5 ms·

Inside a counterfeit analog multiplier

24 points by parsecs 6y ago

ggm 6y ago
If we had counterfeit ICs in the 1970s, how much earlier can we go back, and have widespread counterfeit electronic parts in the supply chain? Since the number of devices from the 1970s which are still extant is small, finding even one, is analogous to finding a fossil: its indicative of a larger population, but without more, you can't tell at what relative scale (bearing in mind of course that fossil electronics, like fossils in real life are subject to bias effects such as what lasts and what is kept)
- kens 6y ago
  My guess is that the chip was manufactured in the 1970s or 1980s but was re-labeled recently.
  - ggm 6y ago
    Where was the chip, these last 30 years?
- jecel 6y ago
  From the end of the post: "Now, counterfeiters re-label RC4200 chips and sell them as the more-expensive AD633 multiplier." The counterfeiters are doing this now, not in the 1970s. They just happen to be using a chip from 1978 since it is still available and cheap to impersonate a more recent chip that is more expensive. My first encounter with counterfeit ICs was in the late 1980s. Though PCs had been designed to use memory with parity, nearly everybody turned off parity checking in the BIOS. That made it popular to sell DRAM SIMMs/DIMMs with 8 good memory chips and one bad one. Sometimes the ninth chip wasn't actually bad but just a cheaper one that was far slower than the other eight.
  - ggm 6y ago
    To get a 1978 ic, they had to have a board to prise it off, or unsold stock. I wonder if there is something in a chip leg, package or substrate which changes in desoldering, such that you can tell how many times it was repurposed.
  - squarefoot 6y ago
    My experience was a lot worse. Dated back in the mid 90s I believe. In a shop I was working with, came a lot of incredibly cheap PC mainboards where the CPU (386SX, IIRC) had been relabeled to 386DX, the static RAM cache chips on the mainboard were just no-die pieces of plastic with pins, absolutely identical to real chips but 100% empty, and of course soldered, then finally the BIOS EPROM was hacked to show a cache was present and other bogus values, but the chip was soldered so one couldn't easily reprogram it with a legit BIOS that would have revealed the scam. We later realized that the manufacturer was smart enough to use on the boards box and manuals the term "virtual cache", so noone could complain. Those boards indeed worked, although of course they were slow as dead sloths.
atoav 6y ago
Amazing I just read counterfeit analog multipler and thought "this must be about the AD633“. I received counterfeits of that chip before and they were so bad they didn't work, mostly just getting hot. The fact that I got my money back speaks volumes.
- kens 6y ago
  Interesting that you've encountered counterfeit AD633s. I didn't realize that counterfeit AD633s were a big thing; I just wrote about this one case. I'd be interested in hearing more about your counterfeits, like how much you paid and where you got them. (I'm guessing on eBay from China.) I wonder if your counterfeits also had the RC4200 multiplier inside. Not working and getting hot is kind of what I'd expect since the AD633 takes a voltage as input and the RC4200 takes a small current. It's kind of strange that they'd substitute a multiplier chip, but one that wouldn't work. It's like the counterfeiters were trying to be clever, but didn't read the datasheet closely.
  - atoav 6y ago
    You are right on the money, it was eBay from China maybe 4 years ago, roughly 2,50 € per piece in a pack of 5. At that time they costed around 7 € a piece at mouser although I don't really would swear on any of these numbers being 100% accurate given the time that passed. I had genuine parts to compare so it was quite clear to me that something was off right away. I complained and described the measurements and got my money back without any further comment from their side.
effnorwood 6y ago
Only divides

24 points by parsecs 6y ago

ggm 6y ago

If we had counterfeit ICs in the 1970s, how much earlier can we go back, and have widespread counterfeit electronic parts in the supply chain? Since the number of devices from the 1970s which are still extant is small, finding even one, is analogous to finding a fossil: its indicative of a larger population, but without more, you can't tell at what relative scale (bearing in mind of course that fossil electronics, like fossils in real life are subject to bias effects such as what lasts and what is kept)

kens 6y ago
My guess is that the chip was manufactured in the 1970s or 1980s but was re-labeled recently.
- ggm 6y ago
  Where was the chip, these last 30 years?
jecel 6y ago
From the end of the post: "Now, counterfeiters re-label RC4200 chips and sell them as the more-expensive AD633 multiplier." The counterfeiters are doing this now, not in the 1970s. They just happen to be using a chip from 1978 since it is still available and cheap to impersonate a more recent chip that is more expensive. My first encounter with counterfeit ICs was in the late 1980s. Though PCs had been designed to use memory with parity, nearly everybody turned off parity checking in the BIOS. That made it popular to sell DRAM SIMMs/DIMMs with 8 good memory chips and one bad one. Sometimes the ninth chip wasn't actually bad but just a cheaper one that was far slower than the other eight.
- ggm 6y ago
  To get a 1978 ic, they had to have a board to prise it off, or unsold stock. I wonder if there is something in a chip leg, package or substrate which changes in desoldering, such that you can tell how many times it was repurposed.
- squarefoot 6y ago
  My experience was a lot worse. Dated back in the mid 90s I believe. In a shop I was working with, came a lot of incredibly cheap PC mainboards where the CPU (386SX, IIRC) had been relabeled to 386DX, the static RAM cache chips on the mainboard were just no-die pieces of plastic with pins, absolutely identical to real chips but 100% empty, and of course soldered, then finally the BIOS EPROM was hacked to show a cache was present and other bogus values, but the chip was soldered so one couldn't easily reprogram it with a legit BIOS that would have revealed the scam. We later realized that the manufacturer was smart enough to use on the boards box and manuals the term "virtual cache", so noone could complain. Those boards indeed worked, although of course they were slow as dead sloths.

atoav 6y ago

Amazing I just read counterfeit analog multipler and thought "this must be about the AD633“. I received counterfeits of that chip before and they were so bad they didn't work, mostly just getting hot. The fact that I got my money back speaks volumes.

kens 6y ago
Interesting that you've encountered counterfeit AD633s. I didn't realize that counterfeit AD633s were a big thing; I just wrote about this one case. I'd be interested in hearing more about your counterfeits, like how much you paid and where you got them. (I'm guessing on eBay from China.) I wonder if your counterfeits also had the RC4200 multiplier inside. Not working and getting hot is kind of what I'd expect since the AD633 takes a voltage as input and the RC4200 takes a small current. It's kind of strange that they'd substitute a multiplier chip, but one that wouldn't work. It's like the counterfeiters were trying to be clever, but didn't read the datasheet closely.
- atoav 6y ago
  You are right on the money, it was eBay from China maybe 4 years ago, roughly 2,50 € per piece in a pack of 5. At that time they costed around 7 € a piece at mouser although I don't really would swear on any of these numbers being 100% accurate given the time that passed. I had genuine parts to compare so it was quite clear to me that something was off right away. I complained and described the measurements and got my money back without any further comment from their side.

effnorwood 6y ago

Only divides