Detecting counterfeit 4110 mosfets

methods

1 GW
Joined
Aug 8, 2008
Messages
5,555
Location
Santa Cruz CA
First, here is a quick refresher on the real 4110 mosfets

Real_IRFB4110.jpg

Manufactuer Name: IRFB4110PBF
Datasheet: IRFB4110 Datasheet

VDSS = 100V
RDS(ON) = 3.7mOhms (typ) 4.5mOhms (max)
ID = 180A (silicon) 120A (package)

So what does all that mean?
Above are the three most important terms.
Translated to shop talk they are:

Max Voltage = 100V
Inline resistance = 3.7mOhms
Max Current = 120A

The RDS(ON) is perhaps the most critical factor.
Heat is the limiting factor in our application and the heat can be calculated in terms of I and R

I^2 * R

I is the current that you are driving through the mosfet
R is the internal resistance.
This shows you that if the internal resistance doubles the heat doubles

Example: I = 100A R=3mOhms

100 * 100 * 0.003 = 30W

This brings me to my point: Counterfeit 4110's

Fake 4110.JPG

There are 2 parts to a counterfeit - Appearance and Performance

Appearance:
I wont get into a big compare and contrast but you can tell right away that the stamp is different above.
These are labeled FB4110 and the date code and logo are in the wrong spot.
This has been discussed: Bobmcree on fake 4110's

Performance:
The way they fool us is by meeting some of the electrical requirements but not all.
In this case, I bet they meet the max voltage requirement.

Max Voltage = 100V
Inline Resistance = 7.4mOhms (Only for example - not actual value)

If you plug the counterfit in and apply 100V everything is fine
If you run it with 20A or 30A they are still nearly undetectable

Trouble is when you try to crank the current up to 100A
The mosfet will still work, but it will create double the amount of heat.

100 * 100 * 0.0074 = 74W :shock:

This is a killer.

A true IRFB 4110 mosfet is a very powerful device.
I personally run a 6 fet controller that does not even get hot at 60V 60A
Inferior mosfets require 12 or even 18 mosfets just to reach 60V 50A

So the point of writing this up is to make sure that the good name and benchmark we have established with the term "4110" is not misused, abused, or confused.
..
There are a rash of new controllers coming out with (what look like) these counterfeit 4110's.


Fake 4110.jpg

I would like to remain unbiased so I am not going to name any names.

Please, if you have the time and resources could you add to this posting by sharing any test results, experiences, or knowledge you have on the subject?

Thanks,
-methods

EDIT_1 - Added note that 7.4mOhms was a made up number for example purposes. Will correct with real number once measured.
 
Someone should explain how to measure the rds-on of both true and counterfait mosfet... I guess it is like a shunt work using 4 wire sensing..

i'll try that.. by activating it.. anyway i'll remove all the FB4110 on the crystalyte and put brand new true IR mosfet.

Doc
 
Here is the Internal Rectifier website:

Search 4110 on IR's website

You will note that they only make a few 4110's in the TO-220 package.
The picture above appears to have more authentic stamping than the ebay 4110's, but I see no version that it would correspond to.

-methods
 
Doctorbass said:
Someone should explain how to measure the rds-on of both true and counterfait mosfet... I guess it is like a shunt work using 4 wire sensing..

i'll try that.. by activating it.. anyway i'll remove all the FB4110 on the crystalyte and put brand new true IR mosfet.

Doc

Thanks Doc.
I was hoping that someone with the time and resources could produce a test like this.

If anyone would like to take the time to characterize these (possibly) counterfeit mosfets I will send you a complementary set of REAL 4110's for your trouble free of charge


-methods
 
methods said:
If anyone would like to take the time to characterize these (possibly) counterfeit mosfets I will send you a complementary set of REAL 4110's for your trouble free of charge

The thread seems to have been purged, but I remember testing a few (two or three out of the ten I bought) of them for their resistance values at one point. They do fall well within 4110 specs, and I know they work with at LEAST the full charge voltage of a 72V SLA pack. I still have the bunch I ordered when I discovered them on eBay.

Unfortunately, I doubt I qualify for the complimentary set, as I don't exactly have the most precise equipment (a $20 multimeter to check voltage, a Doc Wattson to keep an eye on current, and a battery for a handy current source). Plus, while they worked well, I was using two of them in a 30A controller, which is hardly pushing them, I think.

