Chips bought on eBay ... likely to be faulty?

[nickgammon]

I bought a batch of 100 x TL431 voltage reference chips on eBay recently. It was a good price (around $6 for the lot). The are supposed to be able to generate 2.5 V reference (and with suitable resistors use that to compare an incoming voltage to it).

But when I went to test one, I got weird results (like, wiring it the way that was suggested in the datasheet, it generated 1.34 volts). I thought maybe I misread the info, and tried various adjustments (like different resistor values for the current-limit). But always about 1.34 volts.

Then I tried a second one from the same batch. This time I got about 4.8V. Odd. If the circuit was wrong wouldn't it be wrong in a consistent way?

Then a third chip. Finally, 2.501 V. At last. In the same circuit, just unplugged and replaced on a breadboard.

So I am wondering, does this happen often? Is "cheap" code for "too good to be true?".

Would possibly the eBay seller have got a batch of rejected chips and be trying to pass them off as "brand new, in working order"?

[RuggedCircuits]

Is "cheap" code for "too good to be true?"

Sometimes.

Here's a fairly good recent article on the issue:

http://www.eetimes.com/electronics-news/4229964/Chip-counterfeiting-case-exposes-defense-supply-chain-flaw

--
The Gadget Shield: accelerometer, RGB LED, IR transmit/receive, speaker, microphone, light sensor, potentiometer, pushbuttons

[cmiyc]

Perhaps the devices are sensitive to static discharge?  (Just because you don't feel or see a spark doesn't mean one didn't happen.)  In which case, simple handling might have damaged them.

Would possibly the eBay seller have got a batch of rejected chips and be trying to pass them off as "brand new, in working order"?

This sounds the most likely to me.  It is possible the eBay seller is unwittingly pawning off faulty components.

When buying components from an untraceable source, counterfeits are always possible.

[retrolefty]

I've bought maybe 2-3 dozen semiconductor components assortments or small bulk offerings from Asia via E-bay in the last 4-5 years. All from China have been fine and great bargains, one buy of a dozen 3.3v regulators from Thailand were crap, none were at 3.3v output, they varied from 2.8 to 4.2vdc, but none were within +/- .3 of 3.3. It was as if they were rejects from the OEM and resold. So I sense have avoided Thailand listings, which is probably not really fair, but with so many Chinese sellers there is no shortage of sellers to choose from. So as a hobbyist I feel that I can get a lot for my dollar buying cheap Asian offerings, but if I was a business or reselling to others I would most likely stick with US main distributors like Digikey, Mouser, etc.

Lefty

[terryking228]

This is always somewhat of a worry.

I have a good supplier in Shenzhen, China who sells to many big customers and has very, very few problems. He knows what his sources are.

It's not worth it to go out on the street looking for the closeout/surplus guys (there are many)...

From this good supplier I can sell these 79 chips http://arduino-direct.com/sunshop/index.php?l=product_detail&p=214 for $10 and still make some money.

The other problem with counterfeits is when there's a chip crunch and prices for something from the "chip finders" go way up.  Some of those have been out-and-out fakes with phoney labelling on dead packages.  Fortunately I don't buy any chips that cost more than $1 

DISCLAIMER: Mentioned stuff from my own shop...

[pwillard]

This issue does crop up now and again.

In addition to outright counter-fitting to make a buck I have an idea that many large corporations have "incoming test" organizations that validate that the chips they ordered meet specifications.   They send back the ones that don't meet specifications for credit.  So, eventually, there could be a batch of out of spec parts somewhere that may be good enough to sell to hobbyists but not to industry,  so they get purchased by vendors who attempt to do just that.

[terryking228]

I designed Chip Testers at IBM, at a time when IBM was starting to also use chips manufactured by other foundries.  IBM sent teams out to those manufacturers and defined the testing they had to do before shipping to IBM, even what test equipment they had to use.  So the "incoming test" was pushed out to the source.  I understand the biggies like Apple, HP etc. do that too.

Of course some sampling was done, just like the Qualification Lab people did on our own chips :-)

[cmiyc]

Very few consumer-focused companies do any testing of the incoming components.  The cost to setup the testers is outrageous for the return.   Plus almost no consumer-focused does their own builds anymore.  They all rely on contract manufacturers, few who have the expertise to develop the kind of screening necessary.

These surplus reels of faulty components are coming from the original factory.  Rejects go into a bucket (literally).  When a reeling machine goes offline, someone takes the bucket over to it, reels some parts, and prints a label. 

[nickgammon]

Examined  under a magnifier, the chips appeared to have the correct code etched into them. It would be a strange counterfeiting operation to make tiny chips that sell for 6c each retail. I could understand counterfeiting an expensive chip, like a CPU.

I usually am careful handling chips, regarding ESD. However perhaps the supplier wasn't as careful.

It seems most probable that they just onsold a reject batch, quite possibly not realizing it themselves. Maybe they got 1000 of them "as a bargain" from somewhere.

[retrolefty]

It seems most probable that they just onsold a reject batch, quite possibly not realizing it themselves. Maybe they got 1000 of them "as a bargain" from somewhere.

That would be my best guess also. The so called 'grey' chip market (sold outside the chip manufacture and it's authorzied distributors) always carries some risk I guess.

Lefty