The Global Chip Shortage is Creating a New Problem: More Fake Components (zdnet.com) 72
Industry analysts believe that the global chip shortage is creating the perfect environment for counterfeit semiconductors to enter the market. From a report: With demand looking unlikely to calm down, analyst firm Gartner estimates that the semiconductor shortage will last well into 2022, and has warned equipment manufacturers that wafer orders could come with up to 12 months of lead time in the coming months. For some companies, this will mean finding an alternative way of stocking up on chips or shutting down production lines. In other words, the current times are opening up a golden opportunity for electronic component counterfeiters and fraudsters to step in. "If next week, you need to get 5,000 parts or your line will shut down, you will be in a situation of distress purchase and you will put your guard down," Diganta Das, a researcher in counterfeit electronics at the Center for Advanced Life Cycle Engineering (CALCE), tells ZDNet. "You won't keep to your rules of verifying the vendor or going through test processes. This is likely to become a big problem."
As part of his research, Das regularly monitors counterfeit reporting databases like ERAI, and although it is too early to notice a surge, he is confident that the number of reports will start growing in the next six months as companies realize they have been sold illegal parts. The problem, of course, is unlikely to affect tech giants whose reliance on semiconductors is such that they have implemented robust supply chains, and will typically only purchase components directly from chip manufacturers. Those at risk rather include low-volume manufacturers whose supply chain for semiconductors is less established -- but it could include companies in sectors that are as critical as defense, healthcare and even automotive.
As part of his research, Das regularly monitors counterfeit reporting databases like ERAI, and although it is too early to notice a surge, he is confident that the number of reports will start growing in the next six months as companies realize they have been sold illegal parts. The problem, of course, is unlikely to affect tech giants whose reliance on semiconductors is such that they have implemented robust supply chains, and will typically only purchase components directly from chip manufacturers. Those at risk rather include low-volume manufacturers whose supply chain for semiconductors is less established -- but it could include companies in sectors that are as critical as defense, healthcare and even automotive.
Also rejected components? (Score:5, Interesting)
I've been handed systems with fake chips as well, that were apparently originally real components but had not passed quality control. Has that increased lately?
Re: (Score:3)
Re:Also rejected components? (Score:5, Informative)
It won't be Ryzens, they are too valuable. AMD will take the ones with defects, disable the defective parts and sell them as a cheaper SKU. That's all the different SKUs are in many cases, one chip has some bad cache so they disabled half of it and sold it as a low tier part, another runs great at 5GHz so they sold it as an AF edition.
Where you get remarked parts causing problems is things like EEPROMs with some failed cells. It might not be apparent unless you do a 100% memory test, which takes time during production. Even if all the cells are okay it might have failed in other subtle ways, like drawing excessive current during writes or they detected some die defect that means it will work for a while but not get anywhere near it's 100k erase/write cycle specification.
Another one is marking inferior parts as more expensive ones. Say you have a voltage reference rated for 0.1%, it could be a 0.5% one re-marked and the only way you would know is if you measured it after installation on the PCB, or detected measurement errors in operation. If you already had 10k PCBs with the re-marked parts at that point you might try to find a firmware fix with calibration step, rather than reworking every board.
Re: (Score:2)
Re:Also rejected components? (Score:5, Informative)
That's actually a myth. There was a report in a British newspaper about it, but it's not true. The problem was not counterfeiting or stolen formulas, it was just the supply of raw materials being sub-standard.
Dell had a lot of problems back then, e.g. the fault Nvidia chipsets that tended to die after about 12 months. HP was even more badly effected since almost all of their machines had Nvidia chipsets.
Re: (Score:3)
Re: (Score:1)
Re: (Score:2)
Re:Also rejected components? (Score:5, Insightful)
Fake parts come in several varieties.
First are rejects - parts that didn't pass their tests are discarded, but instead of being crushed and recycled, they're harvested. Most rejects aren't rejected because they don't work, they usually just barely work. Example, they may work just fine at room temperature, but fail once it gets hot.
