about 6A at 120V fullwave rectified, but unfiltered, mains AC

Ooer. A 200V (max) MOSFET is not much more than the applied voltage. Mains voltage is typically +- 10%. Less assume 132V RMS. Once rectified, the output voltage becomes Vrms x sqrt(2) or 187.4 volts. Add a little bit of noise when your 'fridge starts or stops and pop! Try a 400v MOSFET.


Use jpegs to save your circuit. MS Paint will convert a BMP to jpg.

 

[John West]

That's why I was thinking it might be inductive spikes, Pete. 200V max is way too close to the continuous AC peaks. Even the smallest of spikes would do it. No margin there for errors or fluctuations of any sort - at all. It reminds me of the 16V filter caps I found in a 15V circuit (designed by a physicist, not an EE.) I was busy replacing them because they kept blowing out. The physicist couldn't understand why.

I replaced them with 25V caps.

 

[Janne]

Like stated by others, voltage spikes from parasitic inductance are a highly likely candinate for cause of blowups. They can account for quite substantial amount of voltage. For example, in my wind MPPT controller project, I measured spikes of over 100V with 30V input on the switch node, without any dampening. With damper networks I was able to reduce the ringing to acceptable levels.

Using a higher voltage MOSFET is advisable, because of your small margin on the rated voltage. But also, it would be better to add dampening to the switch node as well, without it I'd hardly call it robust(and thus not really suitable to be brought anywhere near a windmill where electronics reliability is the absolute pririty, over component count...)

Scope the waveforms on your switch node, and on the fet gate and post them here along with your circuit schematic & photo of the layout for further advice.

 

[Goeytex]

No Schematic = Not Much Help

Simply convert your bmp to jpeg in MS Paint and post it. A photo of the board/ prototype would be nice too.

And please don't do what a lot of folks do here and post a "wishful thinking" or partial schematic instead
of what you are actually using. Make the schematic complete and accurate and include part numbers
for items like FETS, FET Drivers, diodes, regulators, etc.

Unfiltered rectified AC ? Is that the raw output of a bridge rectifier ?

If that is the case then the inductance and capacitance of the wiring alone can lead to spikes > 270v

 

[Goeytex]

Smoke,

I'm gonna let you in on a little secret. Its called an IGBT. (Insulated Gate Bipolar Transistor).
These are very rugged devices ... much more rugged than a MOSFET.

They can be:

1. Logic level. ( need very little current)
2. Rated over 600v
3. Are Internally Clamped at 370v - 420 V

These IGBTs are generally used to drive Automotive Ignition Coils. However they work
great in high voltage high current applications where inductive spikes can smoke an
unprotected MOSFET .

This is the one I used in a high energy ignition system.

ON Semiconductor Part. NGP15N41CL.

This is basically a 15 amp, Logic Level, 600V IGBT with a 410 volt Internal Clamp. (Near Bullet Proof)

Farnell has these in stock.

 

[NoSmoke]

Thanks gentlemen for the help.

MPep: No smoke resulted so I can keep my name for now (probably only because I immediately unplugged the 120 when the load bulbs started full on).

Based on yours and other's comments, it certainly looks like my 200V MOSFETS aren't up to it and that explains the problem. I realized the voltage was a bit close to the limit (187 nominal as pointed out by inglewoodpete) but in my naievety (where's the spell check here?) thought 200 would be enough. Expecting last night though that that may have been the problem, I ordered some 250V units (from Digi-key) but it looks like even that may not be enough.

I'll try them however but am wondering if maybe a cap and/or diode across GS would help dampen the inductive spikes (yes, it's raw outout from the bridge rectifier - my thinking was it will be 3ph rectified in actual operation which would smooth the bumps out).

The IGBT parts look interesting but Farnell doesn't seem to have a NA presence and they want about $30 for non-EU delivery and probably a week or two to get here but Digi-Key has 'em (although the part mentioned by Goeytex is a non-stock item). That's what I will try if my new MOSFET batch doesn't work.

For you eastern hemisphere dwellers who may not be familiar with Digi-Key BTW, they offer Fed-X next day door-step delivery for a flat 8 bucks which IMO is pretty sweet.

