Chip Re-labling


This Web page last updated Mar 7 2024. To email me, see see my ordering Web page for my email addresses.

Background:

There was a discussion in the email list for the VCF (Vintage Computer Federation) Mid-Atlantic USA group, titled "Fake 7400-series memory". It was about TTL ICs which were remarked with the wrong part number. The back story for the initial post is below, with Web links. ON this Web page, I captured that discussion, and the back-story, and added additional commentary from another email discussion group and private emails.

Robert Baruch started a chip-engineering project in early 2017 called Project 54/74. He examined TTL-class chips at the die level, imaging their dies to reverse their design. He obtained an uncommon 74LS189 RAM chip from a Chinese vendor, decapped it, and imaged it - and saw it was not a RAM. Someone not part of Baruch's project, Ken Shirriff, did a detailed analysis and determined it was a DTMF chip, most likely a MK5089 - with the same 2-digit number as the 7400. Baruch says he contacted his ebay vendor, and obtained a refund with the claim the chip was damaged in shipping (!?).

This Web page, I've collected and edited portions of the email discussion. My edits are in []'s. I've also added information from others who obtain vintage chips from "Chinese / Far East" vendors or sources, who obtained similar information about the how and why of these relabling practices. Lee Hart, an old friend and fellow EE, provided some additional opinions on the fallout and moral hazards of these relabeled IC practices. THose interested, can Web-research how relabeled and mis-graded IC's have found their way into military and critical-mission equipment and may have or will become real problems, outside the hobby or vintage-computing worlds. - Herb Johnson

Who is Herb Johnson? I provide a lot of Web pages, about vintage computing preservation and restoration. So I work with obsolete IC's all the time. Some vintage computer owners and restorers, insist on original parts, down to date codes. Or, their computers depend on specific functions found in only one brand or model of IC. So, mislabeled and relabeled chips matter. Check my Home page about the vintage computing I do.

Recent information on remarked ICs

2024: remarked PROMs

Lee Hart has produced 8-bit vintage microcomputer kits for over a decade. They use old-school UV erasable EPROMs, all DIP pins of course. In March 2024 he accounted to me problems of these EPROMS running at as-marked speed and his findings. - Herb

I use 27C256 EPROMs on several of my [8-bit microcomputer] products. I used to get them from BG Micro, then Rochester Electronics, but they no longer have them at a reasonable price. I ordered some from a China eBay seller. They are marked SGS M27C256B-10. Most of them work, so it seemed like a reasonable source. This particular vendor had been supplying me with chips that were *not* re-marked. One can tell because the markings are fainter, they have a variety of date codes, etc.

But I've had customers complain about "bad" EPROMs. When they send them back, they work fine for me. Today, I tracked down the problem. The marked part# says they are 100 nsec access time. They are not!I measured about eight of them, and found the access times vary from 200-300 nsec. So the Far East are re-marking EPROMs as well (harder, since they are ceramic, see update below). So it appears this vendor too has now supplied me with re-marked chips.

My recent 1802 processor kit allowed 250 nsec access time, so half of these parts didn't work. I'll have to sort them and use the slower ones in something else. Being the suspicious type, I checked some EPROMs marked NMC27C32BQ-200 from another far-east source. Their access time is good (200 nsec or less), but they are _not_ CMOS. You can tell by the idle supply current. CMOS parts draw essentially zero current with all inputs tied to VCC. NMOS parts draw 30-50mA with all inputs tied high. - Lee Hart

update, remarked ceramic PROMS I asked Lee Hart about looking at the chip die through the quartz window: I *did* look at the SGS-marked chip inside with a microscope. [They are marked TI inside]. They are identical inside and out, both slower and faster parts. Most EPROMs are manufactured, tested, and marked with whatever speed they pass. A little Goof-Off solvent removed the paint to reveal that it had been re-marked as well, see photo.

