Detecting counterfeit 4110 mosfets

[texaspyro]

he had a big ole Luke style shlt eatin EV grin.

I've always wondered why people grin when they eat shit? There was a little kid on our block when I was in 6th grade. He was a Batman like superhero crossed with Popeye. Beach towel for a cape, his super powers came not from spinach, but from eating dog crap. Grinned and giggled like a mofo when he did it...

[methods]

Working on what you are best at is efficient.

Even though I suggest doing this shit work I can not stand doing it myself. That is why I pawn it off on Matthew..... I can do about 2 or 3 hours of it a month and I have to have a real boner to get started on it.

-methods

[amberwolf]

When mosfets blow I do this:

wikied:
http://www.endless-sphere.com/w/index.p ... leshooting

[liveforphysics]

I have the ability to measure Rds down to fractional uOhm's on devices that can handle high currents.

It's actually pretty neat to have the resolution to watch the uOhm's climbing up rapidly as the silicon warms and bond wires warm etc, then after a minute (depending on the current you're running and how well sinked the part is of course), you can see it increase a whole mOhm, after 5 minutes (or whatever), you see the resistance is double or even 2.5x where it started, because the chip is reaching maximum operating temp.

[methods]

I have the ability to measure Rds down to fractional uOhm's on devices that can handle high currents.

That is just a matter of running the same procedure I just described but at 100A and measuring voltage with a standard 6 digit meter.

Stop showing off or I will have to bring out an arduino and show you that I can make the LED blink.

-methods

[Gordo]

I have the ability to measure Rds down to fractional uOhm's on devices that can handle high currents.

That is just a matter of running the same procedure I just described but at 100A and measuring voltage with a standard 6 digit meter.

Stop showing off or I will have to bring out an arduino and show you that I can make the LED blink.

-methods

I built some silly piece of crap a couple of months ago with a 555 that blinks some LEDs faster or slower depending on how wet your hand is or how tight you grip the wires. Just to show the wife I'm smarter than she is. NOT!

[methods]

The last time I used a 555 timer it was to convert a semiautomatic paintball gun to full auto.

Full Auto Conversion



I used a microphone and LabView to run an FFT so I could graph the firing rate in real time. 15 rounds per second limited by the internal circuit board.

You can tell by the video quality how dated this is

-methods

[liveforphysics]

I hate to be the guy that gets back on topic, but I actually think it would be a valuable resource for ES and the rest of the world doing anything with power electronics to test various real and counterfit FETs and make big shared document providing all the results.


I think helpful, easy, consistent tests for all potential suitable controller FETs would be something like this:


Short the gate low, and put the drain tab on an ice-cube, stick a precision 1mA power supply in series with a ~100-250vdc supply, and slowly ramp up voltage until you see the device start to conduct (the 1mA supply will drop out of CV mode into CC mode when you hit it, then you add the two supplies voltages together).

Now do the same test, but with the drain tab on a hot plate set to 140degC.

Then we have a kinda-sorta ball park range on how much voltage overhead a "100v" fet really has. I've heard stories of folks running real IRFB4110's at >110vdc, and having it work. It would be valuable to know where the real limits are so we can see what margin we've really got.

Then testing the real current handling would be nice too. Not just datasheet Rds numbers * Rth numbers with no thermal pad going to a magical perfect heat-sink, but something real. For example, bolt that thing to a controller case with a typical thermal pad in the stack-up, give it 1 minute of 50amps, see what happens. Then give it 1 minute of 75amps and see what happens. Then give it 1 minute of 100amps and see what happens. I don't expect any TO220 package devices to survive past 75amps for 1 minute, I would be pleasantly surprised if they make it past 50amps for 1 minute. Regardless of when they blow though, it would just be good data to have for everyone doing EV builds.

[John in CR]

Sounds good. You're the one with the fancy tool and the know-how, so I elect you.

It sure is fun to make commitments of the valuable time of others.

[CamLight]

Sounds good. You're the one with the fancy tool and the know-how, so I elect you.

It sure is fun to make commitments of the valuable time of others.

I second that motion.
This is awesome...democracy at work!

[d8veh]

My Ebay Chinese Fets arrived today. One of the 15 was very different (on the right) and the rest look like poor copies. I only want them to switch my battery, so hopefully they'll be OK.

[methods]

Dude those are so fake

From now on nobody is allowed to buy that crap. $1.50 each in quantity 1. That is as cheap as I was selling them WAY back when I was trying to change the world by bankrupting myself.

