SparkFun gets counterfeit parts, buys nitric acid and a microscope, finds lack of silicon.
You may or may not be aware, but there was a pretty wicked shortage of ATmega328s back in December of 2009. Both DIP and the surface mount TQFP packages were hard to find. We had a good mix of sources, and had plenty of ICs coming in, but at the same time our volume was increasing at a blistering pace so we quickly saw the brick wall coming towards us. To avoid going out of stock we started looking for any supplier that had stock.
We normally buy buttons, connectors, and passives from our suppliers in Taiwan, Shenzhen, Hong Kong, and Ningbo. This was our first time with this supplier, and the first time we ordered anything more complex than an LED, so we sort of expected this, but it's a fun lesson all the same.
On the left is an IC from one of our US suppliers (Digikey, Arrow, Future, or Mouser). On the right is what we got from our new supplier.
There's all sorts of weird issues with the ICs that came in from the questionable supplier. We first noticed the date code. On the left we have '1004' meaning this IC was fabbed the 4th week in 2010. The fake on the right reads '0723' or the 23rd week of 2007. But was the ATmega328 even being produced at that time? The oldest post I can dig up is from EMSL dated September 27th of 2008. I think they were some of the first people to actually touch an ATmega328. So it's pretty unlikely these ICs were fabbed in the middle of June 2007, a year and three months before Windell and Lenor (and Jellybean) got a hold of one...
Needless to say, we were suspicious. So before we went on a binge of product building, we built a few units for testing. None worked. We poked, probed, read the datasheet. Maybe the ICs are pulls and came pre-programmed? Maybe the ISP fuse is blown? But alas, nothing worked so we decided to talk to Lindsay, our resident biologist (with friends in chemistry) to get us some access to nitric acid. If Bunnie can acid etch ICs, why can't we?
Chip descapsulation is actually quite nasty requiring hot nitric acid (we had to get our acid bath up to 150C/300F), serious fume extraction and lots of safety gear on hand. Do not try this at home! But find a friend - this is just way too cool...
Under the hood: On the left is the ATmega328. The gold dots around the black border are the remnants of bonding wires that connect the die in the middle to the pins/legs on the outside of the chip. On the right is the 'slug' or counterfeit ATmega328 - there is a lack of bonding wires, and really anything. It looks like a chunk of copper. Better than coal I guess.
The ATmega328 die got damaged a bit with mechanical abrasion. With the 65% nitric acid that Lindsay used, it still took over an hour of bathing and scraping the ICs to get under the hood.
I'm not sure what it is (maybe copper?), but it certainly isn't anything silicon.
The markings on the back of the ICs were a little disconcerting as well. They obviously didn't match, but the counterfeits had 'Taiwan' embossed on the bottom of the IC. Upon closer inspection of the legitimate ATmega328, it says Taiwan (lower right corner) as well so this marking was actually not too bad. The overall thickness of the ICs did not match either.
The reel that the counterfeits came on had pretty good looking labels. We get our regular ATmegas in trays rather than tape and reel so we couldn't compare packaging, but it's obvious someone tried pretty hard to make these look legit.
So what did we learn from this? Trust your gut instinct. From the beginning we knew something wasn't quite right. This is the very first transaction out of over 7,000 where we ended up with the proverbial chunk of coal. Is there anything we can do? Not really. We will certainly never see our money again.
But surely SparkFun has insurance or some protection! From who exactly? While this may sound like it's horrendous, we actually hold great power. With over 7,000 transactions, we have never been burned. Why? Because with most components and suppliers, there is plenty of competition: if any one supplier causes us pain we simply never use them again and reward the suppliers that treat us well. Before these counterfeit ATmegas, we had nothing but very professional transactions. There are far more legitimate businesses that are hungry for our business in this world than there are slight-of-hand professionals. We were in a pinch and didn't vet the supplier like we should have. Live and learn.
We now have a reel of these ATmega328 Slugs, so feel free to pickup your own piece of SparkFun history... For you know, soldering practice, or miniature skeet shooting.