I've converted my crude attempt at schematic drawing to jpeg as suggested and have, I hope, successfully attached it.

Any other comments/suggestions welcomed.........

 

[Goeytex]

If you are considering an IGBT then plug that part number into http://findchips.com and you will see that Mouse
and Digikey also stock that particular one. Findchips is a great resource, especially for those of us in NA.

 

[NoSmoke]

Thanks for the findchips info. Checked it for the NGP15N41CL and they have it alright but they want $20 for FedX shipping vs $8 for Digikey. Digikey also has them but in 400 quantity - may be more than I need depending on how many others I manage to destroy.

BTW, the MOSFETs I just blew are made in China (harking back to a recent thread).

 

[Dippy]

A tricky one , and without some hands-on circuit analysis it's difficult to diagnose remotely.
And now you are in the world of high(ish)-voltage MOSFET switching... probably for the first time?
Now you've discovered that high-speed switching isn't quite so easy huh?

I would also go along with a pesky parasitic inductance as being a possible contributor.
Honest, it can come from anywhere and inexperienced design will make it worse.

Overshoot and undershoot on the driver output never helps and I notice you (apparently) haven't used the Data-Sheet recommended bypass caps for the driver.
"A 1.0 μF low ESR film capacitor and a 0.1 μF ceramic capacitor placed between pins 1, 8 and 4, 5 should be used. These capacitors should be placed close to the driver to minimized circuit board parasitics and provide a local source for the required current."

Obviously from the schematic we can't see the physical arrangement either and that is HUGELY important.

It doesn't take much track/trace/wire to 'create' a level of inductance that will result in a dirty great voltage spike at the gate and/or at the MOSFET junctions. And the faster the switching the worse it can get.
A big fat suitable rated cap between the Drain and Gnd/0V closely placed may help, but be aware of ripple values.
A zener between Gate and ground maybe necessary.
Or slow down the switching speed by increasing the value of the gate driver resistor.
I've said it many times before; MOSFET (and IGBT) driving is a compromise between heat and noise.

What you should have done was experiement with your circuit at , say, 24V and properly 'scoped the significant nodes.
It requires more than a 50p 'scope.
It doesn't take much over-voltage at Gate or Drain to punch-through and kill the device.
Welcome to the world of MOSFET switching!
Good luck.

 

[NoSmoke]

My new MOSFETs arrived today (man, that DigiKey is fast!), IPP600N25N3s (N-CH, 250V, 25A).

Hooked one up to existing circuit and............the lamp load immediately came on full brightness (same as what happened when the old ones blew) Took some resistance readings on the chip and they appeared normal. Tried another - same thing. Rechecked pinout on data sheet - same as other MOSFETS.

Just for the heck of it, tried reversing D & S and it works! WTF?? Did they send me P-CH chips or something? The label on the chip looks correct.

Dippy, I have a 1uF electrolytic cap from VDD to ground (my driver (MCP-1407) has a different pinout than the one shown on Goeytex's schematic - it is single O/P only). I know that's not exactly what is recommended but it's all I had available. As I mentioned, the PWM waveform looks v clean on my scope (I don't know if it's more than a "50p" scope). This all at low frequency (about the lowest the Picaxe will run PWM) of 500Hz. At higher frequencies the waveform becomes distorted/noisy. I guess my question is then, do I still require better bypass cap than the data sheet says at that low a fequency?

Forgot to add, I have added a cap between Drain and ground and will add a 5V zener between Gate and ground.

 

[MartinM57]

Just for the heck of it, tried reversing D & S and it works! WTF?? Did they send me P-CH chips or something? The label on the chip looks correct.

If you've reversed D and S on an N-channel FET so that S is more positive than D, then what about the (now) nicely forward biased "body" diode shown in the schematic of
http://www.infineon.com/dgdl/IPP_B_I_600N25N3+G+Rev2.2.pdf?folderId=db3a3043163797a6011637d4bae7003b&fileId=db3a3043243b5f17012496c9548d199c
? (hint: it's likely to be fully conducting).

Are you sure it's working?

