Somewhere I was watching a chip be x-ray'ed(?) to spot new chip dies, it was fascinating. Not sure if he was you plgDavid, but I watched your video in entirety, and certainly enjoyed it, thank you!
Of course it's only truly a fake if the underlying chip itself is a fake. Taring it and remarking is only somewhat of a fake, more of a deception if you will. Used, abused, tossed around, overheated, etc.
haha, this reminds me when I worked for a very small military subcontractor who painted reliability bands on resistors when they couldn't get the hi-rel yellowed striped required by the contracts. When they were washed in solvent they disappeared. Right in front of the government inspector.
Well if you know your products will be used to protect people ,fi, medical, or in military and you are acting like this you really gambling with lives. Only the guillotine will be the place. Bare in mint you are in a plane with those fake parts! Don't tell me it was an order from upperhand. YOU also collaborate. I really can not understand how dumb people can be. The stuff mentioned here in this film is game stuff and as people only desire "original" they get it, no physical harm is done. Thats the difference.
What you are referring to as "ejector holes" are typically called "ejector pin marks" made by the mold ejector pins as they push the part out of the mold. The "code" markings on the bottom of the chips are usually for manufacturer reference and can be set to anything. Molds wear out, so by putting an identifier mark, they can identify and track mold life. It's also common for identifiers to be put on multi-cavity injection molds, so if there is a defect in one of the cavities they can locate it easily. Still excellent video.
@@plgDavid Not really, as a kid many years ago, I spent a summer as a laborer in an injection molding factory. Horrible job, it's hot, wet (large parts need to be put in clamping fixtures and then in cold water baths), itchy (glass fiber filled plastics), toxic (burnt plastic from clogged nozzles makes your eye's water uncontrollably and mold release is an aerosol that for some reason appears to be neutrally buoyant, so it just floats in a cloud around you) and that's on top of the super monotony. Laborer turn-over rate was 7-working days on average. Most would just walk off the job and the machines don't stop, so there would be a small mountain of plastic parts where they stood. I remember using a snow shovels to clean up messes. We were a small shop with mostly short duration part runs. Sharp's setup would be have been more highly automated I presume.
Why don't they just sell them as reclaimed/re-cycled? we know they aren't being made anymore.. they are legacy chips.. just sell them with the damn scuffs... it's not a problem as long as it's electronically sound! (Pun intended)
Retro hobbyists aren't the main market. They sell to people like military contractors who need new, recently-dated chips which haven't been made in years.
But do you think military contractors let themselves be deceived that easily? I can't believe that especially when even hobbyists like plgDavid investigate and find out.
@@armorgeddon its about the specifications, they don't care if they're old, they just need them to say they're new so they meet 'military grade' and then they can sell them. that's my understanding anyway. it's an everyone benefits kinda deal...aside from hobbyists trying to buy the best chips they can i guess.
@@dextrodemon Makes no sense unless they hope to ask higher price for unused stock. True military application typically require a legitimate supply chain for the lifetime of the device. Also what do they supposedly use it in the military for?
This is an excellent informative video, and one that I hope more subject matter experts start emulating. One of the things I really appreciate is how you show several examples so that, when it's not 100% obvious, the viewer can try to build a "confidence score" to help judge the actual condition. I see the same thing in my hobby (vintage software collecting), where counterfeit items and even counterfeit *shrinkwrap* can artificially raise the sale value. Thank you for this very detailed comparison.
Yep. Solder pulls. Heated with a blowtorch and dropped out. Re-stenciled. Pins bent back out at a splayed angle, and re-tinned. But for stuff such as 74C915, a rare hen's tooth, these Chinese solder pulls are the only game in town. It's a good thing these ics were so over-designed, because these animals heat them way beyond any temperature they were ever meant to endure. Then you get these off of Ebay and only maybe 80% test good. We are damn lucky to have access to them at all.
Wow, this video brings back memories... Back in the 1990's they had a ton of fake L2 dram chips. (Before it was integrated on the cpu) When I was building PC's I was testing them and discovered many fake chips. I cut some open. They were nothing but blocks of plastic with legs/pin sticking out. LOL. Seems nothing really changes.
This explains a lot! I have a few customers with cheep audio equipment and CRT rear screen projectors containing STK power chips. They seem to last a year or so then they end up back on my bench with the same chip failure. As far as the audio equipment these these LSICs usually barely or do not meet spec. I have not taken the time to design a test for the convergence chips. All replacements were dirt cheep on eBay or Amazon! I will ask the seller some pointed questions before purchasing them in the future. Thank You
0:29 - Hey, that's my product! The chips I buy are from a seller in China and are definitely pulled. I take some comfort in the fact that they're not cleaned up very well (pins are bent, dirty, etc), so I hope the puller isn't particularly resourceful. I have gotten some chips that just don't work, but it's probably around 5%. I'd guess that a PSG chip like the 2149 just isn't as lucrative to fake. I have another batch of 20 chips on their way, and I'll take an extra look at them. Really enjoyed the video, particularly talking about the recycle operations. I knew it existed but it was really cool to hear more about it!
Yeah, I talked to Trash80 about it as well afaik? AYs/and 2149 were lucrative to clone in the past, there are quite a few AY-3-8910 clones out there, but the YM2149 needs that extra pin logic so its not a drop in replacement. though AY8930s uses the same logic. You might want to send AY8930 commands just in case :) Extra pulse widths and independent noise on each channel.
It's fun to think about how far some of these chips traveled. They were presumably manufactured in Japan, maybe then shipped to China for the assembly of the original product, then maybe sold to the US, then disposed of decades later in an electronics recycling collection, from where it was shipped back to China, where it was then salvaged and relabelled to then be shipped back to the west. I wish the whole recycling process could be done locally in an honest manner, but I guess labour costs are too high in the west.
Oh man yeah, it is crazy actually. I think labor costs are really the trouble indeed, and HSE guidelines and environmental stuff. In China they just do it while here we have all the extra rules and high labor which makes the recycling not profitable.
@Fratti In the 80s and early 90s the original boards would have been assembled in Japan, the US, the UK, Korea, or maybe Taiwan for the more budget items. China didnt do a lot of name brand electronics manufacturing back then.
I'm a big fan of reusing the stuff, so personally, I'm very happy that chips are recycled but this should be super clearly disclosed so professionals like You are 100% aware of what You are buying. In the end, money only talks so this won't disappear and fakes will continue. :( Without this video, I wouldn't know. Thank You for this :)
In 2000? I was removing a friend Intel Pentium 100 chip from his computer and when we were detaching the heatsink ceramic paint came off the top of the chip showing another date code and the fact it was once a Pentium 75!
That's not the same thing....Back then, uPCs got tested for spec at the rated frequency. Some of them met spec at an even higher frequency without problems. Some of those uPCs then got re-marked with the higher rating....
I actually heard back in the 90s' independent computer shops would sell overclocked computer as the real thing. Essentially selling computer with Pentium 100 overclocked to 120MHz as Pentium 120 gaming computer. The heatsink were glued on so impossible for regular people to tell
Most of my contributions came as a result of me buying 43 YM2608s/OPNAs from Ebay and AliExpress for development of my MSX OPNA cartridge. I bought three from a seller on ebay, all were sanded, and one was partially non-working. All were laser-etched with a datecode from 2012, and all, curiously, had the die revision marked as M. The other 40 I bought came from a seller on AliExpress. They had good reviews, most of the buyers were Japanese, and the picture (sadly) was of a beaten-up but legit, non-remarked chip. I ordered 10 at first, all came in, ALL WERE LEGIT, AND ALL WORKED. I cleaned them out after, buying a second batch of 30 more, once again, all were untouched but obviously recycled, some were chipped, some were scratched, I ended up with one having a broken pin, but of the 40 that I bought, only 2 were damaged, and they had the same fault as the one non-working chip as the very first batch of 3 sanded chips. A total loss of 1/3 (33% failure) for the remarks, and 2/40 (5% failure). I got all sorts of date codes, one duplicate, die revisions L, O, and P, but NOT as the first three were marked to say, M or N. On the subject of the old-style Yamaha datecode: That was LOTS of fun to figure out. Being able to find that out and get that published on Siliconpr0n's Yamaha vendor page (other things still need to be corrected there), and in this video is really cool to see. Looking at pictures of chips and trying to spot patters like manufacturers, package types, date codes, what different letters and numbers mean is such a fun thing to do. The people who originally thought of these schemes and know what they 100% mean aren't around any more, either at the companies or, sadly, on the planet. Being able to dig up and solve small mysteries like this all in the name of preservation and documentation is one of the most fun and fulfilling things out there, in my opinion. On the subject of the remarked EPROMS, I actually had 10, not 1. I bought a batch of 10 27C801s on ebay. Large EPROMs are useful and expensive, and while I may not need them now, who knows? I got them in and they were all marked exactly the same, and like you saw on yours, they all had brand new pins, slightly tarnished, factory straight. All were also blank after being read. EVERY SINGLE ONE could be cleaned, and not with acetone, but plain 91% IPA. The most amusing part is that THREE were remarked TWICE. After being cleaned, Three had the exact same marking, in the exact same place underneath, laser etched into the ceramic. That's something that I've never heard of or seen, but is apparently possible and very scary: There is a possibility that any chip you buy might not be re-marked just once, but twice.
