This is only half the story. The chips discussed in the article do the basic function advertised, but may not have the same tolerances as the genuine chip. But there are other types of counterfeits out there.
When we "recycle" our e-waste, much of it ends up in large landfills in China, where very poor workers melt down old boards, remove chips, and sort them by size and pin count. They then use a machine to scrub off the label and print on a new label of whatever part they want to market (usually obsolete parts that can't be bought from the original manufacturer). These end up in the supply chain of even reputable distributors. Amazingly, some of these counterfeits have made there way into military equipment, and as a result, there's a whole division of Homeland Security that conducts tests on suspect computer chip shipments entering the US.
A circuit board where you replace one of the specified chips with a completely random one which only has the same pin count won't pass even basic QA controls something like what, 99% of the time?
As another commenter said, smells like an urban legend.
Yes, and they were actually controller circuits for CPU voltage regulators.
There is also a market for recycled and semi-properly labeled chips, for example selling low precision grade analog chips as a higher grade or chips tested for the usual 0-70°C operating temperature range as rated for military/aerospace range -55-125°C.
This is harder to catch with QA. A few years ago there was an investigation in the US where they found some examples of such chips in military equipment and also proved counterfeiting beyond any doubt by ordering military grade parts which have never been offered by the manufacturer in military grade - the vendor simply relabeled them.
Do you mean we end up with a labeled chip whose insides are something completely different, and so won't ever work?
Surely that doesn't happen this way. It must be many times cheaper to just buy empty packages than to clean up old ones?
The worst part actually is, when those chips are the correct ones- but have been used and have all the wear. Means, you have them baked on and off board, slight overcurrent damages, wear on internal flash and Elektromigration.
https://en.wikipedia.org/wiki/Electromigration
And this chip is integrated into your new device.
The only way to avoid this, would be a standard start of life vault, that is writeable once with the date and burns of the pathway to to re-write after that.
Manufactureres force you allready to buy new equipment 2-3years, by placing codensators near heatsources and other nice tricks (including the famous printer counter).
In Sweden, we call them "Kondensatorer", i.e using the old form. For me, this causes quite a lot of jumbled/mixing of words, as my friends and I typically switch randomly between English and Swedish versions of component names.
Hi there, I'm very interested in your comment. Do you have a link where I could read more about the techniques you've discussed as they pertain to forced obsolescence? Thanks!
I have no direkt link, i just had - lets say a lesson at my university where one the teachers (who shall not be named) mentioned that there is a pretty exact science of statistically averaging the livetime of a capacitator with the heat it experiences during the devices operation. So if you want to extend a devices life, you can desolder some condensators and put them in areas of good Cooling.
I suspect they do it this way rather than manufacture empty packages because there are a huge variety of chips out there with an equally diverse set of package designs. It's easier to pay unskilled laborers to sort chips into 200 different groups based on size, color, and pin count than to build a machine to replicate 200 different package types.
That's the crux. You don't buy mass-produced empty packages to put the parts in. Packaging happens by putting the IC die, pins and bonding wires which connect them together inside a mold and filling it with plastic.
It's probably cheaper and certainly lower-tech to replace markings on existing chip than produce a new one.
There are more incipient problems. We've recently had an issue where some CCDs - readily available from China - work just as they're supposed to, but have some weird surface coating that causes interference effects. The up to date models, bought from a reputable distributor (e.g. Digikey et al) seem to work fine and look physically different.
It's not clear whether the ones you can buy on eBay or Aliexpress are simply older models (and the manufacturer then removed this coating), or are actually knock-offs. We have no way of really knowing.
It was a good lesson: most of the time it's just not worth it especially if you're not chasing pennies on your product margin.
Personal experience having purchased random TO-3 packaged transistors from ebay labeled as LM323K voltage regulators to repair old hardware. Its "well known" in the retrocomputing hobby.
Ironically SOME sellers on ebay are legit, and some sellers are legit sometimes, so if you play a statistical game buying from ten sellers at a 50th the price of legit on-shore sellers, you still save money. Obviously its a lot simpler to build a temporary test fixture for a 70s linear voltage regulator than a memory chip or obscure CPU support chip, LOL.
When I was working for a microprocessor company (which you use everyday) many many years ago, they were selling this JTAG debugger for a few thousand $ each. The cpu architecture was booming in China at that time, but that $ figure is not something people can afford. Clones started to flood the market pretty fast, a typical clone set you back for a few hundred $, that was already 1/10 of the original price.