 

[Dippy]

A perfect square-duty wave on the MOSFET gate terminal when driving with load?
That's very well done.

Martin, totally right.... and it comes on full with no PWM? Is that right?
Something's wrong.


"it's all I had available"
...mmm... not good , and you could have ordered the correct caps when placing order with Digikey maybe??

Good luck.

 

[Goeytex]

You do realize that your new MOSFETs are not "logic level" and need about 8v - 10 volts to drive properly ?
Remember .. Gate to Source Threshold Voltage (Now 4v) is when the FET just starts to conduct, not what it needs
to fully conduct.

And you will need some kind of filter on the output of the bridge rectifier if you expect any kind of decent switching.
Think about it ... you have a 120 hz half wave going to the load. Then you are switching that at on /off at 500hz
.... meaning the switched current signal will look like random garbage.

If you for some unknown reason do not want to filter the rectifier output then you would be better off
using an SCR.

If the FET s conducting with no gate signal, then I would suspect the possibility of improper grounding.
Try shorting the source and gate together and then powering up, if working properly the FET should
remain off.

 

[NoSmoke]

Well, I screwed up - made a wiring change to add a cap and diode across D-S and reversed the leads. That's why the new MOSFETs wouldn't work at first.

Goeytex, thanks for determining the new MOSFETs are not logic level gate - I can blame that one on DigiKey as I searched for logic level MOSFETs and the IPP600N came up. I should have checked the data sheet I guess.

Anyhow after rectifying my wiring mistake and installing a fresh MOSFET, it worked for a while and then the old problem with the Picaxe halting again appeared. Whilst fiddling with that, the MOSFET again croaked.

Maybe I'll try the IBGTs or SCRs next. Or maybe not - this thing is giving me indigestion.

 

[MPep]

Anyhow after rectifying my wiring mistake and installing a fresh MOSFET, it worked for a while and then the old problem with the Picaxe halting again appeared. Whilst fiddling with that, the MOSFET again croaked.

Do you have enough capacitance for the PICAXE supply rails? Picaxe shouldn't halt at all.
Reconfigure your MOSFET driver to supply 10 - 12V to the gate. Should work okay.

 

[Goeytex]

If this is a load dump for a windmill, then why are you switching unfiltered rectified 120 v ? Most dump DC from
the battery bank. Usually 12v, 24v or 48V dc .... ?

Before running off and getting an SCR or IGBT maybe you should rethink the minimalist approach of omitting a decent
filter capacitor. And you need to understand that board layout is critical when switching high voltage/high current
loads. It cannot be thrown together haphazard with little forethought for ground & power planes, trace widths
wire lengths ,etc.

If you want to try clamping/snubbing the max voltage across the FET then check out this link.

http://www.daycounter.com/Calculators/Snubbers/Snubber-Design-Calculator.phtml

 

[NoSmoke]

As I mentioned, I didn't bother filtering the test 120V as the operational circuit will be rectifed 3ph (from the wind mill alternator) which would smooth the DC substantially. Maybe though feeding the MOSFET with unfiltered DC is causing it some sort of problem (as well as generating a lot of electrical noise which seems to be why the Picaxe won't start when heavier loads are attached to the MOSFET).

This isn't a load dump circuit, it is a load control for the wind mill. A HAWT requires removal of the load in low wind speeds or it will not get up to speed when the wind speed increases i.e. it "stalls" and will only rotate slowly as the blades are not turning fast enough to "see" all or most of the wind passing through the swept area.(unless in v strong winds). This BTW happens m/l automatically if the mill is feeding a battery bank as the battery does not impose a load until "cut-in" is reached i.e. the generated voltage exceeds the battery voltage. Without the load control, it's kind of like trying start a car moving in 5th gear.

A further use of the circuit would be for "MPPT" (Maximum Power Point Tracking) in which the load is continually adjusted to extract the max amount of power at varying wind speeds.

MPep, I was hoping to eliminate the need for a second power supply circuit and therefore went with a logic level gate device (or so I thought in the latest go-around).