I like the idea of re-use but be clear about it. Indeed, one doesn't know how much the chip was hammered in prior use, high-temperature and poor anti-static conditions when extracting.
Interesting video. I used to work in the semiconductor industry. I have seen a case where a customer sent in a complaint about our chip and it turned out to be a counterfeit chip. That was an interesting situation, at that time. I have one detailed comment: at about 15 minutes in the video you explain the date code. And mention that the chip was "fabricated" on that date/week. The entire manufacturing process, starting from a silicon wafer, takes multiple weeks. The date code as printed on the package is the assembly date code. The chip inside could have been manufactured quite long before that and have been stored before being assembled.
Nothing beats comments from actual specialists! Would love to know in your experience are the wafers usually done in the same plant as the molding/assembly or its unrelated? The wafers need to stay in dust/hair/temp controlled env and shipping them out makes little sense right?
@@plgDavid Oh no, the entire manufacturing process is most certainly not done in the same factory, on the same physical location. I guess (but didn't check) that semiconductor manufacturers have promotional videos on their website where they explain the entire manufacturing process. Those videos explain the whole process better than a TH-cam comment box allows (Edit: I found an example video here: th-cam.com/video/M-wNC3Z3ZX4/w-d-xo.html ). Often the entire process can be split into four main parts: (1) get the electronic circuit in the wafers by using e.g. chemical etching and electronic implanting technologies, (2) test each circuit on the wafer and separate into individual dies, (3) put the dies which passed testing into a package (e.g. DIL, SO, BGA, ...), (4) test the assembled product. For logistical purposes do semiconductor manufacturers hold a stock of wafers, either after step (1) or (2). And they hold a stock of final products. Step (1) needs to be done in a clean room to avoid contaminations and to get the manufacturing yield up. The more advanced wafer fabs tend to be in Western countries, whereas Asian countries (e.g. China) have wafer factories which are a few generations behind. Assembly and testing of chips is most often been done in Asian countries (e.g. China, Thailand, Malaysia, etc.).
@@my.own.devices Exactly; CPUs are famous for going around the world several times, every step is performed by the lowest-paid workforce capable of the required quality standards. You'd have the optical transfer to wafer done in Germany (needs precision) and the pins would be soldered on in some hellhole in Malaysia. I was also told by a professor that the silicon crystal seeds are imported expensive, then grown into ingots (?) in medium-standard countries.
27:14 you know you're getting cheap stuff when the packaging is that plastic bag used for Korean ramyun noodles. This looks like nongshim shin ramyun packaging.
Wow, you put an ENORMOUS amount of work into making this video. I can't imagine the thousands of hours of experience, the inordinate number of orders, and just general expertise in a very specialized field that went into producing something like this. You sir, are a scholar and a gentlemen for sharing such knowledge with the world. I hope all the expertise and hours help in building your reputation and business. THANK YOU!
@@plgDavid I meant it! The internet with it's open-source/hacker/builder/creator/etc community never ceases to amaze me. People such as yourself who offer up their time, knowledge, experience, and expertise for essentially nothing but the satisfaction of sharing it with others are what makes these communities such awesome places to be apart of. No one shares this type of info/content for the pennies (if you're lucky) that TH-cam might give you if they decide that talking about this isn't a "sensitive topic". I'm glad you saw my comment as you definitely deserve more gratitude than I'm sure you receive! I'm still in my early 30's and though I still skate a bit, all those years of throwing myself down flights of stairs and handrails definitely gave me "old man knees" much too early in life 😁 I do much more reversing (I'm a reverse code engineer) and programming these days than I do skateboarding, but I will certainly be skating for as long as my poor, abused body will allow me to! Love your content man. Keep up the great work!
Very familiar. This is very well known to be done with certain transistors. All if a sudden we had whole batches failing. It really baffled us because they were just measuring just fine. It became really clear when we decapped a couple. The little metal slug (for transferring the heat) was just half the size of the original. Therefore the transistors were just basically overheating. This was even from a known seller/distributor.
From a molding perspective the letters and numbers on the ejector are usually to help identify the cavity number after the molding process it also helps when rebuilding the mold. You want too make sure the same ejector pin goes in the same location, it's also a standard from particular mold builders or molding shops.
I remember running into fake chips back in the late 90's when I work fro a PCB contract manufacture. They were called "shaved ICs" then. One fake Ic was interesting in our x-ray machine you could see the internals were different. (The ic was an amp and was no longer mad, the board had to get a redesign) Interesting thing even with a legit similar spec IC was would it not run right when being used, diagnostic testing said good. but when system started running it would stop
I don’t know who you are, or how I was recommended this video, I’ve no interest in Audio Electronics, but that was the most interesting 40mins of random TH-cam recommendations in a long time 😃
It’s certainly comforting that the guy that was making some good bucks at home (granted with some unhealthy practices) is now the employee of multi billion corporation that as a middle man will rip the little guys off as well as hobbyists buying this stuff! Thank you for the video!
our factory in China returns on average 20-30% of the parts because they fail pre-production quality control. We have had cases of something getting through with the result being the product when used in low power situations "fails". In that case it was a fake Motorola 4011 logic chip, everything seems to work but I found when the VCC rail had a low voltage (but still within spec) it would cause a .5 volt logic output error. On closer inspection the 4011 chip looked exactly the same as the real deal. When I put it under a microscope I noticed the Motorola logo was poorly done and nothing like the original.
Eugh... I know that where there's demand, supply will follow, but just... don't sell crap like this to people who don't know any better. Sell 'em as recycled or "refurbished" cause I for one would rather gamble on recycled chip than have none at all. Just don't put duct tape on a Rolex and tell me it's brand new =/
your comment will inspire exactly 0 of them to change their behaviour. In fact this video has probably made a few people realize they could make money selling counterfeit chips.
Counterfeit IC and Yahama FM chips are two item of interest of mine, so I was waiting for this video; it doesn't disappoint. Thanks David for taking your time to share your knowledge and expertise on this matter !
I don't know why youtube recommended me this, or who you are, but when I see a man, who is so deep into topic, that he have a book "Counterfeit chips" I put a like. (Probably recommended because I watch LGR now and then)
Excellent. The 6532 RIOT chip market is seeing a crapton of this going on as well. It's a PITA, and there's no community FPGA solution for to replace them, so we spend and cross fingers.
Someone should write code for a common chip tester to distinguish and test these FM chips, allowing any serious hobbyist or reseller to sort incoming chips into fakes, relabels, dead chips and good old stock.
They should definitely be remarked as recycled. The lifetime of the part is a function of how long it has been powered on. People should know upfront if the part has been used.
The numbers and letters in the ejector pin recesses are identification for where they belong in the mold. And could also identify the cavity it came from if there are problems with the molding process.
This is great! I come across these fake chips ALL the time. Once you learn how to spot the tell-tale signs, you can find re-badges and fakes in TH-cam videos everywhere. They always screw up the fonts and date codes!
Your video inspired me to do a little fake sleuthing as well. I decided to see if I could find out what my fake YM2612's really were. Just thought you might be interested in this! www.aidanlawrence.com/the-fake-ym2612-mystery-game/ Thanks again for the video!
First letter in the 4 letter lot code on later markings shows the chip manfacturer. G is Hitachi for example. But yeah, laser marking never appeared on anyone but NEC's chips at least in 90's and NEC didn't produce the FM chips for Yamaha (though they did other sort of chips). NEC always has a distinct font as does Fujitsu and some others. I hope that with this video and general awareness makes the sellers realize that original state is always better than a remark and other concealment methods.
