Is That Part A Counterfeit? Here Are A Few Pointers

If you order an electronic component, how do you know what it is you are receiving? It has the right package and markings, but have you got the real thing from the original manufacturer or have you got an inferior counterfeit? We hear so much about counterfeit parts, and sometimes the level of effort put in by the fraudsters is so high that from either a visual or electrical standpoint they can be hard to spot.

[Robb Hammond] writes for Aeri, with an extremely interesting guide to some of the cues for spotting a counterfeit semiconductor part. In doing so he gives us something of an insight into the techniques used by the fraudsters.

The first feature of a package to be examined are the indents. Relabeled chips often have their old markings sanded off and a coating applied to simulate the surface of an unmolested chip, and this coating can either obliterate or partially fill any indentations. Using comparison photos we are shown discernable hidden indents, and partially filled indents.

We’re shown textures and paints, and how markings can sometimes be shown as counterfeit by washing with solvent. A Cypress-marked part is found to be a cheaper Altera one under the paint, and other parts are shown with misaligned markings and markings placed over indents. Wildly varying countries of origin are claimed while seemingly retaining the same batch codes, an impossibility confirmed by manufacturers.

If you order your parts from legitimate distributors then it’s likely that what you receive will be the genuine article. However with the popularity of online auction sites and online bazaars the possibility has become ever more likely of being left with a counterfeit. Knowing some of these tips might just make the difference between the success or failure of your work, so it’s an interesting read.

Have you had any dodgy parts on your bench? Tell us about them in the comments. Meanwhile, it’s a subject we’ve covered before.

Via Hacker News.

69 thoughts on “Is That Part A Counterfeit? Here Are A Few Pointers

  1. When I worked for a defense contractor we had TVS diodes that were shipped to us by a franchised distributor. The strange thing was that the parts worked, at first. Meaning the parts functioned as a diode because they were diodes, remarked as TVS product. Not till we got the materials xray-ed and de-capped did we verify the product was counterfeit. The way these got through is the distributor purchased the materials as NOS stock from a known and trusted broker. They were wrong. The root of how this problem crops up in military product over and over is because of how the FAR/DFAR regulations are written. They make it very, very difficult to change a design. If there is a shortage and open order (or worse a DX rating); the manufacturer will do almost anything possible to close the supply gap.

  2. I had a FTDI USB/RS232 chip, marked up as such, that the system recognised as a Prolific chip!
    Also, another that the FTDIProgramming/configuration tool was happy with, but refused to “take” config’ changes, such as inverting the TX data line.
    Buyer beware.

      1. Real? We’ve got literally thousands of product in the field with real FTDI chips that work a charm. You’re hard pressed to find equal quality drivers IMHO. Prolific are ok, that Cypress chip is ok, but… I’ll completely concede that their corporate ethics are sketchy deliberately trying to brick fake chips but their products are quality.

        1. Many people thought they had original FTDI chips in their products and got bitten when FTDI decided to brick clones. A lot of customers were not amused, and companies who did nothing wrong but being the victim themselves got stung for it. These weren’t companies who should have known either.

        2. I don’t know much about driver quality for usb serial chips.
          They all just seem to work on my linux box :)

          But I have to admit that I haven’t tried to use “advanced” features of those chips though.

  3. Not necessarily a “counterfeit” issue, but I once tried to repurpose a super-cheap USB cable to power a project. Snipped off one end, soldered the red wire to VIN and the black wire to GND, and turned it on. Of course, my project did nothing.

    After an hour of debugging the circuit, and finding nothing that made sense, I took a multimeter to that USB cable. They had used black for the 5V wire, and red for the ground. Whoever set up the wire spool that day just couldn’t be bothered, I guess.

    1. Or how about the “power only” USB cables that come with some of the USB batteries? Looks just like a real USB cable. It’s just missing the data lines.No easy way to tell by looking at it. But someone saved a fraction of a cent per cable somewhere along the line.