However, that was still considered too expensive - then there was this vendor in Shanghai which came up a clone they sell for around $100 - after talking to them via some unofficial channels, we figured out that they had people stationed in those electronics dump sites to source the chips required. In those place, you don't pay for each chip, you pay for a kg price, e.g. I want 10kg of that chip, how much each kg please?
10+ years on, those engineers who first studied that CPU architecture back then using those cloned JTAG debugger are now pretty experienced/better paid, they are probably the ones complaining fake/reused chips on some Chinese forums claiming they are the victims.
The problem with faking chips is that the victims maybe out of a warranty claim later when they find a problem. Also there might be subtle differences which will only be found when in full production. For a cloned JTAG debugger the risks are lower since they are used in low volumes. Personally I think they it is okay to clone or reuse chips but don't claim to be the original manufacturer. It is quite easy to relabel chip with your own imprint.
I think a bigger problem with the 'chips by the kg' mindset, is that there is zero incentive and zero funding to develop good software to go along with the hardware.
Companies who are successful have spent way more time in ensuring that their software works reliably with their debuggers, across a wide range of chips and platforms. Sure it's expensive, but everything just works.
By 'stealing' this software in this way, cloning firms are dragging the entire industry down with the clones that don't work 100%, or the IoT devices that get infected due to lack of any security. I'm really hoping some certification body like UL, or major distributor like Amazon or Alibaba, steps in and starts ensuring at least a basic level of quality for hardware and software.
When I read this article I wondered, why would they knock them off and then sell branded goods at the rock bottom generic price? Then I clicked through a few more articles and it seems that the answer is so an integrator can "cost-optimize" your board and profit the difference between genuine and fake components, while you are none the wiser.
In one of our board designs, we were using a well-known manufacturer's 25Mhz oscillator, to provide a clock to a gigabit Ethernet PHY. When we wanted to assemble the boards, we could not find this oscillator in any official distributor, so we bought them from a vendor in China, who had some in stock.
When we assembled the boards, none of the boards functioned properly. There was no Ethernet connectivity, the PHY wasn't working
We contacted the "well known manufacturer" and provided photos of the part and they confirmed that the oscillator was not theirs. It was a counterfeit part.
The "counterfeit" oscillator actually did provide a clean 25Mhz clock, but the PHY did not work.
It amazes me there are companies out there that produce counterfeit silicon. I wonder what's their profit margin?
People buying 7 cent chips off Aliexpress and similar know full well that it ain't coming from TI and usually isn't labelled as such - it might be called an NE555 but make no mention of manufacturer and not have one labeled on the chip itself. Obviously finding a datasheet from a different manufacturer won't be applicable...
If you're buying those you're most likely using it for an application that isn't pushing its limits - they're great for experimenting with as a hobbyist or cranking your design to the lowest price possible if you're not on those limits.
Mouser is selling TI NE555DR for $0.07 in quantity, so I am sure that TI's actual sale price is lower.
So, I think someone paying $0.07 on aliexpress is paying "full price" for something less, even if they rationalize they are getting a discount because they are not buying large lots (Buy only 100 at mouser and the cost goes up to $0.15, but for instance Arrow has a lower price than that for this quantity)
AliExpress has another benefit though; zero to very low shipping costs.
In the EU buying electrical components at a hobbyist scale often means having to save up your purchases (e.g., €0,24 for the NE555 at Conrad) until you hit the minimum free shipping limit (in this case €20) or pay the full shipping (€4,95 in this example) for one little component.
AliExpress just has the better user experience if you are learning and experimenting (just buy a couple of spares and be prepared to switch to genuine stuff when needed).
I don't understand why these places can't just use regular mail.
I wanted some discrete through-hole MOSFETs for bidirectional level shifting as described in this application note [1].
The ZVNL110A satisfies all of the requirements [2]. In small quantity they were $0.604 each at the time from Mouser, and I wanted 10, so $6.04 for the MOSFETs. And then another $7.99 for freaking shipping. (I ended up ordering 30, figuring maybe I'll eventually need that many, and then I'm "only" paying 50% for shipping instead of >100%).
These things are very small, and only weigh half a gram each. Just take an ordinary letter envelope, drop them in, put a first class postage stamp on it ($0.49), and mail it. A $0.49 first class stamp allows up to 99 grams, so orders up to a couple hundred of these could be shipped that way.
For those curious, the size constraints on a first class letter in the US are:
0.25" is 6.35mm. A large number of through-hole components would usually fit, such as transistors, diodes, resistors, and disc capacitors. Most surface mount components would fit.
I wonder what the breakdown on the actual shipping vs handling labor costs are for a big distributor like Mouser or DigiKey.
