Having worked with factories in China I'm not sure the sleuthing relating to glue types, etc really mean a lot. Our vendors regularly changed things like that, sometimes without even consulting us first. (Although they should have). Also, we made running changes on things like that all the time as well.
There is no reason why silk screen printing on a counterfeit product should be any less sharp, so I don't understand the relevance of saying the Canon logo will be a little blurred on a fake one.
The premise "How to spot a fake Canon flash" is nonsensical because as mentioned by gaius, often fakes are made in the same factory. In the case of a company I used to work for, their product was copied which included identical custom board layout and a copy of our firmware burned into the chips too. Obviously someone on the inside of our factory took all our designs and ip and was using it to build exact replicas. The only recourse, which worked for us, was to notify the Chinese government and they took action to shut down the people making the counterfeit products.
Congrats on successfully getting the government to step in!
Depending on the nature of your product and your supply chain, you could try to only flash the firmware once the finished products arrive in America (assuming the packaging isn't sealed yet). And do your best to make it hard to completely dump the firmware (maybe make part of it encrypted with a key that's in a one-time programmable memory that can't be read except by the decrypter, if one of your chips has this feature).
I don't know exactly because I wasn't involved in that side. The general thing I realized is that it's in the Chinese government's best interest to make sure large scale counterfeit operations are stopped. When looking for a factory you now have options beyond China, so if fakes in China marginalize your business, you're just going to move your manufacturing elsewhere.
I wonder if these companies could burn a "watermark" into their firmware, or create subtle differences of internal components. Then you give a slightly different variation to each factory; when the counterfeits come out, you know who to stop doing business with.
Some products I know need to be given key material in order to work properly (e.g., games or other software). It's relatively easy to control the product then.
I have been to a shop in China that was very up front about the fact that everything they sold was a qa reject. (mostly department store stuff, not electronics).
What prevents factories from implementing some online verification mechanism for serial IDs? At least for products which cost a few hundred bucks, this should be economically feasible, no? Then, before buying, you could scan a QR code and have the site tell you if it's a valid serial.
I had always presumed this was where NFC would gain adoption within $BrandName.
One could easily and non-invasively (preserving packaging) check authenticity against $BrandName's website. It would cost less than printing UUIDs outside the packaging and maybe make a dent in knock-offs by both increasing cost of reproduction and fear of being 'shamed' by others being able to validate your items in-person. It would also allow Marketing folks to provide backstorys on their products and a way to directly interface with people realizing their goods are fake...
If we're talking about online lookup, then there's a simple countermeasure: if the system has already seen the same serial number 50 times, or it sees the serial number in Shanghai then fifteen minutes later in Frankfurt, or any number of other heuristics, it'll report that it's a fake. True, this might inconvenience the one legitimate customer who had that original serial number.
Also, you need to make sure that the serial numbers are long and random-looking, so that they are not auto-generated.
I don't know if this will inconvenience the one customer unless they bought it used. Someone with connections to the counterfeiters would have owned that valid product for some short period of time and probably kept it.
To expand slightly on this point: I worked for a sub-contract manf company, and we made commercial flash units for a (now defunct) well known UK manufacturer.
We were idiots, and the products we made were not the quality a customer would have expected. But we were fully legit.
I suspect that a team of people creating extra units, skimming those off and selling them via ebay or such, would have been achievable for some products. There'd need to be coordination across a range of departments.
Is it just me or is the point of this article "you can't"? Professionals who deal with real Canons all day every day were fooled for an absurdly long time, the silver bullet relies on extremely niche domain knowledge (the structure of serial numbers) which the counterfeiters don't have to get wrong, and the author was only able to spot minute differences in shades of colors, etc, while staring at a real and fake flash side by side. If I don't have a real flash already, how can I know if the shade of the glass is slightly too blue?
Living and working in China, I quickly discovered that it's a buyer-beware market. Within a few weeks I had learned to always buy my electronics from the official stores in shops like Guomao. From ICs to smartphones, the fakes run the whole gamut.
In shops like Shenzhen's SEG, I've watched them apply brand stickers in public. When asked if they're real, they told me with a straight face that it was and came from the same factory. Yeah right. One of my coworkers lost all his photos from the trip, when his fake SD with a bogus size reached the internal limit and started overwriting his earliest pictures.
One of those fakes cost me my contract. I bought a USB key from a big box store. And used it to store some client work on it. Brought it into their office and nothing worked. I tested it on my servers and everything was fine. A few days later after constantly trying to debug it I copied the version from my USB key to my servers and bam! I finally get the same errors as the client.
In the end I wasted needless hours and missed a deadline over a $20 USB key (you know the cheap ones they keep at the checkout line).
2. If you knew a deadline was approaching why would you waste your time debugging a USB drive? There are plenty of other ways (including a new USB drive) to transfer the data.
1. Because the advertised capacity was 8GB but the real capacity was <1GB. There was a lot of coverage about this a few years ago. Including Bunnie Huang who had a problem with bad, fake SD cards being in his Chumbys'. The controller reported 8GB to the OS and so as it reached real capacity the controller would overwrite the first blocks.
2. If the code works on your computer yet doesn't work on the clients computer you would be lead to believe the problem is on the client's side. Honestly, how many people actually check their storage media? I didn't debug the USB drive I ran the code from the USB drive after recompiling 50 million times.
I don't see a lot of programmers trying to run diskcheck or memcheck every time they have an unexplainable crash.
I'm with the group that thinks that this flash was probably made in the same factory that makes the real flashes. My question is why would the fake flash stop working? Why would a "fake" product, which in this case is more likely a product made in an extra factory shift, be more likely to fail than an official product?
