Igbt <-> mosfet

[eccerr0r]

I've played fast and loose with components for the heck of it (and mainly to use the stock I have) but I have a new substitution to try in order to not have to order components.

I'm going to try to 1:1 replace IGBTs with MOSFETs...

The current carrying capacity of the MOSFETs are not going to be as good as the IGBTs, but the IGBTs seem to be quite overspecced to begin with, so maybe it's not so bad. I'm going to derate other components like fuses for the time being to save the MOSFETs.

Any other things I should look out or is this foolhardy and all I'll get is ?

Hopefully will get some sign of life from the grid tie inverter until I can order bona fide IGBTs. And theoretically should be able to determine if it has a dead time violation that caused the IGBTs to fail to begin with, though I feel like it was more likely lightning...

[CapLeaker]

Re: Igbt &lt;-&gt; mosfet

I am running a grid tie inverter for 10 years now and it never missed a day. Lightning is no problem for anyone that has proper grounding installed.

I’ve seen folks changing out Mosfets for IGBT’s, but you wanna do IGBT’s for Mosfets. Never seen someone do this, but I think while it could be done there are certain things that do come in play. Heat of the Mosfets is one. The other one would be the gate drive resistor. While IGBTs have a higher gate drive resistor the Mosfet have a lower gate drive resistor. However we don’t know what Mosfet you try to sub what IGBT for. I guess there are more details needed to get a better answer.

[eccerr0r]

Re: Igbt &lt;-&gt; mosfet

I'm pretty sure this GTI was not properly grounded as it's an ad hoc inverter not to mention it's "CCS". The "CCS" aspect may be sufficient cause for failure to begin with. The failed IGBTs still look intact (one with all pins shorted, other with C-E short) so it's possible it's not lightning too... *shrug* no clue other than CCS syndrome.

What kind of stuff was changed when someone swapped in IGBTs for MOSFETs?

I was looking at the drive requirements for the two and they were quite similar but the voltage needed for IGBTs were somewhat higher - which makes sense due to needing to also overcome the drop in the BJT. I'm wondering if I stick in a few diode drops but have to do this carefully as it'd affect bother up and down, perhaps inverse parallel diodes.

I haven't decided on a temporary replacement yet, have to worry about voltage and current, I don't have anything close at all so I will have to put in an order at some point.

[eccerr0r]

Re: Igbt &lt;-&gt; mosfet

Oh I chickened out sorta. I found some old IGBTs but they're nowhere near the ideal specs... Soldered them in and ..well.. it looks like the GTI is alive! Well, not sure, hard to tell, need to see my grid power consumption and see if it changed any with it turned on, but for once my solar panels may be making their pay...

It's unfortunate my Kill-A-Watt doesn't deal with back power, but I think I have an idea on how to deal with that...

---

YEP IT'S WORKING! Now I need to go order real IGBTs to do a proper fix!

[sam_sam_sam]

Re: Igbt &lt;-&gt; mosfet

I am glad that you found a solution to your dilemma with this repair

[sam_sam_sam]

Re: Igbt &lt;-&gt; mosfet

It's unfortunate my Kill-A-Watt doesn't deal with back power, but I think I have an idea on how to deal with that...

Did you find a solution to this dilemma with what you want to do here

[eccerr0r]

Re: Igbt &lt;-&gt; mosfet

Actually I leave that as an exercise to the reader

The hint: if you don't like where you're at, define your own zero!

[eccerr0r]

Re: Igbt &lt;-&gt; mosfet

oh. I think I figured out how this thing failed while I was perusing the web.
This GTI is MPPT and doesn't know when to quit. So if MPPT exceeds the design limits, so be it, exceed the design limit.

MPPT of some batteries will far, far exceed the design limits - so I think someone tried using a battery instead of solar panels.

This is sort of a let down though not totally unexpected. It was meant to be MPPT and anti-islanding after all. Really need a regular inverter for that though it's a shame this can't be double duty...at least without a hack...

And may well cut off MPPT at its design limits. (and even less with my ghetto fix, then again it was meant to be a band-aid until I get real componets.)

[CapLeaker]

Re: Igbt &lt;-&gt; mosfet

Or someone put too many panels on a string.

[eccerr0r]

Re: Igbt &lt;-&gt; mosfet

Possible, but unlikely... would need to get a lot of panels whereas a LiFePO4 pack could wreck havoc.

[redwire]

Re: Igbt &lt;-&gt; mosfet

Mosfets are fast switching compared to IGBT's, which can give new grief if the design has really time deadtimes or compensation added for that. They can turn the IGBT on in advance, while turning another off.