Compare with labor costs for either factories or small owner operated export businesses run out of Shenzhen and you can see where there might be a pretty big difference in the handling price.
Labour costs are probably the primary barrier for this, but I can't help but wonder if a big distributor couldn't make it work; after all, the labour costs are already factored into the component unit price, not the shipping costs. For a large, traditional order you get a shipping list, determine the box size needed, gather the components, pack the box, slap a label on, and place it in the outgoing mail trolley.
For these small shipments, you would get a shipping list, see that the order system determined that it would fit in an envelope, grab an envelope, gather the components (perhaps for ten envelope-orders at a time), place in envelope, slap a label on, and place it in the outbox for envelopes.
It doesn't seem that much more labour intensive or more expensive in terms of packaging, and the first distributor in a geographical area to do so might gain a bunch of customers. Even if the base price of components was increased by 10% or 20% for envelope orders a lot of us would still use that service.
Mouser and DigiKey could offer a reduced SLA for small postal orders, meaning they wouldn't send someone out specifically to pick that order, but would do it opportunistically.
Letter envelopes go through roller machines that bend them quite a bit. If you sent rigid objects through it will be delivered in pieces. There was an EEVBlog video about a viewer who sent a PCB or solar cell (i forgot) in a letter... Dave received dust.
Interesting, after doing some searching all listings I've looked at they either lack a logo or have a TI-ish logo but they don't mention it being a TI brand or genuine TI part in any way in the description - which is always a dead giveaway as if they do so they can be subject to punishment. Alibaba is actually pretty harsh on people selling known counterfeit goods and will ban sellers if they catch onto it - so I assume they don't mention it because their buyers don't care much and they don't want to get in shit for it.
I dunno. It seems pretty descriptive to me. It identifies the source of the fakes (China), the product being counterfeited (the venerable 555 Timer), and the price they're being sold at.
I found this on eBay - 100 pcs of thru-hole 555's for $3.29 (3.2 cents each) with the TI logo printed on them. They say brand new and unused - but that's just not possible at that price.
Interesting I wonder what the history of this design is? And how Hitachi's (or their partner's) die really wound up being re-marked as TI? If the die is a Chinese produced copy, why did they keep the markings during retooling?
I wish photographs of the packages were presented also.
Edit: The other "not genuine" die has quite a few similarities and it looks like one inspired the other.
If it's copied, they likely copied it entirely because trying to remove bits of the lithography mask is expensive and time consuming. Since it's not normally scene there's no use in spending the time or money to change it if you can get the mask.
At the start of the ROHS era my startup was scrambling to get parts, and we bought some grey-market capacitors. Which turned out to be fakes. What a mess. It's all about the supply chain!
I'm a huge fan of the knock-off chips that are just a piece of copper connecting all the pins. I wonder why they bother with the copper, instead of just using a block of plastic.
I worked for a huge company with their own manufacturing in China. We did a beta run of a product and they used some black market components for a TI regulator. When we got the boards the output was dead on the regulator so we sent it to TI. They x-rayed the part and it was just an empty plastic package.
Did you have a pretty good idea that this was the case before sending them to TI? Hardware design fascinates me and I'm just curious how something like this goes down.
We actually thought it was a problem with TI's parts, which is why we sent it to TI (we had a dedicated TI rep that came by every 2-3 weeks). When it came back that the package was empty it was actually a little embarrassing for us, since everyone immediately knew they were fakes. A bunch of calls went out and eventually one of the manufacturing managers tracked down the purchasing guy in the factory that said something like, "Oh this was a small run and I didn't want to buy a whole reel so I just ran down to the corner electronics market and picked up a 100."
The package pins start as one piece copper stamping (called leadframe) in order to ease handling and to guarantee correct alignment of the pins. The short between all the pins gets removed pretyy late in the packaging process when the die is actually mounted on to the leadframe.
Even Apple was forced to deal with glued fake plastic capacitors. iMac times, “Oral History of Arthur “Art” Astrin”, wifi pioneer: https://youtu.be/Tj5NNxVwNwQ?t=1h34m
They may be dummy components specifically designed for testing board manufacturing processes, without using (more expensive) real parts with die in them.
I bought a few usb flash things on ebay a while back. One had a weird USB connector (hard to plug anywhere). I asked for money back, surprisingly it holds data.
Another was a microsd, I spent 2 hours running through some linux memory testing program (u3 . .. or something similar in name) to be sure it had the right amount and wasn't failing.