I'm with the two posters currently below me. When the factory runs out of the proper parts, they can do some "counterfeit" or sub-quality runs with near matches and most features will probably still work under most conditions.
The simplest example I can think of would be components value such as capacitors and resistors. If a design uses an uncommon part value (like 5.0uF) and the factory runs out, they can build with a common value (like 4.7uF) and come very close.
This is most likely to happen at the end of a production run. There aren't enough parts leftover for the actual quality device, but you can make substitutions and get some more near-quality units out instead of throwing out all of the rest of the good parts.
Another example I could foresee from the OP is that different revs of firmware were available at different times. e.g. a bug was discovered after the ROMs were masked or flashed and that run of chips should have been destroyed but weren't. (Hell, they might still be upgradeable!)
Ironically, the serial number was possibly intended to distinguish the official runs from sub-quality runs... The "counterfeiters" might not have been trying to fool everyone, but instead, to distinguish between official and ghost shifts. Welcome to the new age! :)
I'd say good materials and QA. Naturally some batches of parts (like LCD screens) might have a higher chance of failure than other batches, which is usually known at manufacture time. Then QA, where "bad" products don't make it to consumers hands.
Counterfeits would probably use more of these defective materials and go through less QA.
I've never seen such a convincing fake product. A lot of attention to detail. If they got the serial number right then it'd be impossible to tell without opening up (or at least peeling that tape).
I don't get this - How are you supposed to know that the serial numbers don't start with letters? It's not like you buy tons of identical devices and then start making diffs on them.
In the article he talks as if the repair shop is the first place he took it to, only one month after buying it. Then later he says "This one had fooled camera repair shops, professional photographers and Canon employees for way longer than we all probably would care to admit". His story doesn't add up.
If you read the comments you can see many people have Canon parts with serial numbers that start with a letter. Unless you think they are all fakes too.
Probably not enough to be a "feature", just that the real factory and the counterfeiters happen to use a different method.
Which is weird, actually. A lot of Chinese counterfeiting happens in the same factory. Say you're a famous clothing brand and you outsource making 10,000 of your ski jackets to a Chinese factory with Gore-tex. They make them and ship them back to you, then they keep the production line running with nylon, make 10,000 more and ship those to market vendors, eBay merchants, etc.
You use a controlled amount, and you use less glue.
This saves cost, but is also better engineering. The glue has data sheets and tests and you know you're meeting the engineering requirements. Some glues 'dissolve' / 'melt' the surfaces to be bonded, and you only need a thin smear to get a good bond.
You need to control stuff like glue or conformal coating over a run of X thousand units, otherwise you find people can use crates of the stuff.
Having said all that: Maybe it's just what was drawn on the CAD drawings, and it's been followed faithfully by workers ever since, and there's no actual reason for it other than the CAD worker chose something that looked nice and not confusing.
How do you think they extruded glue in a grid? Would it be an extrusion nozzle on an XY gantry (like a MakerBot), or am I overthinking this?
Maybe a simpler approach would be to extrude through a mask that has a grid-like pattern of holes? Obviously you could only make isolated dots of glue, but not overlapping lines of glue (you can't have floating islands in your mask layer). So although it wasn't the technique used in this article, I guess it could work just as well, right?
Henkel/Loctite have XY(Z) offerings for industrial uses [1,2] as well as screen printing of UV curable glues. The ability to post cure using UV LEDs makes this much more reasonable, even given the short cure times for some cyanoacrylates.
Can't one just spray the glue in that pattern with some kind of inkjet head? Or is that bound to fail due to glue's ability to ... well ... glue things together?
If the glue is standard cyanoacrylate ("super glue"), then, since the glue is moisture-activated, maybe you could do it without the nozzle getting clogged as long as you have some kind of amazing material for the nozzle that repels moisture.
Maybe someone with experience using cyanoacrylate in industrial settings can speak to this. All I know is that my hobby-store super glue nozzles (cheap plastic) get clogged almost instantly. :P
But, my main concern with any kind of XY gantry (as in an inkjet printer like you describe), is that it would take frigging FOREVER. There's no way you want this glueing step to take 45 seconds per unit or more. That's the puzzle, I think, is how did they create this glue grid so quickly that it rivals the blob approach (1 - 2 seconds)?
Stencil application, usually. Similar to the process used for solder paste or screen printing. You can use polypropylene or acetal nozzles, cyanoacrylate won't stick to those.
I thought I spotted the fake on the left because of the small, white, rectangular piece right below the red LED. On the left, the thing is looking much darker. Because there aren't high res images, ironically, it's a bit hard to tell.
But the best way to avoid a fake is to actually buy it from the official shops. That's not to say it too is 100% guaranteed as someone on the inside could have easily swapped the supply for a cheaper source, but it's still your best option.
So, no visual difference. And no functional difference, as he took a shot and it went fine (so, I presume it synced before those 2 months passed?). That leaves a shorter expected life-without-malfunctions time for a cheaper price. Now, if they'd be only be properly labelled as clones...
There is no reason why silk screen printing on a counterfeit product should be any less sharp, so I don't understand the relevance of saying the Canon logo will be a little blurred on a fake one.
The premise "How to spot a fake Canon flash" is nonsensical because as mentioned by gaius, often fakes are made in the same factory. In the case of a company I used to work for, their product was copied which included identical custom board layout and a copy of our firmware burned into the chips too. Obviously someone on the inside of our factory took all our designs and ip and was using it to build exact replicas. The only recourse, which worked for us, was to notify the Chinese government and they took action to shut down the people making the counterfeit products.