I learned about this the hard way when I ordered a bunch of EPROMs off E-bay that were relabeled. I was supposed to get ST Micro CMOS 150ns access EPROMs and I ended up with National Semiconductor 200ns NMOS (older technology) EPROMs. I was able to tell because they did a poor job repainting some of the ceramic tops and the older National Semi logo was showing through on some of them. Another giveaway was many EPROMs have two unerasable bytes that identify the manufacturer (byte 1) AND the device code (byte 2) many programmers can read these bytes and decode them for you. If by some chance the programmer can't it will usually read back the two bytes as hex values which you can decode yourself by looking at manufacturer datasheets. It was originally used by EPROM programmers to auto identify the chip manufacturer and part number to make programming faster but now it is an excellent way to detect counterfeits :-). They worked fine (only one of them was a dud) and the access time difference wasn't a big deal for my application but they did draw extra current in the circuit due to the older NMOS technology they used.
But you could relabel it as a faster speed grade (lower access time) which might be more desirable. I didn't quite catch if that was the case with the EPROM in this video, but if they simply remarked it with the same info then yes, definitely stupid.
@@TomStorey96 A mil contractor has a qpl drawing and would normally reject a shipment of equivalent or "faster" chips. Even a mischievous prime chip vendor that has a blind process improvement change causes problems. An identical faster chip means evaluating where used and possible board redesigns to tame a faster leading edge etc.
@@anullhandle Sure, but thats why I said "might". To a hobbyist, a device with a faster access time or higher operating speed might be more desirable, but will come crashing down because the parts were rebadged as something they arent. A military contractor probably isnt going to be buying parts from AliExpress...
Great video! Ran into lots of blacktopped ICs when buying off eBay for my YM2149 sound card for the RC2014. Managed to find one eBay seller who could get me non-blacktopped ones after sending over pictures of what I was after.
Wish they'd just recycle them back in and not relabel. No problem with properly recycled electronics as long as everyone knows what they're getting. Saves on waste.
Concerning the ST M27C160, the reason for rebranding is not to have similar products and date codes. The main reason is because the chips were made before 2004, before the enactment of RoHS regulations. The purpose of the rebranding is essentially to pass what's now considered toxic trash as good lead-free chips. They make a fortune selling stuff that's supposed to be considered worthless. I bought a bunch of M27C160 myself and about 35% were defect no matter where I sourced them as the real ones are discontinued. They usually come from the millions of Mah-jong and other arcade machines popular in Asia and now obsolete.
Thanks for publishing this video and share your knowledge. I've been recently in the research of a MOS 6510 CPU and I have found out that at least two AliExpress / Ebay sellers do sell the same counterfeit ICs. Those are very easy to spot as the surface had been sanded down and the date code of the chips were 2010! :D . Whereas one seller did automatically the refund, the other tried to excuse on the fact that they were sold by large distributors and offered a 50% refund. Obviosuly I had opened a dispute and provided the evidences and got the refund. The most scary part was having to plug in that device on my C64 and prove that it did not work. I have never exposed the die of any chip, but I would have loved to know what was that crap really.
good video! i've been working on a guide on identifying fake YM2612s specifically, and it's all based on the package type (particularly ejector pin marks). i've found that getting non-remarked chips is rare at this point, unlike a few years ago, and some chinese ebay sellers are now starting to supply completely fake chips, but luckily they can be spotted easily by the package. there's a die inside, but no idea what it is, certainly not an OPN2...
At 1:07 on the right side page "This report states that the majority of counterfeit incidents were reported by US-based military bodies and electronic firms from the aerospace industry." That is a scary statement. That means counterfeit chips could be putting peoples lives at risk.
How much more expensive would a socket be? They even sometimes socket opamps in nicer audio equipment. Please do it if there's even a chance of the chip going bad or gaining on value.
Super interesting video. You've laid out really clearly something that I had sort of suspected but not been able to articulate clearly. Now I have a better understanding of some of the "special finds" I've come across on ebay.
I have no problem with them recycling the chips, I think it's a great idea and environmentally a nice thing. But they shouldn't be screwed with and should be priced as pulls. Nothing wrong with this if they don't remark.
This right here. This is how you make informative videos. I watched every second, and I learned so many things. Very fun to watch, and I'll be super careful in my projects as well, I believe my OPL2LPT has a remarked chip in it. Sounds perfect so I'll keep it, but I would have preferred to have the original markings. Also of note, the earliest 3812 I've encountered (on an original AdLib prototype) had the old style marking 72 06 77 B on it! Thanks for enlightening me as to what it meant; I originally wondered if the chip had been made in the 70s, which felt absurd! Un abonnement bien mérité. J'attends déjà vos prochaines vidéos!
Merci beaucoup! Wow. this made my day. My early Ad Lib (serial 19792) has dremelled 3812 so can't see the date either! twitter.com/plgDavid/status/1222515025518252038?s=20 Would love to see a picture of your card!
if the chip does what it needs to do, it's good enough (like clone). It's when you get a chip and it doesn't do what is expected little alone printed, then we have a problem.
If they're faking the same code on all of the reprints, why don't we just start going after chips made in week, say, 56 to make them fake that? That way, a chip with datecode xx56 is automatically known to be a resurfaced chip We could start with making an high value auction on eBay with 1000 * [4076] IC's from week 9556 to see if it works
Well, u can think of it as organ donation too. Computers, after all, don't have a conscience, so there's no donation form to be signed. The manufacturer company signing it for them is hilarious, its like a slave-master thing.
35 seconds in, that is a remarked chip. It was sanded, painted black and a number imprinted on it. I used distribute IC's and developed my QC department. Buy 2004 I had to fail 4 out of 5 orders of parts coming in.
Interesting video with lots of details. Thanks! Some time ago I received some fake relasered Atmel ATtiny44A chips that probably contained some totally different chip. I made a video about them which you can find on my channel. As you can see the filing of the top was quite crude and not even level, perhaps to get rid of some deeper some details.
I remember something similar from years ago. Sony had a silicon controlled switch, part number SG-613. Some bad batches were rejected, only to end up on the discount market. These were installed and promptly failed, occasionally taking out other components and making for a much more expensive repair.
Considering in the days of the 286 there were motherboard manufacturers making use of full on fake cache chips (just bits of plastic with pins)...I'm not surprised to see fake/refurbished chips still being easily obtainable.
Lol, this reminds me of the late sixties when battery powered radios and walkie-talkies were marketed by how many transistors they had. This was because transistors were relatively new and replacing tubes. Sears actually sold a walkie-talkie advertising that it had 12 transistors. Six of these transistors were not even connected to the circuit.
The first ICs I bought a few years ago were in a Radioshack kit I picked up surplus locally. The chips were 7400/4000 logic, but clearly counterfeit. The printing on them would wipe off very easily if my fingers were the least bit oily. Like if I touched my face and then inserted a chip, the printing would rub right off. Some didn't even have etching underneath the printing. I also got a different brand transistor kit on amazon with different jellybean transistors and when I opened up one of the 3055s, it was just a blob of epoxy over a die. Whereas the real chips had the die bare and readibly visible. Worst of all though, were some 74hc595 shift registers I bought on Ali Express. The Texas Instruments chips had a picture of the United States for the logo, rather than a picture of Texas as anyone with half a brain would know lol. I've gotta say though, even though it seems the majority of the chips I bought from AliExpress are fakes, I never would have suspected it unless I knew what to look for. In terms of function, they work just fine. I don't have time to characterize every chip I have, but the few I have characterized propogation delay on were within spec. But I realize there's a lot more to it than that. So maybe one day I'll go through a sleeve of those 595s and do a bit more thorough characterization.
I prefer checking the resistances between GND and VCC, if you have another good part you can compare resistances, After this checking with an oscilloscope the chip pins if they have the correct outputs, if you have another good part you can compare the signals, if you don't consult the chip datasheet.
@@plgDavid If you want the analog parts of the SID for real (no DSP or FPGA) it should be possible to build a SID on a PCB and connect it to the DIL socket. However, it would probably take some 15-20 74HC series ICs and at least a couple of quad op-amps to implement the functions. (20-ish surface mounted 14 pin ICs on a double sided PCB does not have to be big though.)
Thank you very much for the video! I was totally surpised when I tried to use DRAM chips which was bought on eBay by me and neither was working from the 40 chips. Maybe it would be a good idea to collect sellers on eBay who are selling fakes. Seller in my case who was selling fakes: "huayi-components". Avoid this eBay seller!
It's also pretty common on vintage transistors that are no longer made. I had a Sony receiver to repair with SK darlington transistors in it that were blown, and I ordered a decent looking pair from eBay. I got them and found that they were relabeled cheap transistors in the same package, as the die was about half the size of the originals.