 

[MPep]

Where does the 5Vdc come from? If you are using a transformer to obtain this, then use a tranny to get 12Vdc, and use a LM7805 to run the PICAXE. This is effectively a 'standard' power supply, and one that should work okay in your situation. Then your MOSFET driver will work ok.

 

[Janne]

Even with 3-phase supply you will need filter caps, you seem to omit the impedance of the generator and the wiring.

I've built a fast switcher like this for just the same application, and It's not nearly as easy as it might seem. Noise problems, ringing etc. in the circuit all became a problem. If you're going to feed power from your controller to heating resistors, I'd suggest going with slow PWM (mayby 1-2s duty period), or with SCR's. The momentum of the generator will smooth out the small ripple the slow PWM would cause you. Then, after getting a bit more experience you could tackle the high speed one, if you still desire that. Fast switching is really only required for battery charging, to keep the voltage converting inductor to a sensible size.

Having a HAWT, you probably also know that the load must be maintained at all times on the generator. An application which requires the maxium reliability, not an application to start skipping on components!

Mayby have a look on this thread, to show what kind of problems I had with my battery charging design.. The heating controller was more troublesome due to the higher voltages involved. http://www.picaxeforum.co.uk/showthread.php?10986-Pb-switchmode-charger-discharger./page2

 

[NoSmoke]

MPep, My 5V power supply is a 12V transformer feeding a bridge rectifier and a 7805 with filter cpas on both sides. I also have a 5V cap on the breadboard.

The driver works fine - the problems arise when I plug in the 120V for the MOSFET load and/or the Picaxe won't start. If the MOSFET doesn't blow when the 120V is plugged in and the Picaxe starts, all is well. The circuit has never failed when it is actually operating.

Janne, I hear what you are saying about alternator impedance. I have dug up an 8,000 uF cap and connected it across the 120V rectifier o/p. Ripple is now v low under load. I would still like to know though what if any problems feeding the MOSFET rectified but unfiltered AC might be causing. In any case I have ordered some IGBTs so we will see if they are an improvement.

I understand your comment about switching frequency - it does not have to be v high but 500Hz seems to be about the lowest the Picaxe will produce so that's what I'm using.

I may have mentioned in a prior thread that I have mechanical blade pitch control so don't have to worry about load loss.

I will certainly take a look at your battery charger circuit. Thanks.

 

[Janne]

The pitch control is good, it will make it easier as you won't need to rely on the controller loading alone to keep the machine under control.
Slow PWM could be achieved on picaxe by using pause and pulsout commands in succession. For reading the speed of the machine, you could use pulsin to measure the period of the AC voltage coming from the generator.

 

[NoSmoke]

Janne, good suggestion re slow PWM - I will try it.

Looked at your battery charger project. Interesting - lower voltage than mine but much higher current. Was also interested to note your blew a MOSFET apparently just by connecting it to the battery bank with some sparking at the connection. I wonder if that's what is happening to me (that's when mine pop)?

 

[MPep]

MPep, My 5V power supply is a 12V transformer feeding a bridge rectifier and a 7805 with filter cpas on both sides. I also have a 5V cap on the breadboard.

The driver works fine - the problems arise when I plug in the 120V for the MOSFET load and/or the Picaxe won't start. If the MOSFET doesn't blow when the 120V is plugged in and the Picaxe starts, all is well. The circuit has never failed when it is actually operating.

What I meant was that if you are now using 'normal' MOSFETs (non logic level gates), then your gate drive voltage needs to go up! Or else, you'll be in the linear region, with all sorts of heating problems.

 

[NoSmoke]

MPep, sorry, I misunderstood the question. I have been using logic level gate devices with the exception of the last batch that Digikey listed as logic level but were not (as pointed out by Goeytex). The IGBTs I just ordered are also logic level (I hope).

 

[Goeytex]

If you ordered the IGBT's I suggested they will be logic level. Be sure to read the
datasheet. The gate current needs to be about 5ma so put a 1K resistor in series
with the gate. You need to use a heat sink just like with the FET.

 

[NoSmoke]

Goeytex, thanks for that info (another thing I didn't realize - I otherwise would have left the 20 Ohm in place). Do I still need a driver chip and if so will the FET driver there now work? If FedX works on remembrance day, I should know soon if I can make it work.

