Chinese branded IGBTs

[thaumatichthys]

Does anyone have experience with Chinese branded IGBTs for tesla coils? I've seen some on usually legit websites such as LCSC that seem to have similar specs to well known IGBTs at a much lower price, such as https://www.lcsc.com/product-detail/IGBTs_XDM-XD075H065CX1S3_C5301882.html.

[Rafft]

I rely solely on the price when making purchases. My primary platform for ordering is Shopee, which offers cash on delivery (COD) service. Currently, I'm facing financial constraints and cannot afford to buy high-quality items from websites that require a credit card.

I believe that the more expensive a product is, the better its specifications will be. Once, I asked a seller about the actual part number of a product because its advertised part number and specifications were significantly different. Unfortunately, I didn't receive a satisfactory response. As a result, I have experienced some failures with my purchases, with some breaking easily and a few failing silently. I predominantly use IGBTs for welding, which are moderately fast but perhaps more durable.

In summary, the price of IGBTs serves as a gauge for me in determining their quality.



to the moderators: just testing AI to rewrite my wordings to be more understandable, english is not my primary language. sometime MY english dont come out right. kindly delete if not apropriate.

[Late]

I haven't really used any Chinese branded IGBTs but as its from a legit website I don't see why they wouldn't work. They might not perform as well as advertised but you never know. Sometimes Chinese electronics have surprised me. I think u should buy some and try them. Then report your results here.I might get some too to test them.

[AstRii]

Using Chinese products is always risky unfortunately. And the fact that a "trusted" seller is selling them is not enough of a proof in my experience. I would definitely not use Chinese IGBTs for some project where reliability is necessary (for example Tesla coils for public shows), if however you only want the coil for yourself and run it once in a while, that's another story..

If you ever buy questionable transistors and want to test them beforehand to tell if they are the real deal, I suggest you watch my video on the topic:

[breezetix]

Does anyone have experience with Chinese branded IGBTs for tesla coils? I've seen some on usually legit websites such as LCSC that seem to have similar specs to well known IGBTs at a much lower price, such as https://www.lcsc.com/product-detail/IGBTs_XDM-XD075H065CX1S3_C5301882.html.

Just look for used IGBTs on eBay or something. Scrapyards, etc. They will work, guarantee, at least most of the times. I've never had a dead or fake IGBT delivered.

[Late]

Yea that is true. Chinese do like to fake their ratings. I've never before have had any issues with parts from LSCS though. And i think most used IGBT bricks you can find online for cheap are the real deal most of the time.

If you ever buy questionable transistors and want to test them beforehand to tell if they are the real deal, I suggest you watch my video on the topic:

Thanks for sharing the video. It will defiantly come in handy when testing IGBTs and fets.

[Felix M.]

Hi all,
I spent a lot of time on my first drsstc the last few months but I was never able to get it running how it should be. After a closer inspection of my IGBT bricks (SKM200GB128D), I have a feeling that they could be fake. Switching times and waveforms are just off no matter what I change. There‘s a lot of information in the Brigde output thread. To summ it up briefly, turn on and turn off times are longer than in the datasheet and the bridge output signals are uneven.

The sides of the case look sanded and the QR code is the same on each brick. The QR code on legit bricks seems to contain name/type of brick as well as part number and (probably factory number). On my bricks however the code is much smaller in size and only contains information about the type of brick.


I think I got my hands on some fake bricks that got rebranded. If someone has any idea, I would be happy to know 

Cheers, Felix M.

[Anders Mikkelsen]

Interesting stuff. You could measure the gate capacitance and Vce_on at 10 A for example, if you're looking for a few more points of comparison against the datasheet and against known good devices.

[davekni]

To summ it up briefly, turn on and turn off times are longer than in the datasheet

My experience with smaller IGBTs and FETs is similar.  DC specifications are easier to check, so counterfeit parts often meet DC specifications.  AC specifications, especially delay, are more work to check.  Most purchasers (including companies) don't bother with such detailed checking.  Slow IGBTs are relatively simple and cheap to make.

[Felix M.]

Interesting stuff. You could measure the gate capacitance and Vce_on at 10 A for example, if you're looking for a few more points of comparison against the datasheet and against known good devices.

Gate capacitance for the fast brick is 17nF and for the slow one it's at 21nF (So way higher than the 13nF the datasheet calls for). Voltage drop accross collector emitter junction is roughly the same for the fast and the slow one. Both measure about 1.87V at a constant 10A when fully turned on with 18Vge.

Idk if it helps but here's a more detailed immage of one brick (the fast one)


(The slow brick I did the test on also made a very silent sizzeling sound when he saw 10A for more than 5s)

[davekni]

Gate capacitance for the fast brick is 17nF and for the slow one it's at 21nF (So way higher than the 13nF the datasheet calls for).

Did you bias Vce at 25V for gate capacitance measurements?  25Vce is measurement condition on data sheet (as is typical).  Gate capacitance will be higher at 0Vce.

[Felix M.]

Thank you for mentioning because I forgot. The fast brick has 14.15nF and the slow has 16.91nF.
Is the faster switching characteristic of the one brick related to it's gate capacity being smaller?

[TMaxElectronics]

To summ it up briefly, turn on and turn off times are longer than in the datasheet

tbh that is the case with almost every IGBT that I have seen so far. Even original new out of box parts. The companies always polish the measurements to such a degree that I personally always measure the switching times instead of beleiving the datasheet. Not sure which exact IGBTs you have, but I have some SKM200 bricks that claim something ridiculous like 50ns fall time. With some of my other bricks even claiming 12ns fall time... So just because the rise and fall times don't match the datasheet doesn't necessarily proove fake IGBTs

Did you open one of them up? Its pretty easy to do, and if you're careful with the gate wiring you won't break the bricks either (although re-assembly could be a pain)

[Felix M.]

I thought about opening them but I'll first contact the seller. It's probably pointless but worth a shot anyways. I thing you have such problems with most components and their data sheets but 10 times more turn on and twice as high turn off times are probably not in the range of tollerance.

[davekni]

tbh that is the case with almost every IGBT that I have seen so far. Even original new out of box parts. The companies always polish the measurements to such a degree that I personally always measure the switching times instead of beleiving the datasheet. Not sure which exact IGBTs you have, but I have some SKM200 bricks that claim something ridiculous like 50ns fall time. With some of my other bricks even claiming 12ns fall time... So just because the rise and fall times don't match the datasheet doesn't necessarily proove fake IGBTs

This has not been my experience.  IGBTs I've measured are usually close to typical values and always within worst-case specification values.

Specification for the STGW60H65DRF parts I use in my large DRSSTC (10 parallel per switch) include the following diagrams showing test circuit and resulting waveforms with definitions of parameters.  Rise and fall times are defined as current rise and fall, not voltage rise and fall.  I presume this is the standard definition for IGBTs, though I haven't researched such.  I have seen this same definition in at least one other IGBT specification.





And here's a snippet from CM600DY_13T brick data sheet.  Looks a bit different, but still shows that delay and rise/fall times are for current rather than voltage:

[Felix M.]

The bricks were the problem! After receiving new ones of the type SKM200GB12E4, building them in and testing, everything works. Phase lead is at around 1us and the coil runns on 400A OCD on static loads. Good thing that this problem is solved but as always, the next one is just around the corner :/