Counterfeit market is a big problem in these days. I buy only in thrusted sellers like Digikey, Newark or Mouser. I buy in two chinese thrusted sellers also. Best Regards.
So I went back and checked on who I got my YM2612's from and sure enough it was from someone sourcing from "centruy sources" Luckily, only one of them was a fake. The other one in the pack I got was sanded down and relabeled. I'll have to test the sanded one and see where it stands.
It's a possibility that the letter and numbers on the underside are the injection mold IDs. The molds have a finite duty cycle and these kind of codes are used to identify molds that are at end of life in quality control stage. For example a batch might come through where the notches in the mold are starting to round which can be seen on physical inspection. The mold needing retooling can be identified by its number code.
Yes. Just wish I had the actual meaning of those. the only thing that we can figure out out of this, is oh this looks like this or that manufacturer's mold codes.
Brilliant video. thank you for the info and showing that the chips are recycled. I don't mind using recycled chips in my projects but I want to know they are recycled.
I am sure from that one that was really badly packed it was in a reused from instant dried noodle wrapper. If those where ESD sensitive I bet they would have fried. Thanks for doing this I really enjoyed this as a electronic engineer that likes to try and scope deals for parts like your doing for your projects.
I bought a 8087 (math co-processor) off eBay. It was "marked" as Intel, but according to a video I watched, that I forget which one it was, this was fake. The 8087 seems to work and pass all tests, but the date code printed on it is wrong. So, it seems to be the same as these sound chips.
Really cool video! You got a pretty good hit/miss rate on the chips :). Though for me a “fake” really means that it is actually not the correct chip on the inside. I am mostly happy they can be found at all and we can still build hardware with them, but of course I’m not doing research like you are. The remarking is very stupid, I bought a few YM2608’s a few years ago and they did look dodgy, clearly sanded. But when I finally had a board to try them on a few years later, surprisingly enough all four of them turned out to be the real chip and are working perfectly. I also had gotten four YM2610B’s which I heard a lot of stories about being rare and often fake, but those too were actually also all YM2610B’s! Tested and 6 channels confirmed. It’s not all sunshine though; a person I know bought a whole lot of YM2610B’s for a hardware project and he’s had much worse luck, the vast majority of them were YM2610’s remarked as YM2610B. I think seven out of ten were fake. It was a nice afternoon with two of us testing each and every one of them, but it did kind of suck really for that project. I posted a picture back then: twitter.com/Grauw/status/1127328867604156422 Luckily only 8 out of the 145 VGMs for OPNB actually use the YM2610B capabilities (despite vgmrips claiming there are 35 of them) so it’s not too huge of an issue in the end. The stupid thing is those YM2610’s are really perfectly fine chips on their own, so if they just didn’t remark them and up the price of the real YM2610B’s, everybody would be happy.
Yes I play a lot with the vagueness of what "Fake" means. In the end if it works ifs _fine_ but still no clue how long they will be fine because of their past treatment. Cheers!
@@plgDavid Yeah that part was very interesting, I did not know they treated them so roughly. Even if they’re not remarked, you still can’t tell that part.
@@plgDavid yeah, I agree. Fake is not the right concept here. You should use remarking. And how this treatment affects the future performance is a topic for your next video. Long term reliability test of recycled YM2612 vs unused ones??
@@BramBramsBerg 'Fake' is used throughout as a term that should be indeed replaced (hence the many double quotes). Fake is a term that captures the imagination, and was used in previous video on the subject on YT. Yes long term reliability studies is a great topic of research. Trying to find some hard facts/studies on this is proving to be a challenge. Just like ESD handling
Finally someone who can confirm my guesses with the currently very cheaply found AY-3-8910 chips. These are offered for €1 each everywhere, and because I can't get AY-3-8912 chips for less than €7 each, I was really interested in using the -8910. But the -8910 sounds very differently, I'm not sure if that's because it's a clone or if it's an entirely different chip that is remarked as AY-3-8910. Also some months ago I tested AY-3-8913's from several sources, and those ALL had non-working noise generators it seemed. Thanks for this video, though it of course does not solve the problem obtaining legit chips... Oh, if you want to know: I use dozens of AY-3-8912's a months for retrocomputer DIY kits and sound interfaces (google if you're interested).
I have a few AY-3-8910 clones in here even from relatively big brands like Winbond! This chip's patents (if any?) were not an issue in the mid-late 90s
@@plgDavid So you say they've been 'cloned' a lot back in the days already? I must be honest: I don't know much about their history, only use them a lot for new retrocomputer hardware. Thanks for your reply by the way!
@@plgDavid In case you or anyone else is interested: I posted about it on my Facebook timeline (have a lot of followers there), with an example photo of some of the AY-3-8910 chips I have in stock: facebook.com/BenVersteegTech/posts/2554979338095107
Wow, I had no idea there were so many YM chip fakes floating around. Makes sense in hindsight!
Thanks for the video, truly useful stuff 👍
Really glad you enjoyed it! If you ever have deep sound chip questions, do not hesitate to ping me here or on twitter. Cheers!
Somewhere I was watching a chip be x-ray'ed(?) to spot new chip dies, it was fascinating. Not sure if he was you plgDavid, but I watched your video in entirety, and certainly enjoyed it, thank you!
as I suspected, they are refurbished, no one have the technology to produce this legacy ICs
Of course it's only truly a fake if the underlying chip itself is a fake. Taring it and remarking is only somewhat of a fake, more of a deception if you will. Used, abused, tossed around, overheated, etc.
You don't get the internet, do you? He called you out. You were supposed to insult his mother.
haha, this reminds me when I worked for a very small military subcontractor who painted reliability bands on resistors when they couldn't get the hi-rel yellowed striped required by the contracts. When they were washed in solvent they disappeared. Right in front of the government inspector.
i see a green version for ESD handmade blackboxes DESTROYER versions or?
Well if you know your products will be used to protect people ,fi, medical, or in military and you are acting like this you really gambling with lives. Only the guillotine will be the place. Bare in mint you are in a plane with those fake parts! Don't tell me it was an order from upperhand. YOU also collaborate. I really can not understand how dumb people can be. The stuff mentioned here in this film is game stuff and as people only desire "original" they get it, no physical harm is done. Thats the difference.
Wow, what a terrible company.
Wow - that's 'go to jail' type stuff they were doing!
for stuff like that, cost should be no object. spare no expense. Penny pinching could literally cost lives.
What you are referring to as "ejector holes" are typically called "ejector pin marks" made by the mold ejector pins as they push the part out of the mold. The "code" markings on the bottom of the chips are usually for manufacturer reference and can be set to anything. Molds wear out, so by putting an identifier mark, they can identify and track mold life. It's also common for identifiers to be put on multi-cavity injection molds, so if there is a defect in one of the cavities they can locate it easily. Still excellent video.
That is the sort of comments I'm looking for! You are in the industry?
@@plgDavid Not really, as a kid many years ago, I spent a summer as a laborer in an injection molding factory. Horrible job, it's hot, wet (large parts need to be put in clamping fixtures and then in cold water baths), itchy (glass fiber filled plastics), toxic (burnt plastic from clogged nozzles makes your eye's water uncontrollably and mold release is an aerosol that for some reason appears to be neutrally buoyant, so it just floats in a cloud around you) and that's on top of the super monotony.
Laborer turn-over rate was 7-working days on average. Most would just walk off the job and the machines don't stop, so there would be a small mountain of plastic parts where they stood. I remember using a snow shovels to clean up messes. We were a small shop with mostly short duration part runs. Sharp's setup would be have been more highly automated I presume.
I knew fakes were a thing, but I'm really surprised by how common these re-marked chips are. Thanks for sharing!
Why don't they just sell them as reclaimed/re-cycled? we know they aren't being made anymore.. they are legacy chips.. just sell them with the damn scuffs... it's not a problem as long as it's electronically sound! (Pun intended)
Retro hobbyists aren't the main market. They sell to people like military contractors who need new, recently-dated chips which haven't been made in years.
But do you think military contractors let themselves be deceived that easily? I can't believe that especially when even hobbyists like plgDavid investigate and find out.
@@armorgeddon its about the specifications, they don't care if they're old, they just need them to say they're new so they meet 'military grade' and then they can sell them. that's my understanding anyway. it's an everyone benefits kinda deal...aside from hobbyists trying to buy the best chips they can i guess.
@@dextrodemon Makes no sense unless they hope to ask higher price for unused stock.
True military application typically require a legitimate supply chain for the lifetime of the device. Also what do they supposedly use it in the military for?