I've also assembled another much small breadboard with close component spacing and short wiring (except for the FET or IGBT and associated resistors & zener which are on a separate board with the bridge rect and filter cap). I'm hoping that will alliveate the Picaxe non-start problem (assuming noise pickup is causing it).

 

[Goeytex]

ALWAYS read the datasheet. Make it a habit. If you are not sure about something, ask questions
BEFORE you hook something up a toast it.

You don't really need the driver with this particular IGTB, However, I would leave it in
place to act as a buffer between the IGBT and the Picaxe. Your choice.

 

[MartinM57]

Worth putting a SERTXD("started", CR, LF) at the top of the PICAXE code or light a LED for 0.25s etc - then you can see if the PICAXE is constantly restarting or locking up completely..

 

[NoSmoke]

Goeytex, I did not order the IGBT you suggested as Digikey did not stock it. What I ordered instead was the STGB35N35LZ-1 ("IGBT CLAMPED 40A 176W I2PAK", 380V, logic level).

I appreciate what you mean by reading the data sheet (which I have done) but it is one thing to read it and another to understand what it all means. For example, I don't see anything that indicates the operational Gate current draw - it may well be there but not as far as I can tell. I do see a specification of Gate current of 1mA at the Gate threshold voltage of 1.6V typical but does that necessarily mean it would draw 5/1.6 * 1 = 3.1mA at 5V? (and I would therefore need a 5/.0031 = 1600 Ohm series resistor)?

I will leave the driver in place as you suggest.

MartinM57, I do have an LED in place and with this latest round of non-start problems the LED glows dimly and continuously as opposed to bright and flashing. So I guess that means the Picaxe is constantly restarting(?). In previous instances, the LED would not come on at all (although the download function would work).

Thanks again for eveyone's help.

 

[NoSmoke]

Gentlemen, I am pleased to report that the IGBT is installed and working fine. Everything keeping cool (maybe a little warmer with the IGBT), clean PWM waveform (@ 500Hz) and no Picaxe starting problems. This is with a 6A load at about 150 VDC (duty cycle limited to 80% to spare the light bulbs). Now to try 9A......

Edit:

Tried 9A load and everything OK except a larger heatsink is needed. Experimented a bit and removed the load power 8,000 uF capacitor with no apparent effect. Then however tried increasing PWM frequency from 500Hz to 5,000. The circuit operated for about a second before the IGBT shorted (good thing I bought three of them). I guess I'll stick to 500Hz - I'd like to experiment a bit more but am getting a bit tired of having to order new parts..........

 

[Dippy]

The characteristics of 'gate-off' time versus major current time are different with MOSFET versus IGBT.

There is a difference between average gate switching current and peak current.

Heating-up power from MOSFET is I x Rds , power from IGBT is I x Vce .
As a result of all of the above and assuming good component selection, IGBTs will tend to run hotter in a similar app. and will get worse as you turn the drive speed up.

If you are having failures due to peak current with IGBTs you should look into desaturation sensing.

I'm lost as to the current (no pun intended) state of your drive circuit. However, it appears your experience is on the light side, so I would recommend some reading-up on driving MOSFETs and IGBTs. International Rectifier have some great App Notes and some experience should reduce your failure rate.

A good design/layout/implementation with MOSFETs at this current level should have been successful but it sounds like the design and/or component selection has let it down , however, it sounds like you are getting closer to the desired goal so onwards and upwards .

 

[MPep]

Well done on getting to a point that your circuit doesn't damage itself!!

Further to Dippy's comments, from IR look up Application Note AN-983 IGBT Charateristics. Gives comparisons, drive requirements etc. Check here.
Most manufactureers will have App Notes about their devices, please look them up.

NoSmoke,
Would you please advise, with circuit diagram, how you are now driving the IGBT? Now that you have it working, this info may well be usefull to others.