There is certainly a risk of getting bad parts from 2nd- or 3rd party distributors. But most distributing companies are honest; they don't knowingly sell junk, and promptly replace or refund if something goes wrong. The challenge is that some parts are becoming notoriously difficult to find, so I'm forced to rely on Far East vendors. Then it's caveat emptor!- Lee Hart

2020 stories on "fake ICs"

From an electronics trade magazine Design News, "Dangers of Counterfeit Semi Chips" by John Blyer, senior editor, Mar 20 2020. He interviews a director of product marketing at Rambus Security, producers of silicon IP security measures.

trademarks on ICs; sold-as not-original

I came across this Web site index of IC trademark IDs in Sept 2019, while looking up some clearly remarked chips. It's a useful reference to identify some manufacturer (or reseller) of IC's. Turns out in the case I was considering, a well-known reseller of obsolete IC's has their own trademark "brand". They apparently retest, re-mark and resell some ICs as just that - so their customers know these are not being sold a particular brand or particularly qualified part - and not an attempt to decieve. The discussions below, describe that and other considerations about re-marking IC's. - Herb

Tutorial on "counterfit" ICs

In Sept 2019, I found this PowerPoint like presentation from 2013 about detecting "counterfit" ICs by Components Technology Institute; an engineering consulting company for the military/space trades. It's a good summary of marking practices, erasing and remarking, and some means to detecting these practices. Most of us won't microscope or X-ray chips, but it's informative nonetheless. It's also a tutorial about designating "counterfit"; the authors suggest any remarking is a deception and therefore a counterfit results.

My introduction to remarked ICs, 2017

Fanzine article on "Chinese chip relabling"

Dan Roganti, Feb 19, 2017: Speaking of custom chips, there was another article about this topic, regarding the [Ebay] auctions from China. Wondering of anyone else saw this. [Jan 2017 issue of "Kilobyte Magazine", page 16. See issuu.com, a Web publisher of private Web-based magazines, for the publication.]

The cause of the fake parts, which has been a hot topic for many years, is the result of the mistaken [re-]identification of components. [This is due to] wear and fade of the inked part number, after 35+yrs while sitting in warehouses, and also from just cleaning the parts. Most of the parts are marked correctly, and they apparently used fake datecodes to distinguish the re-stamped parts.

Presumably they acquired all of these computers from the sales during liquidation or simply just e-waste shipped overseas. Some of the part suppliers in the article, who buy these parts by the thousands directly from these sellers iso [eBay], vouch for these sources. - Dan

VCF discussion

Dan Roganti, Wed Aug 23 2017 "If you noticed both [the original and relabeled] part numbers end in '89'.? I posted in Feb about an article [as above] that discussed the notion of fake parts. If you read the article, they interviewed a parts dealer which works directly with the China dealers. [And so] you will always find fake parts, because the salvage business is not perfect."

systems_glitch Aug 24 2017, has one explanation: "As Dan mentioned, this is probably the result of a mistake in a legitimate, but often poorly communicated, Chinese salvage operation. There are shops in China that specialize in refurbished ICs, which may be old stock, board pulls, etc. Typically the old chips get blasted clean with a mild abrasive (often ground walnut hulls, I'm told), dipped in flux, tinned in a solder pot, then re-etched with a current production datecode and the part number. It kind of loses legitimacy when they put a manufacturer's trademarked logo on the IC, though.

Having bought many hundreds of refurbished ICs from China -- usually things that have been out of production for years -- I've found that most of the sellers and brokers have a hard time articulating that the chips are refurbished in terms that your average English-speaking purchaser would understand. There are of course outright frauds.

As an example, I recently purchased 200x Motorola 6821 PIAs for the runs of reproduction OSI boards we're working on. They were refurb/relabeled ICs with a 2015 datecode. They came with the Motorola logo etched into the top, but the seller was very clear that they were a mish-mash of manufacturers and that the ICs were solder pulls, recycled from old boards but tested after refurbishing. This was 100% acceptable on hobbyist boards, and so far every single PIA has passed test.