@@misium there are plenty of horror story like this one www.theregister.co.uk/2019/02/05/us_counterfeit_chips/
This is an excellent informative video, and one that I hope more subject matter experts start emulating. One of the things I really appreciate is how you show several examples so that, when it's not 100% obvious, the viewer can try to build a "confidence score" to help judge the actual condition.
I see the same thing in my hobby (vintage software collecting), where counterfeit items and even counterfeit *shrinkwrap* can artificially raise the sale value. Thank you for this very detailed comparison.
Re-shrinkwraps are a bane of Vintage Video game collecting yeah. I do collect for some platforms. great parallel.
@@plgDavid I will be creating a video on how to spot re-wraps; I'll make a note to mention it to you once the video is online.
@@JimLeonard Wow, that sounds awesome! I'm definitely going to subscribe to your channel and keep an eye out for that one.
@@JimLeonard Wow, that sounds awesome! I'm definitely going to subscribe to your channel and keep an eye out for that one.
Yep. Solder pulls. Heated with a blowtorch and dropped out. Re-stenciled. Pins bent back out at a splayed angle, and re-tinned. But for stuff such as 74C915, a rare hen's tooth,
these Chinese solder pulls are the only game in town. It's a good thing these ics were so over-designed, because these animals heat them way beyond any temperature
they were ever meant to endure. Then you get these off of Ebay and only maybe 80% test good. We are damn lucky to have access to them at all.
Wow, this video brings back memories... Back in the 1990's they had a ton of fake L2 dram chips. (Before it was integrated on the cpu) When I was building PC's I was testing them and discovered many fake chips. I cut some open. They were nothing but blocks of plastic with legs/pin sticking out. LOL. Seems nothing really changes.
The infamous PCChips M919 Socket3 486VIP motherboard...
This explains a lot! I have a few customers with cheep audio equipment and CRT rear screen projectors containing STK power chips. They seem to last a year or so then they end up back on my bench with the same chip failure. As far as the audio equipment these these LSICs usually barely or do not meet spec. I have not taken the time to design a test for the convergence chips. All replacements were dirt cheep on eBay or Amazon! I will ask the seller some pointed questions before purchasing them in the future. Thank You
What you are talking about is FAKE chips. The Yamaha chips in this video are not fake.
0:29 - Hey, that's my product! The chips I buy are from a seller in China and are definitely pulled. I take some comfort in the fact that they're not cleaned up very well (pins are bent, dirty, etc), so I hope the puller isn't particularly resourceful. I have gotten some chips that just don't work, but it's probably around 5%. I'd guess that a PSG chip like the 2149 just isn't as lucrative to fake. I have another batch of 20 chips on their way, and I'll take an extra look at them. Really enjoyed the video, particularly talking about the recycle operations. I knew it existed but it was really cool to hear more about it!
Yeah, I talked to Trash80 about it as well afaik? AYs/and 2149 were lucrative to clone in the past, there are quite a few AY-3-8910 clones out there, but the YM2149 needs that extra pin logic so its not a drop in replacement. though AY8930s uses the same logic. You might want to send AY8930 commands just in case :) Extra pulse widths and independent noise on each channel.
It's fun to think about how far some of these chips traveled. They were presumably manufactured in Japan, maybe then shipped to China for the assembly of the original product, then maybe sold to the US, then disposed of decades later in an electronics recycling collection, from where it was shipped back to China, where it was then salvaged and relabelled to then be shipped back to the west.
I wish the whole recycling process could be done locally in an honest manner, but I guess labour costs are too high in the west.
If they could talk..... man.....
Oh man yeah, it is crazy actually. I think labor costs are really the trouble indeed, and HSE guidelines and environmental stuff. In China they just do it while here we have all the extra rules and high labor which makes the recycling not profitable.
@Fratti
In the 80s and early 90s the original boards would have been assembled in Japan, the US, the UK, Korea, or maybe Taiwan for the more budget items.
China didnt do a lot of name brand electronics manufacturing back then.
At that time, China's light industry was not developed, so at that time printed With “JAPAN” chips are made of Japanese.
Or back to Japan. Maybe through th same seaport.
I'm a big fan of reusing the stuff, so personally, I'm very happy that chips are recycled but this should be super clearly disclosed so professionals like You are 100% aware of what You are buying. In the end, money only talks so this won't disappear and fakes will continue. :( Without this video, I wouldn't know. Thank You for this :)
Exactly! I would have no problem buying recycled chips as long as that's what was stated up front.
I agree. I'd gladly use recycled parts as long as I knew up front what I was getting. And the part worked perfectly.
Agreed and I've had to in order to repair old Tektronix scopes which have Tek custom devices in them.
I actually thought that these are repros
kinda cool to know they're the originals just sometimes crappily refurbished
In 2000? I was removing a friend Intel Pentium 100 chip from his computer and when we were detaching the heatsink ceramic paint came off the top of the chip showing another date code and the fact it was once a Pentium 75!
That is quite early!
That's not the same thing....Back then, uPCs got tested for spec at the rated frequency. Some of them met spec at an even higher frequency without problems. Some of those uPCs then got re-marked with the higher rating....
I actually heard back in the 90s' independent computer shops would sell overclocked computer as the real thing. Essentially selling computer with Pentium 100 overclocked to 120MHz as Pentium 120 gaming computer. The heatsink were glued on so impossible for regular people to tell
Most of my contributions came as a result of me buying 43 YM2608s/OPNAs from Ebay and AliExpress for development of my MSX OPNA cartridge. I bought three from a seller on ebay, all were sanded, and one was partially non-working. All were laser-etched with a datecode from 2012, and all, curiously, had the die revision marked as M.
The other 40 I bought came from a seller on AliExpress. They had good reviews, most of the buyers were Japanese, and the picture (sadly) was of a beaten-up but legit, non-remarked chip. I ordered 10 at first, all came in, ALL WERE LEGIT, AND ALL WORKED. I cleaned them out after, buying a second batch of 30 more, once again, all were untouched but obviously recycled, some were chipped, some were scratched, I ended up with one having a broken pin, but of the 40 that I bought, only 2 were damaged, and they had the same fault as the one non-working chip as the very first batch of 3 sanded chips. A total loss of 1/3 (33% failure) for the remarks, and 2/40 (5% failure). I got all sorts of date codes, one duplicate, die revisions L, O, and P, but NOT as the first three were marked to say, M or N.
On the subject of the old-style Yamaha datecode:
That was LOTS of fun to figure out. Being able to find that out and get that published on Siliconpr0n's Yamaha vendor page (other things still need to be corrected there), and in this video is really cool to see. Looking at pictures of chips and trying to spot patters like manufacturers, package types, date codes, what different letters and numbers mean is such a fun thing to do. The people who originally thought of these schemes and know what they 100% mean aren't around any more, either at the companies or, sadly, on the planet. Being able to dig up and solve small mysteries like this all in the name of preservation and documentation is one of the most fun and fulfilling things out there, in my opinion.
On the subject of the remarked EPROMS, I actually had 10, not 1. I bought a batch of 10 27C801s on ebay. Large EPROMs are useful and expensive, and while I may not need them now, who knows? I got them in and they were all marked exactly the same, and like you saw on yours, they all had brand new pins, slightly tarnished, factory straight. All were also blank after being read. EVERY SINGLE ONE could be cleaned, and not with acetone, but plain 91% IPA. The most amusing part is that THREE were remarked TWICE. After being cleaned, Three had the exact same marking, in the exact same place underneath, laser etched into the ceramic. That's something that I've never heard of or seen, but is apparently possible and very scary:
There is a possibility that any chip you buy might not be re-marked just once, but twice.
HUGE ANIME BREASTS
@@KokoroKatsura Love em, what about em?
I like the idea of re-use but be clear about it. Indeed, one doesn't know how much the chip was hammered in prior use, high-temperature and poor anti-static conditions when extracting.
Interesting video. I used to work in the semiconductor industry. I have seen a case where a customer sent in a complaint about our chip and it turned out to be a counterfeit chip. That was an interesting situation, at that time.
I have one detailed comment: at about 15 minutes in the video you explain the date code. And mention that the chip was "fabricated" on that date/week. The entire manufacturing process, starting from a silicon wafer, takes multiple weeks. The date code as printed on the package is the assembly date code. The chip inside could have been manufactured quite long before that and have been stored before being assembled.
Nothing beats comments from actual specialists! Would love to know in your experience are the wafers usually done in the same plant as the molding/assembly or its unrelated? The wafers need to stay in dust/hair/temp controlled env and shipping them out makes little sense right?
