#1749

Episode 1749
A generous viewer in Denmark opened the fake LM358 parts to find a C665 die inside
Be a Patron: / imsaiguy

[@iowahank1]

Very cool. Many thanks to your Danish viewer for doing this so we all could learn a little something.

[@windmill1965]

At the semiconductor company where I used to work we would not write the commercial type number (e.g. LM358) on the die, but use the internal project number, which included a version number. Each time when we had to do a revision change major enough we would also have to modify the metal layer to indicate the new development type number. Maybe that second chip (C665) is such a product which is marked with its development number instead of a commercial part number.

[@Oldclunker-ge5zp]

The X-ray picture is already quite helpful to spot a fake part. Die photos can be found the old databooks.If the landing zones of the X-ray don't fit those in the die photo, you know you purchased a fake part. The 358 has a common mode input range that extends beyond the negative supply rail. Many applications rely on that. I wonder whether the fake parts would even work then...

[@d614gakadoug9]

The common mode voltage range of the LM358 extends to only about 0.3 below the negative rail because of the input protection diodes. The positive limit in terms of function is up to about 1.5 V below the positive rail but the inputs can be subjected to the full rated supply voltage of the part without damage, regardless of actual supply voltage used. +32 V is the rating for the "standard" family. This is unusual and due to the fact that the inputs are made with lateral PNP transistors that have remarkably high base-emitter reverse breakdown voltage (most silicon transistors have reverse B-E breakdown of around 7 volts). Lots of op amps have a pair of "back to back" diodes between the inputs, so the differential voltage is limited. There are no such diodes in the 358 so it is more suitable for things like comparators where you might want to allow a large differential at the inputs without current greater than normal input bias current. It is slow as a comparator but still quite acceptable for lots of things. Die diagrams are one of those things that just slowly disappeared without most (I think) people noticing. I have no idea what is even being offered for bare dies these days. I don't suppose there is a lot of demand for hybrid circuits anymore so not much demand for bare chips.

[@paulfasolo8552]

Back in my early engineering career, we would hide our initials in the substrate

[@d614gakadoug9]

"Fuming" nitric acid, either the red or yellow variety, is most definitely not something anyone who is not VERY well versed in handling of dangerous chemicals should be attempting to use. I suppose you can make the case that it isn't as dangerous as hydrofluoric acid, but ... It is really quite disgusting that electronics manufactures have to go to such extremes to assure that the parts they are buying are not counterfeit. If a company is forced to resort to a parts broker because of extreme shortages, which certainly have existed in recent years, I can see it. There have been instances in years gone by of counterfeits making it into legitimate distribution channels.

[@allangibson8494]

Hydrofluoric is safer in a lot of applications… (definitely something a respirator and gloves are mandatory with however).

[@sbreheny]

I have de-capped several ICs at home and it really isn't very difficult or dangerous. The nitric acid method is very fast but there are slower, safer ways. The most common way I've used is to just drop the entire die into a flask and add concentrated sulfuric acid. Then I heat this to about 100 to 150C with a stopper in the flask and leave it for about 6 hours, occasionally agitating it manually. The acid turns a dark brown or black color as it reacts with the plastic package and breaks it down. The plastic and metal parts get dissolved but the die is left untouched. When finished, and after the flask has cooled, I slowly pour the contents of the flask out into a much larger container of cool water, finally fishing the die out of that or if it remains in the flask, after all the acid is gone, I add some cool water to the flask and then pour it out onto a paper towel and retrieve the die. The only real precautions needed are safety goggles, plenty of water nearby in case of getting anything in your eyes, and washing your hands after each time you handle any of it. The sulfuric acid does not need to be pure, you can use acidic drain cleaner (don't use basic drain cleaner like lye).

[@rjordans]

Interesting, the original indeed has the circular patterns that you expect from a bipolar technology. The fake one looks like it is a cmos implementation.

[@rjordans]

BTW, if you're interested in figuring out how these layouts work without a deep dive into the math I can recommend the book from Hans Camenzind, Designing Analog Chips. Free for download from the official site.

[@IMSAIGuy]

yes, that is a great book. Hans is the inventor of the 555 timer chip

[@diemaschinedieviereckigeei2941]

@@rjordansWhat would be your recommendation if one wanted to go for the deep dive?

