Electronics > Repair

Induction cooktop repair - IGBT selection

(1/2) > >>

My induction cooktop quit working, blew a breaker and an internal board-mount fuse. It has two Toshiba GT40QR21 IGBTs in parallel driving the coil. I desoldered both, one shows a short across all three terminals, I can measure the body diode on the other one. The Toshiba IGBT seems to be an obsolete part, so I looked on DK for a potential replacement. There aren't many TO3-P IGBTs available, so I filtered by Vce(max) > 1200V and went through the data sheets. I'm not familiar with IGBTs, so should I reject parts that have too high an input capacitance compared to the GT40QR21? The GT40QR21 has a Cies = 1500 pF, most of the ones on DK have Cies 2-3 times that. The gate driver on the board appears to be a discrete design with a 2SC8050 and 2SC8550.

Current candidates: RJH1CV7DPK, IXDH20N120D1

Any help is appreciated, thanks!

Usually higher gate capacitance means that they'll be rated for higher collector current.
Of course it also means higher gate drive current or longer switching times which mean more heat.

At double the capacitance it may be worth reverse engineering the gate drive circuit to evaluate it or if you have an isolation transformer or differential probe comparing the gate switching times with dummy capacitors.

Is it? They have some in stock at Mouser... give or take the parenthesis-junk they have tacked onto there.

Sort by gate charge, it's really what matters.  Uh, hmm, bizarre, Toshiba didn't even record that parameter...

Cies tends to be a poor representative of overall performance.  The missing part is Cres, the cause of the gate voltage Miller plateau when switching (which I'd point to a figure for, but again, they don't rate charge on this thing..).  The combination of Cies swinging ~10V, and Cres swinging ~100s V, defines total gate charge, and therefore speed at a given R_G.

It doesn't look too bad, as they rate switching time at R_G = 39Ω and it's still doing on the order of 200ns, about as fast as the device can switch anyway.  (IGBTs are a hybrid MOS-BJT structure, with most of the current carried by the BJT; as a result, they have switching performance similar to BJTs, most importantly storage time and fall time are limited by semiconductor properties (doping and carrier lifetime) -- whereas for a MOSFET alone, it's defined by terminal capacitances and how hard you can drive the gate.)

RJH1CV7DPK shows EOL on Mouser, are you sure...you didn't mix up the two??

Another important property you will need to shop for, is using two in parallel.  IGBTs in general, parallel about as well as BJTs, for the same reason -- which is to say, typically not well.  You're looking for a positive tempco in Vce(sat), so that as one device heats up, it carries less current in saturation, putting more load on the other which then heats up as well, and they track.  Most have a negative to flat tempco, so are difficult to parallel.

The RJH1CV7DPK definitely shows higher capacitance (ugh, why can't they use log-log capacitance plots), but shows similar switching speeds at modest R_G so it might not be a bad sub.  Beware that current ratings are slightly lower; maybe it will be more prone to failure as a result.

IXDH20N120D1: pulsed current seems to be smaller, not sure how significant that might be in this application.  Oh, it's probably just that it's measured at TC=90°C for some reason.
Nice that it has a short-circuit rating, but it doesn't have a FBSOA to fill out that picture, go figure.
Also no capacitance curves?  Weird...
Doesn't show Vce(sat) vs. T, and the "Typ. output characteristics" for 25, 125°C look very similar, but notice the one has an expanded axis, so it's probably the case the tempco is positive.
Has a different package, but TO-247 is usually compatible with TO-3P -- good eye expanding your search this way.

Beware that the gate drive is usually toast, as the transistor fails as a three-way short, putting some hundreds of volts momentarily into that part of the circuit.  The gate resistor, drive transistors, and perhaps neighboring logic, or aux supply, may need to be replaced as well.  (Which, if that includes the micro itself, good luck with that.)  If it's totally toast, good thing these aren't expensive or anything... well, give or take if this is a single hob or one of several on a stove, YMMV.



--- Quote from: shakalnokturn on May 12, 2021, 07:10:18 am ---Usually higher gate capacitance means that they'll be rated for higher collector current.

--- End quote ---
The parts I looked it had higher gate capacitance, with only moderately increased Ic.

Thanks for the detailed reply, indeed they are available at Mouser. I swear I checked last night...

I was stumped by the lack of a gate charge specification, based on my limited experience with power MOSFETs. The gate driver is a discrete pair of BJTs, both of which seem to be healthy (~0.6V across all junctions, but measured in-circuit), so maybe the driver survived the failure. It's also interesting that only one of the two paralleled IGBTs failed, likely due to current hogging - there's nothing on the emitters.

I wanted to try to fix this mostly to see if I could; I suppose I should just give in and get a new one (the original is a cheap countertop unit, nothing fancy).

With 2 in parallel I would replace both to get a slightly better chance for matching.


[0] Message Index

[#] Next page

There was an error while thanking
Go to full version