Gentlemen, gentlemen... It seems like everyone is bemoaning a problem to which we all understand the true fix, but no-one is willing to make the first move towards. A severe degradation in the quality attributed to ones company products is the present cost of this unwillingness to move production of parts and assemblies either on-shore or to a more reliable (and expensive) environment where quality is assured. The mega-dollar question is whether quality is worth the higher cost of this change and how it will impact company futures. In the mean time caveat emptor for those who are making stuff overseas in several countries known to engage in the counterfeiting practices.
Counterfieting of elctronics products is very common because it is very lucrative once established.
I feel the analog ICs percentage is higher just because they are in existence much earlier than their digital cousins and its relatively easy to find swap parts. With digital ICs its hard to find compatible parts because of their nature.
The counterfeiting business is more or less tied with the regions where genuine parts are manufactured. Once a paricular IC's data sheet is matched for quiescent conditions the fake is ready to be on the shelf.
As a former Division GM of an international IC company invoved in Analog IC, I can tell you what is my experience on the subject.
One of the problems is the managementof the wastes. It is very easy for a supervisor of the night shift to put apart the ICs falling into a minor defect bin, or the ICs rejected for minor cosmetic problems, and resell them on the black market, which is, in China, very well organized. Just consider the money one can make selling ~1000 parts ( the rejects during one single night for a given IC) at 20 or 30 cents, when his salary is just the equivalent of 150 or 200 USD/month.
It is possible to control that-we did it- but it requires a very strict discipline enforcement, which has a cost in term of additional staff and controls.
To solve this problem, the first step is categorizing the different types of counterfeits and the relative quantities of each that are making it into the supply chain.
By different types of counterfeits, I mean IC packages that contain one of the following: (1) salvage (waste) die that was meant to be scrapped/recycled; (2) Commercial grade die but package markings indicate Industrial or Military grade; (3) Substitute die (different from package markings); (4) Used parts sold as new -- and any other categories I might have missed.
The first group is one that the industry should be able to control as part of its waste management procedures.
The second group most likely consists mostly or only of packages that have been re-marked, so perhaps a more thorough physical inspection could reveal the fakes.
The third group probably consists only of analog ICs. This is the most insidious and difficult group to detect, because here the counterfeiter is literally doing his own wire bonding & packaging, substituting the die of his choice with similar functionality (e.g., cheap op-amp instead of expensive one).
The fourth group, like the second, might be detectible by more thorough physical inspection. An IC that has at some point been soldered onto a board should have some tell-tale signs.
Things like re-labeling or selling used parts as new are easy to comprehend, but it boggles the mind to imagine that there are criminals out there who could manufacture ICs in categories #1 & #3. That means they have access to real IC assembly facilities -- which one would assume are all legitimate businesses that would never agree to assemble counterfeit ICs.
It seems to me it's probably both. Analog parts, especially op-amps and comparators are probably easier to counterfeit. Many applications can work, sort of, with a generic op-amp or comparator: not all specs are critical to all applications. For an AC-coupled audio amplifier, an out-of-spec DC offset voltage might go unnoticed. However, in an application where the input offset voltage is critical, the problem becomes obvious. There are enough such applications that the counterfeit parts are detected and reported more often... but probably not as often as they find their way into products.
I think analog parts, such as op amps and converters, are "easiest" to counterfeit or substitute with a lesser part because the fucntionality is the same with the fake as the genuine--only the full-range performance may suffer. A "decent" counterfeir op amp works the same as a really good one, except at the extremes or in some specific parameters.
And most users don't have the time or expertise to fully test these parts for specs such as bias current, drift, or linearity on the bench--and certainly can't afford to do it in production. They count on the vendor's part meeting the operational parameters, especially after they have qualifed the prototype and first run and are satisfied that all is well.
Dylan - the enormity of the problem is really troubling. Do you have any insight on why analog parts (as well as the other top five) are the most frequently counterfeited? Or, are they just reported on more frequently?
Didn't know how prevalent this was. It almost feels like businesses need to vet through all their chips, just to ensure they are in working order. Regardless, this is a pretty crazy situation.
Member of the semiconductor group at Patexia:
Blog Doing Math in FPGAs Tom Burke 2 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...