General consensus was to be wary, as they might perform as well as a Mexican 4110, but it's questionable as to whether they do it for very long.
 
i have run into several controllers with the counterfeit parts, and have not found any more failures among them than with the real parts, but most people do not run the parts at anything close to 100v. I would expect that if there are processing quality control issues that the major issue would be the breakdown voltage which is a function of the thickness of insulation layers. I would not pull them out of a working controller if i were using it myself, but i would not warranty one with them in it. now that the price has come down to 1.70 in qty 100 there seems little reason to use fakes.
 
Link- you can't just throw a meter across them to test them... You have to dynamicly load test them as you would test a battery Ri...


Methy- You have a hundred thousand dollars of testing equipment at work for testing electrical stuff. Clamp a real IRF4110 down to a sink, switch it into shorting across a shunt of the appropriate resistance, perhaps something like a long coil of 12awg wire in a 5gal bucket of water, get an accurate current reading passing through the load, and measure the voltage difference across the FET. Use V=IR, and you can know the resistance. Using something like a 10k switching signal and throwing it on a scope with leads across the FET to watch the voltage difference would be neat to see too.

I know from my dicking around with an ebay lot of 10 of them that they got hot enough to melt the thermal epoxy I mounted them down with before they did a neat cracking/arcing/shatter effect. I've never had a genuine 4110 be able to melt the thermal epoxy yet (haven't even been able to burn one up yet! but I keep trying, and I'm sure I will get it :p ). From my own super unprofessional observations, I would say the heat output ( which IS the same thing as a FETs Rds(on)), is around 4x higher. That is only a crude observation from watching the striking difference in the heating of the heatsink between playing with the fakes and the real FETs. I want to see some real testing! And I'm too lazy to do it myself :) :) :)

If my super super rough estimation is in the ball park, it would mean that the tiny wallet sized 6fet Infinion would be making less heat than the 18fet controllers running the fake chips. But again, I wana see some real testing! Somebody out there get to work getting us some data :)
 
searching on my past electrionic class, i found something interesting i noted;

Rdson is the drain-source resistance of a MOSFET during its conduction period.
We should know that Rdson changes drastically (maybe 300%) based on the amount of current flowing through the device. also changes due to gate-source voltage and temperature, but most due to current though the drain-source of a MOSFET. therefore we should think, not to measure a specific value, but a range. take for example a 10A continuous rated MOSFET. how to test?

-Using a 10A power supply with digital over-current protection.

-Measuring the drain-source terminals with a digital volt meter set for very low Ohms.

-Connect the MOSFET drain to the positive terminal of power supply, and source to the negative terminal.

-Setting the over-current on lowest level and voltage to some low voltage, say 20Vdc.

-Having a seperate power supply (with common ground system) turn the MOSFET on, and keep it on.

-Then turning on the main power supply and it should go immediately into current control mode.

-At every 1A interval measuring the voltage across drain-source. this voltage should be very low, so use a good volt meter. plot the results in excell, Rdson vs. current.

-Rdson = Vmeasure / ShortCircuitCurrent

It can be interesting, to repeat experiment but change the gate-source voltage to half of its value.


According to the IR website and the datasheet of the IRFB4110, the rds(on) is measured at this condition:

VGS = 10V, ID = 75A Pulse width ≤ 400μs; duty cycle ≤ 2%.

here is the link to the IR spec:http://www.irf.com/product-info/datasheets/data/irfb4110pbf.pdf

Doc
 
Another possible difference (between real and counterfeit) to be concerned about is the Miller capacitance of the MOSFETs. If the counterfeit ones have a greater capacitance they will turn on and off slower. This keeps the MOSFET in its linear region for a longer time which can easily overheat it.
 
liveforphysics said:
Link- you can't just throw a meter across them to test them... You have to dynamicly load test them as you would test a battery Ri...

Link said:
I don't exactly have the most precise equipment

Face saved?
 
I guess you guys all know that International Rectifier has a manufacturing plant in Xi'an, China, don't you?

It seems at least possible that these parts come from the Xi'an plant rather than the Tijuana, Mexico, plant.

If these really are counterfeit, then I would be really surprised if IR aren't on the case in China, notwithstanding the usual difficulties with IPR in that part of the world.

Maybe we shouldn't be too quick to condemn these parts as forgeries until we get some hard evidence that suggests they are. As far as I can see, there doesn't seem to be anything to suggest that these Chinese supplier parts are sub-standard, they just seem to be marked differently from those sold to Western markets.