So, how can they sell PIAs so cheaply? I suspect they're being paid to take eWaste, subsidized by the Chinese government to not just dump it in a pit, subsidized on the tax/shipping since it's an export business, and they are in the end selling the IC for money. It's like being paid to do scrap cleanouts, but with government assistance along the whole way. - Thanks, Jonathan"

Dave McGuire, Aug 24 2017, presents another scenario. "As far as selling fake 74LS189s, look at it this way. Try not to apply American labor economics to the situation. There's a market for 74LS189s. It's a very small one, but it does exist.

A Chinese businessman comes across a few hundred thousand completely useless chips, like DTMF generators. A few searches on completed eBay listings tells him which 16-pin DIPs sell and which ones don't. Chip labeling equipment is easily reconfigured on the fly (think date codes), so he makes a few hundred "74LS189s" and puts them on eBay. And a few hundred of these, a few hundred of those...and of whichever ones sell, he makes more.

And while he's putting them on eBay, he's also shopping them around to companies that perform equipment repairs, to the military parts channels (through a front), etc.

Just think of what you could do in business if you had access to a labor pool that is endless and essentially free, all of the cast-off equipment that American corporations are liquidating when they shut down factories, government subsidies, lots of greed, and a complete lack of scruples. That's half of the businessmen in China today.

This is all extensively studied, well-documented, and is not a new problem. It is a very large problem in the electronics industry today, and not just because of occasionally inconveniencing a hobbyist. We have fake components getting into military supply chains by the tens of thousands." -Dave

Discussion with Lee Hart about relabeled chips

In Sept 2017, I showed the above discussion to Lee Hart. He's an EE with decades of design experience. Lee is a current designer and producer of small digital devices, often intended for the hobby vintage-digital community - TTL logic, not Arduino, etc. Here's his comments and opinions and some dialogs. - Herb

I see info on this topic from various electronics trade journals, as well as from an occasional hobbyist who got "burned".

There are honest businessmen in China, as well as incompetent ones, and crooked ones. Like many things, if it sounds too good to be true, it probably is. The guy who finds an incredible deal on some rare part is more likely to get burned, than a repeat customer who buys in quantity and maintains a dialog with the seller.

I've often ran into mis-marked and re-marked ICs; though rarely obviously bad or wrongly marked parts. That may be because I'm more cautious. There are only a few Chinese sellers I will buy from. When I've had a problem from them, they fixed it. It seems as if they buy parts from other dealers, and may get bad parts themselves sometimes.

For example, I asked [a particular supplier] if they had any Hitachi HM66256BLSP 32k RAMs (the "S" means in the skinny 0.3" wide package). They said yes, they can get some, and sent me ten standard 0.6" wide HM62256 RAMs that were *remarked* BLSP. I complained, and they said, "Sorry; Our supplier did not understand." It took a few weeks more, but they replaced them with the correct part (which worked, and physically matched some real Hitachi HM62256BLSP parts I already had).

Ah, well. There's that Chinese curse "may you live in interesting times". - Lee Hart

Another example: 65C02

Additionally, in Oct 2017, Lee and I discussed his purchasing of quantities of obsolete Rockwell 65C02 processors. They have unique features over conventional 6502's. - Herb]

"[After I manage to find them,] THEN I have to test them. The 6502 in particular: I'm using the Rockwell R65C02 CMOS variant. It has a few unique instructions, and I use software which depends on them. It also has a couple hardware features, such as a built-in clock oscillator and of course it is CMOS so it [is low power and] works on batteries.

I bought [many] from [a well-know parts distributor]. Then I got them from Chinese ebay vendors; but [those] turned out to be random 6502-class chips with the markings removed and "R65C02" printed on top. 1/2 to 3/4 of them work. I have also been getting them a handful at a time from other distributors, surplus houses, and ebay.

I've had better luck with the Chinese 1802s. They're more likely to work, probably because I'm not doing anything "interesting" with it on the Membership Card, so any age/brand/variant of 1802 will work.