Second are functionally identical parts - these are usually turnkey parts. You may remember fake FTDI USB to serial adapter chips a few years ago. These are functionally identical - the fake chips are a completely different architecture internally but work the same when hooked to a USB and serial port.
Third are outright substitutions - for chips where a part may function similarly to another part, the parts are re-marked. Example, transistors - two transistors may have similar specifications, but different part numbers. But maybe a parameter keeps one from being an easy substitute to the other. Nonetheless, unscrupulous dealers will use one and pretend it's the other.
And then there are outright fakes. Take one popular chip, find one with similar packaging and then alter it so it looks like the popular chip. No surprise, these chips don't work - because they're the wrong chip. They just look like the one in demand. The hope here is to sell it to some collector who may not actually use the chip, or for someone stocking up their parts bin and may never actually get around to using it for years.
Re: (Score:3)
The last one is a distinct problem but it has been growing in the DIY PC space. There have been some well reported instances of supply chain tampering where shipments of CPUs will have the real ones stolen and replaced. Sometimes the thief's will put in effort and relabel an older chip and swap them. Sometimes they will just toss in a print out of a picture of the chip taped to a weight. They just need it to make it to retail so they can't be caught.
Re: (Score:2)
Substitutions are an interesting one.
Taking FTDI as an example, a lot of companies have switched to using cheaper parts from other companies that are more or less functionally identical. Years ago drivers were an issue, the FTDI ones had a few advantages over the standard Windows CDC driver, but not these days.
So why bother substituting if they are functionally identical and work fine? A lot of it was down to the perception that these new Chinese brands must be crap, so pay 5x more for an FTDI part. Now tha
Re: (Score:2)
"Functionally identical" is a very, very dangerous phrase, popular among purchasing agents who skimp on specs and lead to enormous issues down the line. One of my favorites from long ago in my career was when somebody had taken out all the ceramic capacitors scattered all around the board and put in one larger tantalum capacitor at the bottom of the board on the new design. To salvage the boards, someone on site had to pull the new boards, manually solder in ceramic capacitors across the same leads on the n
Re: (Score:2)
Well, let's say you are outside the legal jurisdiction of a customer. Just take the money and ship a box full of sand. Or if you're subtle, just ship normal chips that aren't in low supply instead. Then spend the money. You don't actually ever need to send a part that pretends to work, or chips that failed QA, etc. For the ultimate in criminality demand payment in bitcoin.
Re: (Score:2)
Wherever someone thinks he gets away with it, and the marge is high enough, you have counterfeit products.
Re: (Score:3)
Wherever someone thinks he gets away with it, and the marge is high enough, you have counterfeit products.
And with our lives being chip-based these days, perhaps counterfeit punishments need to increase considerably.
Of course, problem with that is you'd have to start with distributors like Amazon, a card-carrying member of the Donor Class...
Re: (Score:2)
You can increase punishments in your own country, but often the counterfeit products come from countries that knowingly protect the criminals.
Re: (Score:3)
Why do you point to China?
Because China is the main culprint - not the only one, though. Amusingly some of the Chinese fabs actually include bug fixes for errata that never made it into the silicon parts they're ripping off, such as the CS32F103 from CKS.
Everybody is a potential culprit. Back in the Reagan era they deregulated the aviation parts market. That went well for while until an airliner fell apart in mid air thanks to fake parts https://en.wikipedia.org/wiki/... [wikipedia.org]. That incident, among others, led to an investigation in the US that discovered the market was awash with fake aircraft parts and that counterfeit parts had been found in Airforce One. Recently the US airforce was being plagued by fake Chinese chips: https://www.themanufacturer.co... [themanufacturer.com] and ap
Re: (Score:2)
Re: (Score:3)
>we had 487 chips rewired to run as 486 etc.pp..
That's impossible. The 487SXs were ALREADY fully wired DX chips, that took over ALL CPU processing once it was installed.
Also, there was no need of course for a 487DX, as the 486DX included a math coprocessor already.
As per wikipedia:
"The i487SX (P23N) was marketed as a floating-point unit coprocessor for Intel i486SX machines. It actually contained a full-blown i486DX implementation. When installed into an i486SX system, the i487 disabled the main CPU and
Re: (Score:2)
Re: (Score:2)
How?!