      1. I have been burned by several of those recently! I don’t trust the 6″ cables that come with USB batteries, but recently, I’d had a bunch of 3′ cables that were “charge only.” These shouldn’t have the USB symbol on them.

        1. Advantage is that you can use it to charge from public places without being hacked.
          Is it a feature or a bug? I guess it’s about the markings, it would be a feature if there was some standard marking to let you know. Maybe the USB foundation should release a logo.

      2. Failure to lable them is egregious; but USB cables that lack data lines have their virtues for charging from things that are probably OK-ish on voltage but can’t be trusted in terms of what they might try.

      3. There is a trend in uber cheap USB cables to be in clear blueish insulation, with very good stress relief, shielding and twisted pairs. Searching for USB clear data cable yields good result, often cheaper than charge only cables.

    2. I cut apart a USB 2.0 cable and found it was a basic four conductor cable. The USB spec call out a twisted pair for the data pair, which did not exist. It worked ok, as I found you can cheat on short lengths. But if your cable acts flaky, just keep in mind where it is made.

      1. Every now and then I need a long cable with weird connectors (Serial, usb, whatever).
        Instead of going to a shop and buying long cables which also take up a lot of room for storage, I changed to making changeover connectors to be able to use standard CAT-5 cable.
        I’ve used such a cable with avrdude (low-speed usb) to re-program an avr that was mounted in a weird place over a 10m long CAT-5 cable without any problems.

  4. Sometimes even trusted brokers or suppliers get a batch of counterfeits, I’ve gone through a few such batches with some of my vintage computer kits. Also there are vendors that “recertify” and relabel parts; they will usually be upfront about this if it’s a trustworthy vendor.

  5. “A Cypress-marked part is found to be a cheaper Altera one under the paint” – i don’t agree with that logic. Camouflaging a PLD as a microcontroller makes no sense – they are NOT the same, so a “cheaper one”? – the seller just had a shitload of the old Altera PLD’s in stock and remarked them instead of having “empty” chips manufactured.

  6. This is an interesting (and “save-worthy”) article but from experience in the manufacturing world there can be a huge difference between what manufacturers write down (somewhere) as “acceptable” and what a production line (which is paid to produce to quantity) will actually ship.

    I can’t help but wonder what simple mistakes (paint striping etc.) are out there on legitimate parts as well.

  7. For years we have had a fetal heartbeat detector with a built in rechargeable battery, obviously the battery eventuality needed replacing, upon opening the unit I noticed the board is marked “shenzen bestman instruments co ltd” all the transistors and chps have had their markings ground off but at least the battery is on a plug in lead, I un plugged it and measured the output, it’s a 9.6v pack but +ve is blue and -ve is red.

    1. Yes, they probably are, and then they’ll have to “up their game” to get fakes past us, some won’t bother, their gig is up and they’ll find some other market to pass junk on. Fake malaria medicine is big in 3rd world countries.

  8. This may be helpful to someone.

    I am color-blind so I see things differently and I can spot a fake chip by the color-temperature of the printed part number.

    Color-temperature is like a shade of contrast.

    I view part numbers under a 5700 degrees Kelvin (color-temperature) light. The printing on fake chips has a color-temperature between about 3000 to 3800 degrees Kelvin.

    I wonder if normal color-perceptive people could use UV light or something like that. Perhaps UV lighting and viewing via a phone camera.

      1. Well that is better than the “What color is Red to you?” question.

        It’s neurological rather than vision-sensory. My brain does not process color information. I see by shape-depth and color temperature as a contrast to ambient color temperature. I also have dyslexia.

        I also can’t tell easily between left and right as my vision is symmetrical or rather processed that way.

        I do very poorly on VCS testing.

        The clue came when I discovered that my tinnitus was neurological rather than physical.

        I have traced the problem to a defective human leukocyte antigen gene and mycotoxin exposure.