@@plgDavid Oh no, the entire manufacturing process is most certainly not done in the same factory, on the same physical location. I guess (but didn't check) that semiconductor manufacturers have promotional videos on their website where they explain the entire manufacturing process. Those videos explain the whole process better than a TH-cam comment box allows (Edit: I found an example video here: th-cam.com/video/M-wNC3Z3ZX4/w-d-xo.html ). Often the entire process can be split into four main parts: (1) get the electronic circuit in the wafers by using e.g. chemical etching and electronic implanting technologies, (2) test each circuit on the wafer and separate into individual dies, (3) put the dies which passed testing into a package (e.g. DIL, SO, BGA, ...), (4) test the assembled product. For logistical purposes do semiconductor manufacturers hold a stock of wafers, either after step (1) or (2). And they hold a stock of final products. Step (1) needs to be done in a clean room to avoid contaminations and to get the manufacturing yield up. The more advanced wafer fabs tend to be in Western countries, whereas Asian countries (e.g. China) have wafer factories which are a few generations behind. Assembly and testing of chips is most often been done in Asian countries (e.g. China, Thailand, Malaysia, etc.).
@@windmill1965 Thank you very much for all those details!
@@plgDavid AMD CPUs often say (or used to say) things like "MADE IN USA DIFFUSED IN SINGAPORE"
@@my.own.devices Exactly; CPUs are famous for going around the world several times, every step is performed by the lowest-paid workforce capable of the required quality standards. You'd have the optical transfer to wafer done in Germany (needs precision) and the pins would be soldered on in some hellhole in Malaysia. I was also told by a professor that the silicon crystal seeds are imported expensive, then grown into ingots (?) in medium-standard countries.
27:14 you know you're getting cheap stuff when the packaging is that plastic bag used for Korean ramyun noodles. This looks like nongshim shin ramyun packaging.
Wow, you put an ENORMOUS amount of work into making this video. I can't imagine the thousands of hours of experience, the inordinate number of orders, and just general expertise in a very specialized field that went into producing something like this. You sir, are a scholar and a gentlemen for sharing such knowledge with the world. I hope all the expertise and hours help in building your reputation and business. THANK YOU!
This made my day, thank you so much! And great name too. I'm 46 but I still skate :)
@@plgDavid I meant it! The internet with it's open-source/hacker/builder/creator/etc community never ceases to amaze me. People such as yourself who offer up their time, knowledge, experience, and expertise for essentially nothing but the satisfaction of sharing it with others are what makes these communities such awesome places to be apart of. No one shares this type of info/content for the pennies (if you're lucky) that TH-cam might give you if they decide that talking about this isn't a "sensitive topic". I'm glad you saw my comment as you definitely deserve more gratitude than I'm sure you receive! I'm still in my early 30's and though I still skate a bit, all those years of throwing myself down flights of stairs and handrails definitely gave me "old man knees" much too early in life 😁 I do much more reversing (I'm a reverse code engineer) and programming these days than I do skateboarding, but I will certainly be skating for as long as my poor, abused body will allow me to! Love your content man. Keep up the great work!
Very familiar. This is very well known to be done with certain transistors. All if a sudden we had whole batches failing. It really baffled us because they were just measuring just fine. It became really clear when we decapped a couple. The little metal slug (for transferring the heat) was just half the size of the original. Therefore the transistors were just basically overheating. This was even from a known seller/distributor.
From a molding perspective the letters and numbers on the ejector are usually to help identify the cavity number after the molding process it also helps when rebuilding the mold. You want too make sure the same ejector pin goes in the same location, it's also a standard from particular mold builders or molding shops.
Part of me is very sad to think of all those old (probably working) Sound Blasters being scrapped - I'd gladly buy the whole card!
Vink Same
I remember running into fake chips back in the late 90's when I work fro a PCB contract manufacture. They were called "shaved ICs" then. One fake Ic was interesting in our x-ray machine you could see the internals were different. (The ic was an amp and was no longer mad, the board had to get a redesign)
Interesting thing even with a legit similar spec IC was would it not run right when being used, diagnostic testing said good. but when system started running it would stop
I don’t know who you are, or how I was recommended this video, I’ve no interest in Audio Electronics, but that was the most interesting 40mins of random TH-cam recommendations in a long time 😃
That made my day, thanks!
This video was captivating. Thanks for all the effort.
It’s certainly comforting that the guy that was making some good bucks at home (granted with some unhealthy practices) is now the employee of multi billion corporation that as a middle man will rip the little guys off as well as hobbyists buying this stuff! Thank you for the video!
I am researching this for two weeks now. Thanks for uploading and clarifying
our factory in China returns on average 20-30% of the parts because they fail pre-production quality control. We have had cases of something getting through with the result being the product when used in low power situations "fails". In that case it was a fake Motorola 4011 logic chip, everything seems to work but I found when the VCC rail had a low voltage (but still within spec) it would cause a .5 volt logic output error. On closer inspection the 4011 chip looked exactly the same as the real deal. When I put it under a microscope I noticed the Motorola logo was poorly done and nothing like the original.
Damn! I didn't think they would fake even such mundane ICs as CMOS or TTL 'glue logic'.
Eugh... I know that where there's demand, supply will follow, but just... don't sell crap like this to people who don't know any better. Sell 'em as recycled or "refurbished" cause I for one would rather gamble on recycled chip than have none at all. Just don't put duct tape on a Rolex and tell me it's brand new =/
your comment will inspire exactly 0 of them to change their behaviour. In fact this video has probably made a few people realize they could make money selling counterfeit chips.
Counterfeit IC and Yahama FM chips are two item of interest of mine, so I was waiting for this video; it doesn't disappoint. Thanks David for taking your time to share your knowledge and expertise on this matter !
Thanks! Really had to kick myself for finishing it.
@@plgDavid Procrastination is a beach (with a T)
I don't know why youtube recommended me this, or who you are, but when I see a man, who is so deep into topic, that he have a book "Counterfeit chips" I put a like. (Probably recommended because I watch LGR now and then)
Excellent. The 6532 RIOT chip market is seeing a crapton of this going on as well. It's a PITA, and there's no community FPGA solution for to replace them, so we spend and cross fingers.
I'm only 5 minutes in to this video, but it's fascinating so far. Thanks!
Someone should write code for a common chip tester to distinguish and test these FM chips, allowing any serious hobbyist or reseller to sort incoming chips into fakes, relabels, dead chips and good old stock.
They should definitely be remarked as recycled. The lifetime of the part is a function of how long it has been powered on. People should know upfront if the part has been used.
The numbers and letters in the ejector pin recesses are identification for where they belong in the mold. And could also identify the cavity it came from if there are problems with the molding process.
Thats actually something new! (the where they belong in the mold part). Thanks!
This is great! I come across these fake chips ALL the time. Once you learn how to spot the tell-tale signs, you can find re-badges and fakes in TH-cam videos everywhere. They always screw up the fonts and date codes!
Your video inspired me to do a little fake sleuthing as well. I decided to see if I could find out what my fake YM2612's really were. Just thought you might be interested in this!
www.aidanlawrence.com/the-fake-ym2612-mystery-game/
Thanks again for the video!
First letter in the 4 letter lot code on later markings shows the chip manfacturer. G is Hitachi for example. But yeah, laser marking never appeared on anyone but NEC's chips at least in 90's and NEC didn't produce the FM chips for Yamaha (though they did other sort of chips). NEC always has a distinct font as does Fujitsu and some others. I hope that with this video and general awareness makes the sellers realize that original state is always better than a remark and other concealment methods.
I learned about this the hard way when I ordered a bunch of EPROMs off E-bay that were relabeled.
I was supposed to get ST Micro CMOS 150ns access EPROMs and I ended up with National Semiconductor 200ns NMOS (older technology) EPROMs. I was able to tell because they did a poor job repainting some of the ceramic tops and the older National Semi logo was showing through on some of them.
Another giveaway was many EPROMs have two unerasable bytes that identify the manufacturer (byte 1) AND the device code (byte 2) many programmers can read these bytes and decode them for you. If by some chance the programmer can't it will usually read back the two bytes as hex values which you can decode yourself by looking at manufacturer datasheets.
It was originally used by EPROM programmers to auto identify the chip manufacturer and part number to make programming faster but now it is an excellent way to detect counterfeits :-).
They worked fine (only one of them was a dud) and the access time difference wasn't a big deal for my application but they did draw extra current in the circuit due to the older NMOS technology they used.
especially stupid to put blacktop on top of a ceramic DIL package as in the case of the EPROM - since these are never black from the factory...