I'm happy to be proved wrong, if someone can provide some evidence.

Jermy
 
I would say for starters, they have the "B" in the name, which is IR's nomenclature for lead free.

Yet, they have the familiar whiteish powdery haze which is exclusive to lead based chips. All the Pb-free stuff has that blinding almost polished looking finish that is an inherent metallurgy property of the lead free solders they are tinned with.

Any thoughts on that Jeremy?
 
Jeremy Harris said:
I'm happy to be proved wrong, if someone can provide some evidence.
Jermy

Why dont I just call IR and clear this up then?
I call and tell them that a distributor is trying to sell me a part "FB4110" and I ask them if those are genuine.

Can everyone agree that this would clear the air?

I think it is equally as dangerous to call a bad egg good as it is to call a good egg bad so lets put any doubt away right now.

-methods
 
wrobinson0413 said:
Sorry for the long posting...should have pruned the PCN reports a bit instead of just copying and pasting.

Any info is good info, thanks for helping out!
No need to prune.

I am calling right now.

-methods
 
Here is the marking spec for 4110s from the IR datasheet (sorry for the low quality image - I wanted it to be shown inline):

irfb4110marking.JPG
You should be able to read a FET, identify the factory code, and work week and year of manufacture, and establish (perhaps with an email to IR) whether the part is genuine. I know this image says "IRFB4310G", but it IS from the 4110 datasheet.

BTW, to my eye, some of the above images look definitely counterfeit, while others are much less clearly fake.

Does anyone know if 4310s are also (allegedly) being counterfeited?
 
I made the phone call.
IR has agreed to send me written confirmation of the fact that they do not produce a part number FB4110 anywhere in the world.

I am at this point going to stop with the niceties and say the following -


Parts that are not marked as above are counterfeit.
End of story.


IR does not make a part number FB4110.
They dont make it anywhere in the world.

I thank everyone who has helped and I ask that further proof, experiences, and measurements continue to come in.

-methods
 
I made a second call to a different department to ask a few more questions.
Turns out the TO-220 is ONLY built in Mexico - nowhere else - as stated above.

Regarding the FB, there are certain instances where the IR can be cut from the part number for very small parts.
I was told that these were usually only the surface mount automotive parts marked with (AUIR)
Protocol is to first drop the AU, then the IR
IR has no record of ever needing to drop the IR from IRFB4110 due to the large package size

The part is also marked "Lead Free" (The P on the 744P) and these parts obviously are not lead free.


-methods
 
Good work. It'll be interesting to see what IR does about the bogus parts coming from China now it knows about them. As it has a big investment in it's Xi'an plant there, my guess is that it will want to stamp down hard on forgeries.

BTW, the Pb free thing is a bit of a "Western world only" thing, as lot's of products in China have lead in them still. China has to comply with the lead-free regulations for export goods, AFAIK, but doesn't for goods intended only for domestic use. The Pb free designator on the part number is the suffix "PbF", BTW, as well as the "P" in the assembly code. IRFB4110's were sold as a non-Pb free item a while ago; newer ones have the PbF suffix on the part number but some suppliers still list the non-Pb free datasheet.

What puzzles me is why they bother to fake the IR logo, yet make an obvious error with the part number. It doesn't seem to make sense, does it?

For what it's worth, my money is on this being a "Chinese domestic use only" part, one that's probably not internationally qualified and that may well be made by the IR plant in Xi'an but never intended for export. I wouldn't mind betting that IR HQ wouldn't want to own up to making non-PbF parts in China. Not good PR in the current environmentally aware climate. This would seem to make more sense than them being made by some sort of back-street fab plant as forgeries, if only because fab plants cost megabucks, even in China.

Jeremy
 
There are boot leg semiconductor manufactureing plants. As far fetched as it sounds, it's actually getting to be a pretty substantial problem. If you manufacture a FET, and you are just looking to make some income, you may as well label it as a premium well respected FET, and then try to copy the design specifications as best as you can. Seems like a lot smarter game plan than comming up with your own brand X name/part number.

If you have the equipment/facility available to make a TO-220 package, you can go right ahead and make whatever you like, and label it however you like as well. In applications with front hubmotors, a failed/shorting FET could actually result in death in the wrong situation (fail closed, lock the hub). The ones thermal epoxyed to my CPU sink that I played with failed pretty impressively, and got that sink very hot before failure. The real IRF chips only make the sink Luke-warm, and I haven't been able to kill them yet.
 