        In brief, my HLA gene defect causes an over-reactive cytokine (autocrine signalers) response that has damaged leptin receptors in my hypothalamus. My hypothalamus now can’t correctly signal my pituitary gland which causes endocrine dis-regulation and poor immune system function. From there the process repeats.

        This has caused a broad collection of neurological impairments.

        I am taking a selective toxin binder to artificially tag the toxins so that my weakened immune system can recognize them and remove them.

        I’m expecting to make a significant but mostly incomplete recovery. Incomplete because of the large time it has been since the per-cursory event. It has taken a long time mostly due the neurological impairments. At the end of any recovery I will have a per-disposition to multiple sclerosis should my life extend that far.

        This may all sound like bad news but it’s far better than the potential alternative outcomes.

        So as I often say – There is no life-saver in the gene pool so get your genetics checked.

        Is that interesting enough?

        1. Fascinating. I’d be very curious about the unabridged version of the route you took to arrive at this diagnostic conclusion. Incredibly unusual. Best wishes for the best outcome possible.

  9. I got some remarked ATTINY2313’s once. They worked fine. But the first time I used them, my app ran slow. Then I read the fuses and found out that they were all set to 4MHz (default is 8MHz). And then I had a better look, and found that they were ground down or blacktopped (not sure which). After reprogramming the fuses, every single one of them worked fine. So most likely they are just pulls that had been rebranded with a newer date.

    I posted a bit about it on Facebook. Not sure if this link works, but Facebook says it does. :)

    1. I’ve intentionally bought a few hard-to-get parts that were remarked in this fashion. Aside from the reprocessor using the original manufacturer’s logo, I view it as totally legit IF the chips are marketed that way (they’re often not).

      1. I am using them for hobby purposes, so I don’t care much. But secretly I was hoping that they were ATTINY4313’s (4KB flash instead of 2KB). Ah well, sadly not, so I still had to optimize my code for space. :P

    2. I have an ATMEL STK-500 from “the old days”. It worked well but I was working with the ATtiny12 at the time (~5 yrs ago), which required HVSP to program them. Eventually, one of the I/O pins on the AT90S8535 was blown and I could no longer use the STK-500, so I had to find a different programmer. I found the “STK-500 repair guide” online, which allows one to use the ATmega8535 since the AT90S8535 is no longer available. Since using the ATmega8535 requires using a hacked copy of the original AT90S1200’s controller code, I decided to hold off. I wasn’t really using the STK-500 any longer anyhow because the AVR Dragon was smaller and worked better for my purposes.

      About 2 years ago, I found a vendor on eBay that had AT90S8535’s for a reasonable price, so I ordered 4 of them. They arrived in the usual 3 week time period. I verified the partnumbers and they looked correct. I considered myself lucky because I could now repair the STK-500, which was in storage 200 miles away.

      A few weeks ago, I was at said storage location and came upon the STK-500, which I promptly packed into my vehicle for future repair “surgery”. Once I returned “home”, I decided to see if I could repair it. Having the proper tools, I was able to remove the old AT90S8535 with only one lifted trace from the PCB. I grabbed one of the replacement AT90S8535’s and soldered it into place then visually checked for solder shorts and set out to verify that I could communicate with the newly replaced AT90S8535. The repair guide spoke of desoldering the AT90S1200’s RESET pin and tying it to 5V and soldering a jumper between two pins, so the AT90S8535 could be accessed via the ISP header. I didn’t like that idea, so I found that intercepting and breaking the RESET line from the AVR DRAGON and tying it to the AT90S1200’s pin that drives the AT90S8535’s RESET pin and then tying the ISP header’s RESET pin to 5V did the same thing without any soldering.