But you could relabel it as a faster speed grade (lower access time) which might be more desirable. I didn't quite catch if that was the case with the EPROM in this video, but if they simply remarked it with the same info then yes, definitely stupid.
@@TomStorey96 also stupid because everyone should be able to spot the modification since ceramic packages never have a black plastic coating...
@@TomStorey96 A mil contractor has a qpl drawing and would normally reject a shipment of equivalent or "faster" chips. Even a mischievous prime chip vendor that has a blind process improvement change causes problems. An identical faster chip means evaluating where used and possible board redesigns to tame a faster leading edge etc.
@@anullhandle Sure, but thats why I said "might". To a hobbyist, a device with a faster access time or higher operating speed might be more desirable, but will come crashing down because the parts were rebadged as something they arent. A military contractor probably isnt going to be buying parts from AliExpress...
@Zero Cool stupid in a way that any half-way experienced person will immediately spot the fake!
Great video! Ran into lots of blacktopped ICs when buying off eBay for my YM2149 sound card for the RC2014. Managed to find one eBay seller who could get me non-blacktopped ones after sending over pictures of what I was after.
This would be talk worthy topic at Chaos Communication Congress or similar event!
Wish they'd just recycle them back in and not relabel. No problem with properly recycled electronics as long as everyone knows what they're getting. Saves on waste.
These counterfeit chips should be marked with "D" for 10 years of Doom and "H" for Heretic.
Your comment should be marked D for dumbass.
There aren't any counterfeit chips shown in this video. They are all genuine, original chips.
Some may call them reused, I call them "pre-tested"...
yes, I'm showing myself out :D
@@johncoops6897 he never said the ones in this video just counterfeit ones should be called that
@@ducksonplays4190 - you seemed to have missed the first word in the original comment. He said " *THESE* counterfeit chips..."
He was talking about chips in general
This is an amazing video and a great service to the retro-computing community.
Concerning the ST M27C160, the reason for rebranding is not to have similar products and date codes. The main reason is because the chips were made before 2004, before the enactment of RoHS regulations. The purpose of the rebranding is essentially to pass what's now considered toxic trash as good lead-free chips. They make a fortune selling stuff that's supposed to be considered worthless. I bought a bunch of M27C160 myself and about 35% were defect no matter where I sourced them as the real ones are discontinued. They usually come from the millions of Mah-jong and other arcade machines popular in Asia and now obsolete.
@@francoisleveille409 That's a piece of info I haven't heard before, makes a lot of sense!
This was a really good video
Thanks for publishing this video and share your knowledge. I've been recently in the research of a MOS 6510 CPU and I have found out that at least two AliExpress / Ebay sellers do sell the same counterfeit ICs. Those are very easy to spot as the surface had been sanded down and the date code of the chips were 2010! :D . Whereas one seller did automatically the refund, the other tried to excuse on the fact that they were sold by large distributors and offered a 50% refund. Obviosuly I had opened a dispute and provided the evidences and got the refund. The most scary part was having to plug in that device on my C64 and prove that it did not work. I have never exposed the die of any chip, but I would have loved to know what was that crap really.
Destroying a machine by putting a bad component in is much worse indeed.
good video! i've been working on a guide on identifying fake YM2612s specifically, and it's all based on the package type (particularly ejector pin marks). i've found that getting non-remarked chips is rare at this point, unlike a few years ago, and some chinese ebay sellers are now starting to supply completely fake chips, but luckily they can be spotted easily by the package. there's a die inside, but no idea what it is, certainly not an OPN2...
Awesome! I would be curious in looking at this guide when it is out.
At 1:07 on the right side page "This report states that the majority of counterfeit incidents were reported by US-based military bodies and electronic firms from the aerospace industry." That is a scary statement. That means counterfeit chips could be putting peoples lives at risk.
Yes indeed! There is little risk with synthesizer projects, but medical/military/transportation ... its really scary!
Quickly scanned my stash of Chinese-sourced OPL-2s, all different, thankfully. I test them before soldering but you gave some good tips.
How much more expensive would a socket be? They even sometimes socket opamps in nicer audio equipment. Please do it if there's even a chance of the chip going bad or gaining on value.
Super interesting video. You've laid out really clearly something that I had sort of suspected but not been able to articulate clearly. Now I have a better understanding of some of the "special finds" I've come across on ebay.
I have no problem with them recycling the chips, I think it's a great idea and environmentally a nice thing. But they shouldn't be screwed with and should be priced as pulls. Nothing wrong with this if they don't remark.
I agree.
Btw you have the same joy as i do when i handle retro ics. They are a thing of beauty indeed.
33:21 it literally hurts me seeing those chips get unwrapped from packaging foil, lots of ESD going into those poor chips
ESD is the least of their worries with the crude recycling they suffered, but I agree
But are they cmos? If not ESD wont break them...
@@plgDavid so what is about whether these are operational?
@@MacIn173 false representation and the potential latent defects of being treated like hell.
YM2612s aren't CMOS from what I remember
I think that YM3438s had a CMOS variant
Report those vendors to Ebay. If people report them, they may be banned.
They will pop up again using a different seller name. When you can by new.
This right here. This is how you make informative videos. I watched every second, and I learned so many things.
Very fun to watch, and I'll be super careful in my projects as well, I believe my OPL2LPT has a remarked chip in it. Sounds perfect so I'll keep it, but I would have preferred to have the original markings. Also of note, the earliest 3812 I've encountered (on an original AdLib prototype) had the old style marking 72 06 77 B on it!
Thanks for enlightening me as to what it meant; I originally wondered if the chip had been made in the 70s, which felt absurd!
Un abonnement bien mérité. J'attends déjà vos prochaines vidéos!
Merci beaucoup! Wow. this made my day. My early Ad Lib (serial 19792) has dremelled 3812 so can't see the date either! twitter.com/plgDavid/status/1222515025518252038?s=20 Would love to see a picture of your card!
This is brilliant information. I really appreciate you taking so much time to detail everything for us viewers. I really learned a lot!
Great video! I've had to buy from Hiific and a few others before... it's hit or miss, but most of the time they are the only ones selling what I need.
Great video and great research. Lots of time and good work required. I know because I've done things like this. Thank you for sharing!
if the chip does what it needs to do, it's good enough (like clone). It's when you get a chip and it doesn't do what is expected little alone printed, then we have a problem.
If they're faking the same code on all of the reprints, why don't we just start going after chips made in week, say, 56 to make them fake that?
That way, a chip with datecode xx56 is automatically known to be a resurfaced chip
We could start with making an high value auction on eBay with 1000 * [4076] IC's from week 9556 to see if it works
I was not expecting the dive into chinese remanafacturing culture. Eye opening seeing how those guys are making a living.
Makes me wonder how many Mega Drive's and dead Sharp computers are cannibalised for their YM2612 chips.
This makes me cry a bit. Not gonna lie. :(
My SB AWE32 with 2 mb extra ram is in a locked drawer and will follow me to the grave ;)
Well, u can think of it as organ donation too. Computers, after all, don't have a conscience, so there's no donation form to be signed. The manufacturer company signing it for them is hilarious, its like a slave-master thing.
This certainly has my admiration. That's high level research! Namasté
35 seconds in, that is a remarked chip. It was sanded, painted black and a number imprinted on it. I used distribute IC's and developed my QC department. Buy 2004 I had to fail 4 out of 5 orders of parts coming in.
At 36:20 m"...the best thing would be to ship an extra chip..." Yup! I always order 50% more parts than I need. Saved many a project that way.
Thanks for putting the time in
This video is fantastic! I didn't know of this subject or any of the connected details.
Interesting video with lots of details. Thanks! Some time ago I received some fake relasered Atmel ATtiny44A chips that probably contained some totally different chip. I made a video about them which you can find on my channel. As you can see the filing of the top was quite crude and not even level, perhaps to get rid of some deeper some details.
Really interesting video and it looks like you have bunch more I'll like.
Thank you for this video, David. Fantastic information.
The sheer amount of counterfeit C64 SID chips alone out there is enough to make your head spin
Very interesting. I'll go and cjeck my IC's. I have bought alot of them cheap on ebay, but I have mixed in a few that I know are legit new ones.
31:30-31:45 "Look Mom, I'm fake" - yeah, that had me laughing, but yeah, pretty informative video
I remember something similar from years ago. Sony had a silicon controlled switch, part number SG-613. Some bad batches were rejected, only to end up on the discount market. These were installed and promptly failed, occasionally taking out other components and making for a much more expensive repair.