I don't doubt that there are boot leg facilities about, but these FETs seems to be pretty close to the IRFB4110 spec in as far as we've able to tell from some pretty crude assessments. The obvious screw up with the part number marking still seems odd. Why go to the bother of setting up a multi-million dollar bootleg fab plant and then make an error that is so obvious?

Maybe we'll never know quite what the true story is here, whatever it is is sounds pretty fishy. The curious side of my nature would just really like to know who makes these "FB" marked parts, and what the real difference is between them and Tijuana manufactured IRFB4110s.

Anyone have access to a high power semiconductor test facility, complete with curve tracer that works up to around a couple of hundred amps with short pulse duration?

Jeremy
 
I agree that the FB is odd.

I am not prepared to believe that they are actually being produced by IR.
If they really were IR parts then logic would follow that they would be just as good as Mexican IR parts.
There would not have been multiple failures, they would not run hotter, and we would not be having this discussion.

IR is not going to produce a version of their own part that sucks.... If anything IR underrates all the parts they design because they want the big government contracts. We love to use IR parts because our quality people know they are easy to qualify.

It is more likely that these are some standard mosfets being produced by some standard company.
Someone is sanding them down and re-stamping them with new logos.
I have seen at least 2 different versions - one that was totally wrong and one that was almost right except missing the IR.

Guys like Kenny know that these mosfets are not up to par and that is why they are only rating their 18 fet controllers at 48A... I mean come on... They rate the 4310 12fet units at 48A already.

I am going with the re-stamp theory.

-methods

P.S. I am not convinced that we have tested these enough to believe that they are "close" to the real IRFB4110's either. I dont know anyone who has run even close to 100V or anything close to what I would call high current on these. My bet is that when we test them they will be inferior to an astounding degree.
 
Jeremy Harris said:
Why go to the bother of setting up a multi-million dollar bootleg fab plant and then make an error that is so obvious?
Jeremy


That's the thing Jeremy, you don't open a multi-million dollar plant just to bootleg a chip. That wouldn't make any sense at all.

Right now, counterfeit commercial airliner parts are a major issue. I've seen reports that claim there are more counterfeit airline parts in use than genuine parts. Nobody goes out and invests in the real full start-up costs to make fake airline parts. That would cost many millions. What happens is some plant upgrades equipment, or maybe an outdated facility closes, and the machines are liquidated in auctions, and eventually wind up being sold for pennys on the dollar to who ever wants them. Then they can setup a crude but functional factory anywhere, and have a legit manufacturing plant that also happens to make some counterfeit stuff on the side. In China, this is not something difficult to get away with...


In college, I worked as an engineer for a company named G-speed, which built a lot of the aftermarket car parts. Most got re-sold/re-branded under a variety of names. The manufacturing was based in Taiwan and Russia. They landed a super fancy titanium forging setup that was part of a NASA program at a government auction that retailed in the 10's of millions. They got it so cheap, shipping it to Taiwan ended up costing twice what the equipment cost. It got all setup in a warehouse, and we started getting proto-type dies designed for parts, and fooling around with it, and then POOF! Somebody broke in and stole the control equipment for the forge, along with all our Ti blanks... Just the control equipment alone would have cost over a million bucks, so they ended up just canceling the lease on the building and leaving it. lol. That was a major bummer, because we could have made a fortune with that setup :(

Manufacturing processes are always improving. This means outdated equipment is always being replaced. It ends up somewhere, in someones hands, and they can make whatever they like with it, often for comparatively minimal start-up costs.
 
Be good to get some accurate test data, if only to show just how bad/good these things are. I may try and measure some of the characteristics, as I have a pack of the dodgy 4110s that I got from Ebay a while ago. I'm sure I've got a few genuine IR ones about somewhere, so if I get the time I'll see if I can measure some of the low power characteristics somehow. I don't have much in the way of modern test gear, but if the differences are big enough I should be able to spot them, all the same.

It might be a week or so until I get enough time to give it a go.

BTW, I've some experience of setting up fab plants, in that I ran a major government contract that ended up months late because one custom ASIC supplier decided to relocate their fab plant from Italy to Switzerland. Although they moved all the kit, so the new plant was identical to the old one, they still spent six months getting decent yields from the relocated plant, simply because the learning curve for the new operators was so steep. One thing I learned from that is that making wafers is as much art as science - the kit isn't the challenge, it's the skills embedded in the workforce.

Jeremy
 
Back
Top