-methods

[Kin]

Uh, so, I apologize for what is certainly a stupid question...but...

Methods, with respect to replacing fets, a page back you said,

"12 fets, 2pcs"
"18 fets, 3pcs"

You also say, paraphrasing, "don't be stupid, replace the whole bank."

As there are three phases, doesn't a 12fet have 4 fets on each phase? And a 18 fet, 6 fets?

[John in CR]

Uh, so, I apologize for what is certainly a stupid question...but...

Methods, with respect to replacing fets, a page back you said,

"12 fets, 2pcs"
"18 fets, 3pcs"

You also say, paraphrasing, "don't be stupid, replace the whole bank."

As there are three phases, doesn't a 12fet have 4 fets on each phase? And a 18 fet, 6 fets?

There's a high side bank, and a low side bank for each phase.

[Kin]

Righto. >.<.


Thanks!


I suspected something like that, but http://endless-sphere.com/forums/viewto ... =2&t=41159, shares my sentiments <<but then how the bloody hell does a 9FET work?>> I'll keep reading, it seems I should have searched that earlier. [General point summary: in the 9fet, 2 fets are on one side and 1 on the other, and the switching all happens on the two fets.]

[flez1966]

My controller just blew after <5 mins of use....

Fets have a lovely IR logo, date stamps, very very well printed, and the part number is FB4410 .....





The case moulding is very like real ones....

Guess i need to find some 4110's and upgrade while i repair..

[amberwolf]

What conditions was it being run under?

Also, does the rest of your controller look like the one here:
viewtopic.php?f=4&t=44226
because that one has some definite QC (lack thereof) problems, including the way the heatsink bar mounts to the case which prevents heat from being able to escape properly, and that would also blow up even genuine FETs. (and it has an SMD resistor just sitting on the PCB, not soldered to anything; probably other issues I haven't found yet, and it has not even been powered on yet).


Another potential problem I see with yours is that it appears to have wrinkles in the kapton tape. If thsoe also exist behind the FETs, they will not sit flat and they cannot get their heat out of the FETs and into the heatsink bar.

Similarly, there appears to be gobs of white paste on there, too, which will do the same thing, as while it is better than air it is not nearly as good as a direct physical connection, and often large gobs also have large air pockets, blockign heat transfer.

There may be other issues, too, but that's what I can see in that single pic.

[flez1966]

yeah I never thought of the tape, I have had it apart before use, I always expect QC to be non existent and so i strip everything i buy and bench test before use.

I was 'happy' with the build on this, I just read about the bar issues and must say mine was sat nice to the case and had transfer goo to the case etc.

what I dont like is the fact that pcb is not supported by the case at all, it had no support on one side, and is hanging by the fets on the other, I'm old school, heatsinks have screws and are screwed to the pcb, the fet then has no weight on its legs and its legs should have a S bend for thermal expansion, and thats how it will be modded when it goes into real service...

It would run fine on the bench, would trip out after 10-15 seconds and get worse like it was heating up or losing volts, i just just kept going to destructively get to the bottom of the problem, as poking about with a meter was getting me nowhere

so, eventually it got very weak, and then would not go and the motor started 'cogging' badly so i knew a bank had shorted a phase out. blue phase to +pack, I think thats the more common failure mode but not sure why.

as i never had more than a 15 second 'burn' and never got above 20mph, eventually seeing it fail at about 10mph, in my mind it was weak fet that failed, stressing the remaining two so they failed, no part of the case was hot, but the bar was warm near the middle where the fets went pfft

there always a chance that the original fault still exists, maybe in the bank that switches the phase to ground... that could have caused excessive current to flow (and maybe the tripping) through the high side set taking them out...

all good fun these things...

so, now trying to source genuine ones preferably in the uk so if they do turn out to be fakes i can send them back...

[CamLight]

Fets have a lovely IR logo, date stamps, very very well printed, and the part number is FB4410 .....
<snip>
Guess i need to find some 4110's and upgrade while i repair..

Something to consider if replacing your FETs is that, if genuine, they are IRFB4410Z FETs and not IRFB4410. They are close in spec but the 4410Z version has a very reduced safe-operating-area at higher voltages compared to the 4410. Most of the other specs seem pretty close.

And be sure to add a thin film of thermal compound to both sides of the tape! To me, it appears that the FETs do not have any compound underneath them in your photo. There only seems to be compound underneath the tape.