      Once I had a connection, I was able to communicate with the AT90S8535 BUT I found that the newly replaced AT90S8535 was reporting to be an ATmega8535! I looked at the other 3 chips and saw that two of them had a date code of “0209” and the other one was “1328”. Hmmm… an AT90S8535 manufactured in 2013?? I realize that I had been dupped! Unfortunately, I had installed one of the chips with the 1328 date code in the STK-500. :( Well, the ATmega8535’s code only needed a single byte changed to work with my verison of AVR STUDIO. I really don’t use the STK-500 but at least it is repaired. If a firmware upgrade is required, I may have a problem but for now, at least the STK-500 is working.

      I have not had a chance to check the two AT90S8535’s with the 0209 date code. Mostly because I don’t have a means to temporarily connect to the ISP pins.

      Peace and blessings,
      Johnny Quest (and Bandit too!)

    1. It’s not a matter of remarking (though I’m sure you can buy fake Monster Cables…for whatever that means, heh), just you can order basically any cable configuration you want from China. Want a braided power cable with paralleled separate wires for phase/neutral/ground? No problem. Make no mistake, that’s where most of this “audiophile” and enthusiast stuff comes from.

    2. this is my least favorite part about the modern technology market-place. everyone knows the $1 LED lamps (A19) are crap, and the USB cable that came free with your budget phone, and so on. but you can pay $20 for an LED light and it comes from the same batch of QC rejects. or you can pay $100 for a cable and it’s still non-conforming garbage. there are good and bad out there but they are absolutely not reliably sorted by price.

  10. i’ve seen similar problems with medical supplies. people think they are getting a great deal on a Cat tourniquet being sold on Amazon for half the wholesale cost and leave a glowing review, but as soon as they try using it(hopefully only in a training environment) they find out that it is not, in fact, and actual CAT. in fact there are a lot of medical and first aid supplies being counterfeited like that. in general if the price seems too good to be true, it is.

  11. Sourcing components for mass production can be interesting.. We had big issues sourcing a Freescale/NXP MCU, out of stock on all major dealers, Factory in Shenzhen failed to source. Out of desperation we even asked 10 random alibaba electronics resellers and they could not find it either.

    Then I had an excellent mail conversations with Dave at components-center (the HK site) he always have an almost infinite number of components, and I got whatever price i want for them. He even accepted 1/10 of the price compared with his initial offering for parts the manufacturer confirmed did not exist…

    We ended up redesigning the PCB as we could not live with the manufacturers 49 weeks lead time..
    This time around we made sure to select parts that was in stock in great quantities from manufacturers with low lead times. It even ended up decreasing the BOM cost.

  12. In my last job, we had some watchdog chips that had an internal NV flash, authentic parts from two different suppliers, it wasn’t until we verified against the datasheets that the minor difference between them caused an issue.
    We had a bunch of micros not working, turned out that they were essentially just lumps of copper and nothing else inside.
    My current job we see counterfeit parts a lot, due to the scarcity of a required part that is long obsolete, mixed in with genuine parts are counterfeit, they were identical, except for a very very slight typeset difference in the characters.

  13. It seems like there should be a database of high-quality macro part photos to help verify this stuff. Reputable dealers could upload quality shots of what their wares look like, and buyers could confirm with visual inspection whether their stock looks like the expected items.

    Of course, if this becomes a big enough issue, maybe we’ll see physical anti-counterfeit measures e.g. hologram stickers, hidden CPU_ID type instructions in micros, etc.

  14. I once bought some replacement power transistors for a vintage amplifier on ebay. once they arrived they looked quite different from the authentic ones, the plastic was shinier and the markings weren’t clear, so I pried one open and lo and behold, a tiny wafer about 1mm sq. instead of the beefier one that was in the original. this was before I learned that bootleg components were a thing.

  15. I’ve given up on power SSRs, can’t find a trusted source that isn’t 10x over my budget. E.g. how do you get a 25A or 40A SSRs that doesn’t have something like a 6A triac in it? Last try I got some 40A fostek units at a realistic price. The epoxy fill was good and they looked OK. They worked for a while and then burned out. I managed to pry the triac case off the epoxy and it turned out to be a 6A part…

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s