Counterfeit parts, in MY bins?

[phaeton]

It's more likely than you think...

Occasional lurker, second time poster, so please forgive me if I'm in the wrong place or talking about the wrong stuff... Anyway, I'll try to keep this concise:  I have a dilemma, and it's my own fault.  Over the years I've built up my inventory of parts from Radio Shack, Hamfests, Jameco, Mouser, Tayda and Futurlec.  Over the last several years I've also been buying parts and small parts lots on ebay.  I've recently come to grips with the fact that I likely have some parts that are counterfeits/fakes, factory seconds, or otherwise junk.  Radio Shack, Jameco, and Mouser sources are tried and true, and I also want to believe the same about Tayda and Futurlec.  Any fakes I might have will likely be from ebay.  The problem is, they're all intermixed.  For instance, I have probably 40 or 50 TL061, TL071, TL072, TL082 in DIP-8 packages in my parts bin, and I don't always recall which ones came from where.  Repeat that for about a dozen or two different IC types.  I have GOBS of stuff.  After reading a bit about it, it sounds like picking out the wheat from the chaff is going to be an ordeal.

My best strategy so far:

1) Pull out all the obvious stuff- things with bad logos, date codes that don't make sense, or ones where the appearance deviates from the datasheet (see question 1).  A quick look and I noticed a set of 8 or 10 TL072s in my collection with all the same 'look' and lot code, but the lettering was all shifted left or right from chip to chip.  So those are all suspect.

2) Test the non-obvious stuff-  This is where I face the punishments for the errors of my ways.  Again, I will sort them into similar sets, based upon appearance or lot codes. 

Quick tests might be the acetone wipe or the scrape test, but others would mean reading the datasheet and figuring out how to put them in a circuit to exceed published specs somewhere.  I'm sure there are established but non-destructive tests for semiconductors, but I don't know them.

3) Hope for the best.  There are a handful of parts that are still in tubes or distributor packaging that I can either remember or verify the source, but there are a lot that are not.  I'm not wholly confident that I'll be able to ferret out all the crap, but I can probably at least get some of it.


I go back and forth between just tossing everything and starting over, but I know there are lots of good parts in there too.

My questions:

1) Do packaging styles change over time?  If I pull up a current datasheet for a dual opamp, can I assume that the image it shows of the packaging type has stayed the same since the beginning of time?

2) What gets counterfeited most often?  Obviously the more expensive the more likely, so microcontrollers, CPUs, FPGAs and stuff, but do they even bother with TO-92 BJT transistors, linear voltage regulators and diodes?  Passive and electromechanical stuff is easy enough to measure or verify visually, and probably least likely to be faked.

3) If any of you have found yourself in a situation like this, how did you approach it?  I would appreciate any recommendations.

Again, I understand this is my own fault and I'll never buy stuff from ebay again.  Thanks for any suggestions, and sorry.

[ArdWar]

1) Do packaging styles change over time?  If I pull up a current datasheet for a dual opamp, can I assume that the image it shows of the packaging type has stayed the same since the beginning of time?

Yes certain parts may change a lot, especially now where chip bonding and packaging are often outsourced. With certain parts it feels like PCN are flying around every other batch.

[magic]

Fake jellybean opamps are typically something very cheap, like a Chinese clone of LM358 or RC4558 with fake markings. Once upon a time I have decapped a dozen different junk opamps from AliExpress and all of them were like that (link).

LM358 is easy to spot because it works down to the negative rail (as a voltage follower, for example). RC4558 is a bipolar opamp almost identical to TL072 (but BJT instead of JFET inputs) so it may be less obvious. Bias current is the most important difference, slew rate is another one.

I have also seen some "high end" JFET opamps which were NE5532 inside, or some unidentified (yet) FET input substitute, and I suspect it may have been CMOS rather than JFET. If that's the case, then the main difference will be increased low frequency noise.

[Kleinstein]

There is essentially nothing to simple to be a fake - there is a report about false 0 ohms resistor. So 1 cent parts.

Fakes are especially found with somewhat more expensive parts, that are in some demand for repairs / DIY. So slightly exotic (e.g. extra low noise, parts with good reputation and not longer made) TO92 BJTs are very well candidates (will likely end up as a random cheap BJT).
Higher power audio BJTs are also common fakes - may work initially, but with a higher chance to fail due to a lower power part.
From the used marked there are also recycled parts - so not intentionally false parts, but reclaimed parts that could be damaged or the wrong type by mistake.

The cases and lables did vary over time, especially change in the fab (the packaging part) and printing technique. Old lables were often rather poorly printed new ones tend to be better and maybe laser written. Especially cheap parts can vary.

[thm_w]

Look at your ebay purchase history.
Separate that stuff out if you want to, specifically opamps as mentioned.

Basic logic ICs might be fine.

[phaeton]

Thanks for the tips everyone.  I've been going through my ebay history as suggested, and making lists.  I can also do that to some extent for Mouser, Jameco, Tayda, and Futurlec, but there's going to be some overlap.

I'm also working full-time and going to school part-time so this might be a slow process.  I think the most difficult will likely be the transistors, because they're so small.  And so numerous.  At one point I noticed that some of the TO-92 types were going out of print in favor of SMT, so I started 'stocking up' I guess.  I'd guess that any garbage ones are less likely to be re-marked types (except for maybe some JFETs) and more likely to be original factory seconds or non-working tosses.

I feel like I'm the #1 tosser here though.