Another aspect is HV mosfets have high RDS(on) and at some point IGBT's end up more efficient. They have a huge 1.7V VCE drop but once a mosfet is above this, it's cheaper to use IGBT's. IGBT's are also much tougher when partially on, if your inverter has poor gate drive.

Not sure what voltage/current this inverter is or the old part numbers.

Mains transients can be huge at Cat. IV that's why many smart meter fires were happening, they couldn't take the 8kV+ spike. It's worse if you are at the end of a long run of distribution power lines and/or near the pole pig.
Utilities are putting MOV's on the pole pig input bushing.

I think I would use latest generation IGBT's and their voltage ratings are up too. Would have to do an autopsy on the old IGBT's top see why they failed.
Apparently it's very common for companies to purchase raw IGBT dies and do your own packaging/bonding wires.

[eccerr0r]

Re: Igbt &lt;-&gt; mosfet

Indeed I was concerned the IGBT would be slower than the mosfet. At the very least, coupled with the transition time and threshold, the MOSFET will turn on sooner. Nevermind turn off, these IGBTs are dastardly slow compared to even MOSFETs which have ways of mitigating shutoff. Was hoping there was enough dead time designed in for this so that they could sub in whatever part was cheaper at the time of manufacture.

So far the "baby" TO220 IGBT is surviving at 130W or so tie-in. The remaining IGBTs that weren't fried are rated for 2.5x the voltage and 6x the current of the ones I subbed so I'm definitely not going to get anywhere near the original design target. Luckily power is fairly stable around here...

New hopefully exact match IGBTs are coming through the mail...and hope I don't get baby IGBTs in TO-247 packages...

[eccerr0r]

Re: Igbt &lt;-&gt; mosfet

Well, on a bright sunny day today, the baby IGBTs blew... LOL
I put in MOSFETs this time, still nowhere near the current capacity of the original IGBTs but a bit ... bit closer. Unfortunately I simply do not have high current devices...

... and so far it's holding up at least for now... Let's see how these work.

[CapLeaker]

Re: Igbt &lt;-&gt; mosfet

What were the IGBT’s that blew and what mosfets are you going to use as a temporary replacement?

[eccerr0r]

Re: Igbt &lt;-&gt; mosfet

I did not want to give part numbers as I wanted this to be completely theoretical...

but if you insist...

G60N100 was the original big honking IGBT
RCP10N40 was the baby IGBT that died, well of course it's way too small, albeit I was using 1/6 the rated power of the whole device before it failed which incidentally is 1/6 the current rating of the original...
FDFP18N50 is the MOSFET in now since that's what I have on hand. Going to see if the sun kills them tomorrow.

I am not going to bother trying: 75N06 ... IRF640 ....

I'm thinking I should buy some more FDFP18N50's too....seems like a good MOSFET to have on hand.

[CapLeaker]

Re: Igbt &lt;-&gt; mosfet

Well… I got curious when the these baby IGBT’s blew again.

[eccerr0r]

Re: Igbt &lt;-&gt; mosfet

I thought the RCP10N40 was a really obsolete device so no loss... probably could be replaced by a modern mosfet with nothing but upside ...

[eccerr0r]

Re: Igbt &lt;-&gt; mosfet

The 18N50s survived one day, peak current was 1.3A ...
still waiting on new G60N1000s...

[CapLeaker]

Re: Igbt &lt;-&gt; mosfet

Interesting it survived that long. What’s the grid tie inverter brand? Any pics?

[eccerr0r]

Re: Igbt &lt;-&gt; mosfet

Finally got replacements from China.

I was a bit suspicious about the picture of the IGBTs that I bought off of fleabay but now I know why the photos were suspicious. I thought I was buying Fairchild branded units (which were the originals) but I got post merger On Semi units...
I think they look good enough... at least I hope I didn't get fakes.

Going to wait for a rainy day to put them in, the FDPF18N50's are still working however. But I think I'd rather have the G60N100 IGBTs installed instead of the tiny MOSFETs (or the tiny IGBTs), even better when the G60N100s at least have a metal back plate whereas the FDPF18N50s are plastic all around. I hope that the noise emission also goes down with the slower IGBTs, not to mention that they will be balanced with the two remaining working IGBTs.

Just so suspicious the old IGBTs had no failure marks on them, Still look fine, just shorted all the way.

While we're at it, MOSFETs are easy to parallel unlike BJTs. However, how about IGBTs? On the datasheet it says IGBTs are "easy to parallel" but is it?

(as an aside, I happened to have gotten the FDPF18N50s from the same seller as the G60N100s.... probably by coincidence. Hoping that if the 18N50's lasted these few days as temporary fixes, the G60N100s should also last a while...)