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
There are other ways to create counterfeits. One common thing is to make more wafers from owned lithos, completely off record, and maybe without all the process controls in place, then ship those to a third-party packaging company that then cooks up finished product. Really hard to tell the difference in any way until the products get into wide use and then the failures crop up. Another technique is making copies of the lithos, and then dragging them to another foundry. Variation from one foundry to another
Re: (Score:2)
The best place to create a counterfeit is in the factory where the genuine part is made. The only place to obtain reject parts that can be relabeled as good is the place where the genuine part is made. Often, but not always, that place is China.
China's wild west business ethics and lax regulatory system means Chinese production costs are lower, so it's a financially attractive place to offshore production to, but you also take the risk that dodgy side hustles might be a factor in a supplier's low bid. C
Discovered just this today!! (Score:5, Interesting)
Re: Discovered just this today!! (Score:1)
Re: (Score:1)
Electronics nerds are the worst of all types of nerd pedants.
Re: (Score:2)
It's this kind of discourse that draws me to /.
Re: (Score:2)
My FET fakes bring the nerds to my yard...
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
I had a batch failure like this. We were making miniature radio transmitters, and generally got very consistent results on RF output. Then on one batch, almost all transmitters failed, due to low RF. Turns out the current gain of the RF transistor was lower than normal on this batch. Unfortunately for me, the defective transistors were still just within stated specs. I scrapped the whole batch, and bought a new reel of transistors from a different manufacturer, and everything was fine again. I had always su
It really depends on the component (Score:4, Insightful)
Some can be counterfeit, others really cannot. For example, an MCU can basically only be copied and it requires an unscrupulous Fab for that to work. The component will simply not work if it is not a precise copy of the original design and that will be immediately obvious. Power semiconductors, on the other hand, can be replaced with ones that have worse specs for reduced lifetime and reliability.
Re:It really depends on the component (Score:4, Interesting)
Counterfeitung is a big problem for STM32 MCUs.
Often, the fake is actually an STM32 clone (they tend to be fully compatible for the core, but have differences in the peripherals), that had its markings sanded off and reprinted to look like an STM32. Sometimes is is actually a lower-end STM32 with markings changed
While the problem has been around of a while [hackaday.com], it has gotten worse lately, to the point where some report that it has become virtually impossible to get an original.
Re: (Score:2)
I doubt that someone is cloning an entire MCU, that would be difficult and costly (you would have to have a team of silicon experts, photograph the MCU, slice off the silicon layer by layer to figure out how it was contructed and then label millions of transistors) . Much easier to repaint.
Re: (Score:3)
People clone STM32 MCUs.
The CS32F103 is a Chinese clone of the STM32F103. AFAIK the main differences are the marking and that the CS32F103 fixed some of the errata still present in the original.
The GD32F103 is another Chinese clone of the STM32F103. AFAIK the main differences are the markings ands that it supports higher clock speeds and has higher power consumption than the original.
As long as they do have the correct markings, those are examples of rather good clones. But there are others.
Re: (Score:2)
That isn't correct at all. MCUs are often counterfeited. They typically use other cheaper MCUs to do the heavy lifting and fake the model code. You as the programmer are none the wiser until your device doesn't work. You can even attach a debugger and see that you're correctly writing registers but for some strange reason function X just isn't working. Yeah function X isn't present and that register you're writing is doing something else entirely.
Digital components are faked all the time. The most classic e
Re: It really depends on the component (Score:1)
Re: (Score:2)
You did not understand my statement. The problem with, for example, fake power transistors is that they work for a time. A MCU will fail immediately.
Re: (Score:2)
That is also completely untrue.
Both fake transistors and fake MCUs can work completely fine, or can fail early. The only thing that really is common among all fakes is the inability to meet their exact specs, and that is common to MCUs and transistors.
Re: (Score:2)
You still do not understand what I am talking about. Well, no matter. You have demonstrated cluelessness in the past.