[@rjordans]

@@diemaschinedieviereckigeei2941 skywater 130nm with xschem and magic, get your hands dirty and 'build' things

[@rjordans]

On KZbin, bminch has a lot of very nice tutorials

[@jackevans2386]

Thanks also to your Danish viewer. Great video.

[@simonlyons5681]

Interesting. IMHO one of the highest volume counterfeit parts currently (with IoT everywhere) is the Dallas/Maxim/Analog Devices DS18B20 temp sensor. Even reputable outlets are selling fake parts. These work, but are not as accurate or stable as the real ones. It's concerning!

[@splitprissm9339]

Who in dog's name would counterfeit am LM358, a part that very likely is actually much more common than the mythically common 555 and could easily be recycled - take any random piece of junk electronics to parts and LM358, LM324, TL431, LM393 are literally the most likely chips to be among the parts you get. Would have guessed uPC665 or uPC6650, but none of these are opamps....

[@bryede]

The only reason would be because they're so popular that you can move a ton of whatever these are by calling them 358's and inserting them into the supply chain somehow.

[@dieSpinnt]

@@bryede Also sorting, testing and providing a guarantee for the parts (second hand parts) would be out of the scope of these scammers. If they could, they wouldn't hesitate to sell you black beans directly fished out of their chilly as integrated circuits. Hehehe:)

[@LaserFur]

LMC660 is a quad op amp. LMC662 is a dual op amp. but then the numbering jumps LMH6645,46,47

[@hirokazuamaike4932]

Same fake OPA128LM　Low bias ＞60fA FET input OPamp. I had bought 5 items from China. I tested -5pA,-7pA,-7pA,-10pA,+4pA . I had loss many hours.

[@IMSAIGuy]

+4pA that's a good trick

[@phuang3]

You probably got LF356 or some cheap fet-input ones.

[@dimBulb5]

Very interesting! Thanks!

[@johnwest7993]

It's a genuine C665! I am awed!

[@IMSAIGuy]

digikey doesn't carry them 😥

[@charlesnelson619]

You should trying to run it at up to 36V supply like the LM358 can use !

[@lavacat720]

The only thing running is you after getting blown smoke at by a lm386

[@dean5263]

Pretty cool, I use knock offs for hobby play and have found that proto typing a circuit is useless because of the poor tolerances. Thanks to your Danish viewer who helped you with this.

[@kennethlung3371]

Excellent vid. First time I've ever viewed the inside of an IC.

[@winstonsmith478]

Chips a la Antoine: an IC chip de-capping recipe without chemicals CuriousMarc Mar 8, 2020 "Antoine is back, and demonstrates his simple method to decap plastic encapsulated chips, which does not use any nasty chemicals." kzbin.info/www/bejne/kILIeXuPf76KjtE

[@steelplasma256]

I was also thinking of this video when he was talking about fuming acid like it was the only way to decap a chip.

[@SeanBZA]

Yes to the scammers if they need a dual opamp they will get a dual opamp from one of their suppliers in the right package, and sand the top, or blacktop them, and mark as needed. Might even be able to order them as plain top direct, cheap. After all you make better if you sell something that QC can check, see it is an opamp and works, and put in stock, so you get a few more orders before they realise the parts are all failing.

[@dean5263]

I performed warranty work in the 1980's and Emerson had a bunch of issues because of that.

[@nil0bject]

i think i got these fake op-amps on fake lm2596 modules. the buck did absolutely nothing... good video. Thank you

[@maItre_gonzo]

Interresting video. Out of curiosity and for science I baught some LM6172 from ali at an alarming low price (1.2€ each). As expected it was not original parts and furthermore almost all of them were deffective/non-functional parts (instantly blew up), few of them did not immediatly blew up but were behaving like a damaged part (output would not exceed 2V while powered with +/-12V). Maybe they re-badged them the wrong way (rotated 180° ?) I had prepared a test fixture to compare their performances against real ones I have in stock, I did not even need that. A bit disapointed as I was hopping to have some re-badged low cost op-amp, but it seems they did not even care doing that. Anyway, just from looking at them you could tell the package has been grinded to put a new marking on them...