What I see on the Chinese 1802s and 65C02s are ICs with identical absolutely perfect markings on top; but many minor variations in the physical package. And, lots of variations in what is marked on the bottom of the chip.

But [my 65C02 use] depends on a CMOS version of the 6502 with the Rockwell enhancements. If I had time, I might be able to figure out what the "duds" actually are. Probably chips from scrapped Apple, Atari, and Commodore computers that had various 6502 variants in them." - Lee Hart

Another example: 1802

Here's notes from Ian May in 2018, about production marks on 1802 microprocessors. The COSMAC 1802 was a CMOS processor produced by RCA in 1975. They licensed it to other manufacturers over time. There's a variety of printed markings on IC's like the 1802. There's also unique "injection mold pin" marks, which can't be sanded off of course.

Implications of relabeled IC's - Lee Hart

later in Oct 2017, Lee Hart had some thoughts about the problems of relabeling ICs, which he wanted to discuss and include. Here's some of our discussion and his edited remarks. - Herb

Herb: My reading of the information I sent you, is that these Chinese sellers who re-mark used chips, have the idea that they are selling chips "close enough" to what the markings (used to) say. They consider it, their version of "normal" business practice.

Lee: Yup. They don't know (or don't care) that there are differences between chips, and customers have reasons for wanting a particular brand or sort level. The speed of a memory chip, or the accuracy of a voltage regulator, or the input offset of an opamp, or temperature range of a micro. Or it's going into something that requires traceability for agency approvals.

Herb: Apparently, they can mark chips in arbitrary numbers, immediately upon request. I can't prove that, but your remarks suggest it's possible.

Lee: It certainly seems possible. Some will even tell you that they'll mark whatever number you want on the chip! I suppose the take-away is, "what you get may not be exactly what is advertised; but it should be close enough for hobby purposes".

For example, let's say you buy a tube of "74LS00" from [a major distributor of new and used ICs]. The tube might actually contain a mix of TI SN74LS00, Motorola MC74LS00, National DM74LS00, or any of a dozen other brands that made these chips. [But,] they will each be marked with the [original and] correct part number, and all of them should work interchangeably (in *most* circuits).

Some Chinese suppliers just take this one step farther. Their tube still has a mix of brands; but they may have *re-marked* all of them with the same number. This makes them look nice (like new parts). It will be OK for hobby purposes; but its a problem if you need a *specific* brand, or are using them for commercial purposes that require tracking and authentication.

...and the next step, is not a mix of 74LS00 chips, but a mix of 74S00, 74F00, 74HC00, etc. Or possibly 4000 CMOS because the number ends in "00" - like the specific example detailed on this Web page. Only physical tests of die, or close examination of chip signals, will reveal that level fraud.

Yes, that's true. Some sellers are clearly dishonest. But do not attribute to malice what can be explained by simple incompetence. I think many just don't know any better, and are only out to make a fast buck.

[So,] if you're making a product [and not some one-off for yourself], fakes are a problem! But if you're a hobbyist, you probably care mainly about "cheap", [So,] you get what you pay for. The cheaper it is, the more suspicious. If you get a bad or counterfeit part, and complain to the vendor, most of them are honest enough to replace it. Note that if you don't complain and just write it off, you are *encouraging* bad vendor behavior.

But who *knows* if relabeling correlates with price? Some reseller can buy cheap and sell for whatever, right? Right. And, the cheap relabeled chips also get bought by outfits in the US, who essentially "launder" them by reselling them as new. And the buyer doesn't test, to save money and because he assumes he's getting prime parts. Every businessman along the chain is committing a "little fraud" to make money. So an old salvaged IC winds up in a critical safety circuit, which fails and kills someone." - Lee Hart


Copyright © 2024 Herb Johnson for this edited content. Other extended quotes may have copyrights by the original authors.

Herb Johnson
New Jersey, USA
to email @ me follow this link