Re: (Score:2)
I’m pretty sure Disney would argue that the MCU is counterfeited any time a new torrent is posted for a Marvel movie.
Re: (Score:2)
Ah yes, the Counterfeit but also copyright infringement argument. Only the best of lawyering for disney
Re:It really Does depends on the component (Score:3)
I used to design Gamma Detectors, using PMT's. :) :)
We used Zener Diodes in the circuitry, and suddenly They started failing by the systemfull.
It came down to a Zener diode that cost less than $00033., that's $33 per 10,000.
A rectifier diode just happened to break over at 200v, and they used those, remarked to our part.
Not being zeners, they failed rapidly. Different internals, not suited for the purpose, so they failed fast.
I did learn something from it: Zeners and rectifier diodes have opposite temperatur
Re: (Score:2)
Fascinating. Some people are just completely broken and will do massive damage for a tiny profit.
Re: (Score:3)
Once you have a working product, there will be a team or project involved in cost reduction. It doesn't matter that you sell the software for $1 million, they will demand that the profit margin improve on the devices by 5 cents. Sometimes it seems like the cost savings doesn't even cover the cost of all the labor expended to do the cost savings.
So, sell a game console that comes with a free copy of Tetris. Someone will try to remove that Tetris because it's costing thousands of dollars a year in revenue.
Garner? (Score:1)
Are they still around? There are still people who believe they don't simply make everything up?
Re: (Score:2)
Everything is made up, including the words you use. The question is if someone agrees with it, and the basis for saying the things you do. Now I assume since you're taking the moral high ground you can point to a study on the accuracy of Gartner's claims over time showing that they "make everything up"? Though considering you can't even spell the company name...
Bid (Score:2)
Sabra Price is Right
Chris Farley: It's just that I've never heard of the Pinnacle corporation before.
Tom Hanks: Is Sony guts, is same thing.
Not New (Score:2)
This shall cause nightmares for IP lawyers (Score:1)
Semiconductor fabs (Score:2)
Having spent one career working for a (then) major semiconductor company, I learned that our semiconductor libraries and processes were so poorly documented and designed that we got a fair amount of variation sometimes even over one wafer. This was for our analog RF/IF parts (cellular handset front ends, PA chains, and I/Q mixer, oscillator, and baseband blocks), where small changes can have large impacts to performance. Die were screened on wafer to see how good/no-so-good/bad they were, then diced and pac
Re: (Score:2)
I am actually surprised how inept and disorganized many hardware engineers can be. People in software may bemoan that they're just wannabe engineers, but usually they will actually have a process in place for software design and maintenance. For example, software people know enough to use a code repository and why it's important; whereas I have seen many hardware engineers writing VHDL code and not understanding the need to have a repo for it, or what branches are for.
Cheap tat horror story (Score:2)
One of my customers made electric car door locks. Small electric motors drove the mechanism. These were basically the same kind of motors used in toys, but adequate for purpose, as long as they maintained a consistent performance. One time, a batch of motors arrived from China, and they just did not have the oomph like the old ones. This caused major ructions. There was no chance of passing off the defective locks. I was told that a whole container load of motors was quarantined. The problem then was provi
Similar to the .com boom (Score:2)
Back in the late '90s, there were stories very similar to this. People shipping "chips" that were just plastic. When someone proved that there wasn't any silicon in them by melting one, the supplier said that they had voided the warranty by applying the heat to the chip. What's old is new again.
Glad I stocked up... (Score:2)
...back in the 2000's I went around and purchased up old electronics stores going out of business. I scored millions of components from their basements for pocket change.
I guess that stock must be worth a fortune now - not a single counterfeit in that collection, just NOS components from factories and stores going bust.
Brazilian Advice (Score:1)
Very common (Score:2)
I know of a manufacturer who had some products made overseas for the first time.
All good until they started failing after 6 months - well over 50%. Unfortunately the product was installed in facilities all over the place and the manufacturer had a guarantee in place meaning they would cough up for the replacement cost INCLUDING removing and re-installing a new one [it was built into the high retail price as part of the offering but they didn't expect 50%].
That was a very expensive lesson. Turned out the man