[@WolfmanDude]

Yes, please do more videos on fake chips, that sounds like a very interesting!

[@IMSAIGuy]

kzbin.info/www/bejne/hITKooyCgbWIidUsi=NkEytthZjl0OBe2t kzbin.info/www/bejne/b3LaoYyef7unmsUsi=97x-yR_UErVbyshA kzbin.info/www/bejne/nXnZpXmpab15Z6csi=DQp6g4MNa3FJET9P kzbin.info/www/bejne/nabIeWysYrecpqssi=GTUq9o8-InOGVCp4 kzbin.info/www/bejne/mYCmapSglM-nrdEsi=ubEXO59sUZkaA3a3

[@graemedavidson499]

I wonder if someday manufacturers will add RFID in the IC package as an aid to detecting counterfeit products.

[@frankowalker4662]

That was interesting. Great photography.

[@TheStuffMade]

It's amazing they'd fake a jelly bean part like the LM358, but I guess someone had a warehouse full of these C665 that no one wanted so they became LM358 instead. Then again most people probably wouldn't notice as it's a very general purpose opamp, it would be different if they faked a 300MHz bandwidth opamp and you got one with only 50kHz bandwidth. It's understandable if your Danish viewer works for something like Terma then they need to make sure every part is genuine and within spec.

[@johnfox2483]

Well, why they do this? LM358 is not an expensive circuit. Actually it's dirt cheap. Customer buy it once, and then never buy again from the seller. Neither is 358 difficult to manufacture. At least part of low quality. Make plenty cheaply, add TI logo, sell for higher price. And only some customer will be suspicious, because of noises etc. The die from film contains two symmetrical parts. Maybe its an opamp. Maybe audio stereo amplifier. If you have better pictures, its possible to recreate schematic. Since LM358 is cheap, I don't see much sense in faking it. You need to make many thousands of them .. and find customers, to make reasonable money. And basis chip are not cost free either. Maybe it's real manufacturer mistake - someone took wrong batch of dies, or chose wrong label to print. Quality checking discovered error and they were throw to garbage.

[@phuang3]

You have no idea about this business in China. Theses are extracted from junks which imported into China. The whole village is doing this kind work. Recycled op amps are sold to the fake parts dealer measured in KGs! The parts are sorted out, cleaned and remarked. They are not interested in cloning a LM358.

[@zukjeff]

i miss 'ElectronUpdate'. he used to deconstruct devices and decap that parts , we good enough to locate the gates on the chips. He has been missing since Covid.

[@nalinux]

I have a few "LM35", with are in fact simple NPN transistors.

[@johnhansen4794]

LOL, from the thumbnail I thought this was a poorly designed DnD dungeon.

[@user255]

Does the industry commonly produce out of specs components? Just thinking if the hypothesis for cheap Chinese parts can even be partly true.

[@BeesKneesBenjamin]

I wonder about how bad these chips actually are regarding reliability. If they work for an hour after the first time being powered, I don't see why it wouldn't last as long as a western produced chip. Especially regarding the machines and technologies they use is still better than people had to work with 50 years ago... About 5 years ago I bought a bag with 200 NE5532s, they also didn't measure up with the real deal, but I sacrificed a couple to make my own datasheet and treated them as a different part. They're worse in about any aspect. But is on par with a 741 atleast, good enough in the majority of cases (I did have to "derate" them to about +/- 12 supply voltage to prevent them from frying themselves out of nowhere). I used like 15 of them in a measurement bridge for a vacuum meter. It's been on for a year 24/7 and it's still up and running, being as accurate as the day I built it...

[@y_x2]

Unbelievable such an inexpensive chip!

[@pentachronic]

I don’t understand why these Chinese companies don’t just produce datasheets, do all the testing/qualification and then sell the parts on the open market. Surely they could win market share that way.

[@graealex]

Well, at close to 100K subs, you might have already received sponsorship proposals for fibre lasers. They can be used to decap as well.

[@KeritechElectronics]

The applied science of chemical decapping... I'd probably shoot some photons of coherent light at it. Either way has its hazards and advantages as well as disadvantages.