I honestly have a hard time believing this--rubycon is a good capacitor brand; why would you put a 'decent' capacitor in a fake product instead of using some cheap chinese crap?
Or maybe recycled, note the short legs. Part of this e-waste you are now obliged to dispose of "responsibly" lands on dump sites in Asia or Africa. Some of that ends up harvested for parts.
The 555 is probably simple enough so that the "fake" version (which several times is just an alternative version from a different manufacturer) works as well as the original one
They probably work but probably not within the same maximum/minimum ratings as the original chip. Your design might need a max frequency, or a minimum operating voltage, or a certain current-driving capability etc. that the fake chip cannot provide, causing the final product to fail when the design prototype worked fine. There are many variants of the 555 with wildly varying characteristics, not all are interchangeable.
What are your thoughts on people not respecting Intellectual Property rights? Will it lead us down a path of gridlock where nobody is going to create anything new for fear of someone else using their work without payment/credit? Also, why is using someones creative output or deriving from it OK in some fields (like applied sciences) but not in others? (Here I'm not including the people who claim others' work as their own, merely people who clone)
I bet there could be an actual business model of recycling components from e-waste. Obviously it'd require more QA work than what the counterfeit "recyclers" do, but, with enough automation, I bet you could structure it so that you at least break even on the e-waste disposal side, and then any resold components become pure profit.
As the article says, "If only have they marked it "CN555" with proper data-sheet - it would have been totally legitimate and fair product."
It's nice to produce new chip designs, and they should do more of that, but it's immoral and inefficient (it hurts everyone integrating your product) to say that you're producing X1 when you're producing X2.
If anything, the fact that a lot of chips you can buy from China do not match their specification is exactly one thing that keeps China's semiconductor industry back. If some Chinese company built a cheaper and in every regard better NE555, they would have a very hard time marketing it because noone would trust that sellers offering them wouldn't be selling them recycled bad-quality TI parts instead.
Although the author calls these designs "fake", they do perform the same functionality, but just simply don't meet the voltage tolerance and noise resistance levels of higher grade designs. So they're more of a "knock-off", "counterfeit", "imitation", or "derivative".
ok. Sorry I had to look up the definition of fake: "not genuine; counterfeit". So I guess fake is valid in this case. I was thinking fake meant "not real" or that it didn't actually do the same functionality.
"same functionality" includes a LOT more than just the basic functions. It also includes voltage tolerance, noise resistance, and all of the other things you dismissed.
But still as the author explains "If you're not reaching for the limits (like max VCC or precision) and have some luck - that might not cause issues." And I didn't dismiss that...I said "just simply don't meet the voltage tolerance and noise resistance levels of higher grade designs".
I guess what triggered me to write my first comment was the fact someone buying a "TI" something chip from eBay when it lists "China" as the seller's location and with a seller's name other than something easily verified to be TI, then it seems there is no pretense that the thing being sold is genuine TI. But the word "fake" holds the connotation that there is some deception going on. So that is why I don't think fake is the most appropriate word. I just hear that word used too much to disparage chinese made derivative imitation products. Someone who knows they don't need the noise levels and voltages tolerances can be perfectly happy and even benefit by using these instead of the more expensive genuine TI.
Well there is no way to verify that the device was manufactured by TI. That itself doesn't let you know as a matter of truth, but that does let you as a smart shopper know that it is likely fake. You can make an inference since (A) you are buying on a website that is known for providing a means for counterfeit goods from China, and (B) the seller is located in China and has no info that reliably identifies it as a seller of genuine parts.
That may be a valid point for a TI-specific part. But if the buyer is just wanting a generic 555 Timer, those aren't part of the specs. The 555 is effectively a black box (https://en.wikipedia.org/wiki/555_timer_IC). Although again I do realize now that the author was primarily attacking the use of "Ti" here.
No, your parent poster is right. Noise-tolerance, induction-safety, aka as working in every case as specified in the data-sheet - are part of the core-product. That is not just some luxery quality issue, this is core functionality. Without it, the chip may fail when placed near some EM-Sources onboard or go haywire on hot days. There is more to Layouting, then connecting A to B.
When we "recycle" our e-waste, much of it ends up in large landfills in China, where very poor workers melt down old boards, remove chips, and sort them by size and pin count. They then use a machine to scrub off the label and print on a new label of whatever part they want to market (usually obsolete parts that can't be bought from the original manufacturer). These end up in the supply chain of even reputable distributors. Amazingly, some of these counterfeits have made there way into military equipment, and as a result, there's a whole division of Homeland Security that conducts tests on suspect computer chip shipments entering the US.