 

[Goeytex]

I should have mentioned this but it is in the data sheet. It has to do with rise times and fall times of the IGBT. These will determine maximum switching frequency. IGBT's are generally not high speed switching devices. ( compared to MOSFETS) 5000 hz is way to fast for this IGBT. It was designed for driving ignition coils & fuel injection solenoids that will never see 5000 hz.

Why do you need (or want) 5000 hz ? With the cap removed that will just be slicing the voltage humps up into little pieces with the current pulses ranging from 0 to max. It will look like garbage (and work like garbage). But then again I already said that.

Even with an SCR and no CAP you want to switch so that you get at least 1 complete cycle for your lowest "DUTY" , meaning a switching speed of 120 / 10 or 12 HZ that will allow a duty of 0 to 100 with 10 duty steps of 10,20,30,40, etc.

@Smoke

Your determination to eliminate the filter capacitor contrary to the advice given here is gonna cost you in blown parts ... Leave it in and keep the switching speed as slow as is practical. Going faster is not gonna gain you anything anyway. I would actually drop it to about 100 hz (or the lowest possible with the Picaxe) to improve efficiency.

Note: The datasheet on the ST IGBT clearly states that there is an internal gate resistor of 1.5Kohms where in the other IGBT it was only 70 ohms. Thus my recommendation of if a 1K gate resistor for the other device. I have no experience with the new IGBT but I suspect that a gate resistor is not necessarily needed with a 5v drive signal.

 

[NoSmoke]

Dippy, thanks for the explanation of why the IGBT runs hotter. As to the current state, I have attached another schematic. I'll look into the "desaturation sensing" if I have further difficulties (at 250Hz which is what I am at now and is the lowest one can go with the Picaxe at 4MHz).

MPep, as mentioned another schematic is attached. As for driving the IGBT I started with the original MOSFET driver and, after that worked, tried it directly from the Picaxe, both with a 1K series gate resistor. That worked as well but, as noted, the circuit failed at 5,000Hz. I'd try it again at 5000 with the driver reinserted but don't want to blow any more parts and it would be an academic exercise anyhow. In any case, I think I'll leave the driver in as Goeytex recommends.

Goeytex, the trial at 5,000 Hz was just experimentation - it worked with the MOSFETs, as well as 50,000 (until they blew later on power-up) so I thought I'd try it with the IGBT. Thanks for the explanation of why it popped. I realize 5000 is not required for the operational circuit and, as noted above, reduced the Picaxe to 4MHz to get 250Hz PWM (the PWM wizard gives an out of range error if one tries to go lower but I'll try changing the code directly to see if 100 can be reached with the 28X2).

I am not at all determined to remove the filter cap (or any other part for that matter) - just trying things to see what works and what doesn't. Removed, it does give a completely scrambled waveform but I saw no problem with that in itself with a resistance only load. Of course, if it also causes component problems, that's a different matter. When the circuit is used with the wind mill 3ph o/p, the ripple will be much less, but, I suppose, the long lead-in wire and alternator inductance may add it's own crapulence so all in all the filter cap seems a good precaution.

I did note the IGBT 1.5K internal gate resistor but was (still am) not clear how that related to the requirement of the external resistor you suggested or, if required, how to calculate its value.

 

[eclectic]

@NoSmoke

I don't know how fast your 08M2 needs to run
for your main program.
However I've just tried this little program with an Oscilloscope on pin C.2

#picaxe 08M2
setfreq k31 ; k250 ; k500 ;m1 ;m2 ;m4;  try different freqs

pwmout  pwmdiv16, 2, 248, 498
stop

The frequency is very slow.

e

 

[NoSmoke]

Crap, spoke too soon. The IGBTs are now blowing on load power plug-in just like the MOSFETs. I wonder if this is related to the instantaneous 60 cycle line voltage at the time of contact (which may explain why the problem is intermittant). The first IGBT to blow had been in the circuit for a while and had undergone several power-up/off sequences but the replacement blew the first time (E found shorted to C in both cases). I don't know how to get around this (assuming the surge in voltage is causing the problem) except maybe for disconnecting the load as well as maybe jumpering E and C before it's plugged in.

E, thanks for the tip - I was not aware the Picaxes could run at such low frequencies (I'm actually using a 28X2).