[@IMSAIGuy]

I have seen people do laser decap but they don't end up with a clean die

[@FelinoAnticomunista]

I buy 10x TL084 in AliExpress... But.. they all was no working properly... I decapped one of them and.... Tadaaaaaa All them was LM324 remarked.

[@__--JY-Moe--__]

jingle bells or marble chimes....Hmm. pretty cool...

[@jessewilson8676]

C665. Getting close not religious but imagine if it was C666?

[@R50_J0]

Very good.

[@winstonsmith478]

Single op amp ripoff of this - Microchip Technology Inc. MCP665-E/UN Op Amp, Dual, 60MHz OP w /CS, E Temp, 10 MSOP 3x3mm Tube?

[@nicolaspinto2927]

Plausible, the MCP66X is a third of the price (possibly bc it's a SMT part) and has a much faster slew. I wouldn't call it a ripoff right off the bat myself without seeing a die shot of one of these chips first. More likely they packaged the cheaper die in a PDIP to sell and make a profit. The 665 is a dual opamp with CS pins which are internally tied low so if you leave them floating and just take the other eight pads out to pins, you have a strong candidate.

[@rjordans]

I'd expect to at least have the bond pads for the missing pins in here then. Looking at the layout the fake chip still is a dual opamp though. V+ mid top, V- mid bottom, outputs top sides and the other one are the inputs. You can see both halves are nicely mirrored, the top has the bigger transistors to drive the output, and in the middle there is a bias current generator shared by both halves of the chip.

[@nicolaspinto2927]

@@rjordans I had thought about the number of bond pads myself and that's why I believe that pins 5 and 6 are left floating with the assumption that the square pads on either side are those bond pads. I am not entirely certain of course. I am not very well versed with dies and those assumed pads don't look the same as the others so I could be entirely wrong. I couldn't find a die shot of a confirmed 665 and lack much to validate either way. Just figured on putting some conjecture out there.

[@Agent24Electronics]

@@nicolaspinto2927 Or maybe it's a copy of the MCP655 but with the CS pins and associated circuitry removed to be cheaper.

[@rjordans]

The mcp662 could be an option but that should be easy to test as it won't stand a supply voltage above 7v or so

[@issyvarsano5622]

but who know what they are packing

[@fer_fdi]

super!

[@tommiller1315]

Why would you want more fakes? Was it price, or unique function?

[@ciano5475]

The point is not get the random fake but the original chinese they put inside

[@tommiller1315]

@@ciano5475 Fake "is a really good part" so I'm guessing it had unexpected attributes which would be desirable to repeat. (I should have asked what was special about that version).

[@chodnejabko3553]

basically the chinese part had better slew rate then the original LM358 from Ti. He explains this at 7:44 in the old video.

[@testboga5991]

People are faking lm358s already? Tf?

[@tenminutetokyo2643]

TI now makes their real logo out of stripes which is hard to fake. So if you see any new chips with a solid TI logo most likely it's fake.

[@IMSAIGuy]

I have not see the stripes. be aware, this is a quote from TI directly: "Yes, the logo pattern can differ between parts based on what manufacturing site they were created in."

[@issyvarsano5622]

sorry , my mistake , C655 actualy is 2 bilateral switches , absolutely no OP. chinese scam. it is probably one of the many atempts to copy the TI SN74LVC2G66 . both 1G66 and 2G66 have chinese tofu copies marked as C665 and C655.

[@IMSAIGuy]

??? again, this device is an opamp, see video 1. in fact it is a nice opamp. it has nothing to do with switches.

[@issyvarsano5622]

C665 is actualy SN74LVC1G66

[@IMSAIGuy]

why did you comment? that part is 5 pins, this chip has 8. I would love to know your thought process. not to mention it actually functions as a opamp

[@enoz.j3506]

@@IMSAIGuy SN74LVC1G66 is a bi-lateral stitch LOL i would say C stands for copy + look up no.

[@killsalot78]

ah, propaganda video not all chinese crap are fake dies, and a lot of them aren't, its really only super super simple chips like these that actually have fakes also really weird you didn't talk about the differences in the xray pic at all, almost like you either don't have a clue what you're looking at or... you needed to leave things out on purpose to help your narrative