Fake XILINX Spartan6 chips from Aliexpress?

[Maxis]

Dear forum members,

I've decided to put my problem in writing since I've never encountered anything like this before. For many years I use Aliexpress as a reliable (now very questionable) source of SDRAM/FPGA/uC/Flash and many other expensive chips, which I can get for peanuts. I have saved tens of hundreds of $$ by sourcing the ICs for my prototypes from Aliexpress. And so far I've never had a single problem with the originality. I.e. in the worst case the chip was unsoldered before being packaged into the tape (this happed twice), but it was original.

Now, recently I've been looking for Spartan6 LX75 chip in fgg484 BGA package. And I bought two XC6SLX75 from the Aliexpress vendor. Seems like they have thousands in stock but nobody before me dared to buy these chips from them.

To cut the story short:
- Upon reception I've noticed little bit uneven shape of balls (assumed that this is process), and some very little color gradient on the PCB carrier of the chip on the bottom side.

Now, I've soldered both ICs and they have the very same result. JTAG is accessible, no short circuits, idle current is a bit high (Vcore Istandby=100mA), but device isn't recognizable by Xilinx since the ID doesn't match any known device to Xilinx. Also something pulls IO_L65P_INIT_B_2 pin low internally with the current of 50mA.

Since both chips behave closely (the Idle current differs by 30% between the two chips) I wonder whether this chip lot can be a fake?


I've contacted the vendor and they told me that this is the original lot.

Have you seen anything like that? What would you do in this case? I haven't lost a lot (40$ only), but mainly I'm cautious to buy this article again from Aliexpress....

[Gyro]

The PCB colour gradient combined with the uneven ball shape might point to the devices having been 'blowlamped' off E-waste PCBs and then cheaply re-balled. Being overheated might account for higher current draw too.

I'm not sure about the ID issue though, or the volume available. It seems unlikely that Xilinx would loose control of reject parts.

[Psi]

Could be a custom XILINX part, something they made for a specific vendor and not available or supported publicly, and that's why the ID is non standard.
The people in china pulling chips off PCBs wouldn't know if it's a custom part and not something generic.

[Maxis]

Exactly - it seems to me that it is the Xilinx custom chip.

Here https://forums.xilinx.com/t5/FPGA-Configuration/Unknown-JTAG-ID-code-for-Xilinx-XC6SLX75-2FFG484I/td-p/1211706 I've placed the report. The chip is recycled - I found it out when using Acetone on the top of the chip. The original marking is well readable for exception of the speed grade.

And the original speed grade mark isn't "2I" it's more like A5I or something close to this, which I haven't found anywhere.

Likely these are the very special ICs. Especially the boot modes aren't the same. Initially I thought that the chip isn't soldered properly (I had 2 ICs from the same seller). I've checked all the ESD diodes - all fine. Mounted the second chip - same thing. Then I've started digging up and found that the chip is not described in Xilinx public documentation. It is a variant of XC6SLX75, but a different one.

Sad news for many amateurs (big companies tend to buy the ICs via the network of official distributors), since the re-cycled and re-marked chips are getting closer and closer to the original. Seems like almost all these chips on AliExpress are recycled.

[noobiedoobie]

If its some consolation... wont all this be temporary? once the global chip shortage passes and supply chains are back to normal these things will reduce drastically