Considering in the days of the 286 there were motherboard manufacturers making use of full on fake cache chips (just bits of plastic with pins)...I'm not surprised to see fake/refurbished chips still being easily obtainable.
Lol, this reminds me of the late sixties when battery powered radios and walkie-talkies were marketed by how many transistors they had. This was because transistors were relatively new and replacing tubes. Sears actually sold a walkie-talkie advertising that it had 12 transistors. Six of these transistors were not even connected to the circuit.
The first ICs I bought a few years ago were in a Radioshack kit I picked up surplus locally. The chips were 7400/4000 logic, but clearly counterfeit. The printing on them would wipe off very easily if my fingers were the least bit oily. Like if I touched my face and then inserted a chip, the printing would rub right off. Some didn't even have etching underneath the printing. I also got a different brand transistor kit on amazon with different jellybean transistors and when I opened up one of the 3055s, it was just a blob of epoxy over a die. Whereas the real chips had the die bare and readibly visible. Worst of all though, were some 74hc595 shift registers I bought on Ali Express. The Texas Instruments chips had a picture of the United States for the logo, rather than a picture of Texas as anyone with half a brain would know lol.
I've gotta say though, even though it seems the majority of the chips I bought from AliExpress are fakes, I never would have suspected it unless I knew what to look for. In terms of function, they work just fine. I don't have time to characterize every chip I have, but the few I have characterized propogation delay on were within spec. But I realize there's a lot more to it than that. So maybe one day I'll go through a sleeve of those 595s and do a bit more thorough characterization.
I prefer checking the resistances between GND and VCC, if you have another good part you can compare resistances,
After this checking with an oscilloscope the chip pins if they have the correct outputs, if you have another good part you can compare the signals, if you don't consult the chip datasheet.
Yeah I do this and also using diode check
LGR blurb got me here.. 👍
I wish someone would clone SID chips properly, the prices of them are crazy
オーエンピーター Yes a real HMOS clone of the SID would be amazing
What price do they go for?
Locally they run for $70+ on sites like ebay.
it's the sound chip from the Commodore 64: en.wikipedia.org/wiki/MOS_Technology_6581
@@plgDavid If you want the analog parts of the SID for real (no DSP or FPGA) it should be possible to build a SID on a PCB and connect it to the DIL socket. However, it would probably take some 15-20 74HC series ICs and at least a couple of quad op-amps to implement the functions. (20-ish surface mounted 14 pin ICs on a double sided PCB does not have to be big though.)
Such a great IC-docu, thanks for doing this plgDavid!!
Thank you very much for the video! I was totally surpised when I tried to use DRAM chips which was bought on eBay by me and neither was working from the 40 chips. Maybe it would be a good idea to collect sellers on eBay who are selling fakes. Seller in my case who was selling fakes: "huayi-components". Avoid this eBay seller!
Excellent video thank you for sharing!
Wow, thanks for wonderful explanation 👌🏻👌🏻.
It's also pretty common on vintage transistors that are no longer made. I had a Sony receiver to repair with SK darlington transistors in it that were blown, and I ordered a decent looking pair from eBay. I got them and found that they were relabeled cheap transistors in the same package, as the die was about half the size of the originals.
Counterfeit market is a big problem in these days.
I buy only in thrusted sellers like Digikey, Newark or Mouser.
I buy in two chinese thrusted sellers also.
Best Regards.
So I went back and checked on who I got my YM2612's from and sure enough it was from someone sourcing from "centruy sources"
Luckily, only one of them was a fake. The other one in the pack I got was sanded down and relabeled. I'll have to test the sanded one and see where it stands.
W27C512 from ebay - "brandnew" and all are counterfeit...
Yep.
Awesome post. Such a murky area...trying to find the real deal...even official suppliers come to grief occasionally.
I am still looking for the SAD4096 made by Reticon, all I got so far where broken of fakes for eBay...
It's a possibility that the letter and numbers on the underside are the injection mold IDs. The molds have a finite duty cycle and these kind of codes are used to identify molds that are at end of life in quality control stage. For example a batch might come through where the notches in the mold are starting to round which can be seen on physical inspection. The mold needing retooling can be identified by its number code.
Yes. Just wish I had the actual meaning of those. the only thing that we can figure out out of this, is oh this looks like this or that manufacturer's mold codes.
here from lgr, incredibly fascinating yet eye opening video. subscribed.
This is really fascinating
Brilliant video. thank you for the info and showing that the chips are recycled. I don't mind using recycled chips in my projects but I want to know they are recycled.
You're doing good work my dude, keep it up! Someone has to do it!
I am sure from that one that was really badly packed it was in a reused from instant dried noodle wrapper. If those where ESD sensitive I bet they would have fried. Thanks for doing this I really enjoyed this as a electronic engineer that likes to try and scope deals for parts like your doing for your projects.
I've learnt some new things from this - and I thought I already knew a LOT of things on this subject!
Keep up the fantastic work!
I bought a 8087 (math co-processor) off eBay. It was "marked" as Intel, but according to a video I watched, that I forget which one it was, this was fake. The 8087 seems to work and pass all tests, but the date code printed on it is wrong. So, it seems to be the same as these sound chips.
Really cool video! You got a pretty good hit/miss rate on the chips :). Though for me a “fake” really means that it is actually not the correct chip on the inside. I am mostly happy they can be found at all and we can still build hardware with them, but of course I’m not doing research like you are.
The remarking is very stupid, I bought a few YM2608’s a few years ago and they did look dodgy, clearly sanded. But when I finally had a board to try them on a few years later, surprisingly enough all four of them turned out to be the real chip and are working perfectly. I also had gotten four YM2610B’s which I heard a lot of stories about being rare and often fake, but those too were actually also all YM2610B’s! Tested and 6 channels confirmed.
It’s not all sunshine though; a person I know bought a whole lot of YM2610B’s for a hardware project and he’s had much worse luck, the vast majority of them were YM2610’s remarked as YM2610B. I think seven out of ten were fake. It was a nice afternoon with two of us testing each and every one of them, but it did kind of suck really for that project. I posted a picture back then: twitter.com/Grauw/status/1127328867604156422
Luckily only 8 out of the 145 VGMs for OPNB actually use the YM2610B capabilities (despite vgmrips claiming there are 35 of them) so it’s not too huge of an issue in the end. The stupid thing is those YM2610’s are really perfectly fine chips on their own, so if they just didn’t remark them and up the price of the real YM2610B’s, everybody would be happy.
Yes I play a lot with the vagueness of what "Fake" means. In the end if it works ifs _fine_ but still no clue how long they will be fine because of their past treatment. Cheers!
@@plgDavid Yeah that part was very interesting, I did not know they treated them so roughly. Even if they’re not remarked, you still can’t tell that part.
@@plgDavid yeah, I agree. Fake is not the right concept here. You should use remarking. And how this treatment affects the future performance is a topic for your next video. Long term reliability test of recycled YM2612 vs unused ones??
@@BramBramsBerg 'Fake' is used throughout as a term that should be indeed replaced (hence the many double quotes). Fake is a term that captures the imagination, and was used in previous video on the subject on YT. Yes long term reliability studies is a great topic of research. Trying to find some hard facts/studies on this is proving to be a challenge. Just like ESD handling
Finally someone who can confirm my guesses with the currently very cheaply found AY-3-8910 chips.
These are offered for €1 each everywhere, and because I can't get AY-3-8912 chips for less than €7 each, I was really interested in using the -8910.
But the -8910 sounds very differently, I'm not sure if that's because it's a clone or if it's an entirely different chip that is remarked as AY-3-8910.
Also some months ago I tested AY-3-8913's from several sources, and those ALL had non-working noise generators it seemed.
Thanks for this video, though it of course does not solve the problem obtaining legit chips...
Oh, if you want to know: I use dozens of AY-3-8912's a months for retrocomputer DIY kits and sound interfaces (google if you're interested).
I have a few AY-3-8910 clones in here even from relatively big brands like Winbond! This chip's patents (if any?) were not an issue in the mid-late 90s
@@plgDavid So you say they've been 'cloned' a lot back in the days already?
I must be honest: I don't know much about their history, only use them a lot for new retrocomputer hardware.
Thanks for your reply by the way!
@@ByteDelight yes. its a very simple design if you look at it. Not even sure Yamaha's own YM2149 paid license fees for it.
@@plgDavid In case you or anyone else is interested: I posted about it on my Facebook timeline (have a lot of followers there), with an example photo of some of the AY-3-8910 chips I have in stock: facebook.com/BenVersteegTech/posts/2554979338095107