STM32L151RET6 chips faulty/bad/fake ?

[DeanA]

A local contract PCB manufacturer has asked me to look into whether a recent patch of STM32L151RET6 chips he has soldered onto customer’s boards could be faulty/bad/fake.  I’m not very familiar with these devices (yet) or the operation of the boards they are used on.
 
The newer chip on the left is not working as expected (ending with 1.3).  If the chip is replaced with one of the older batch chips (on the right, ending with 1.1) the boards work as expected.


The programming sequence seems to work the same for both devices (although verify comes up with an error for both, perhaps the memory is locked or something).
The boards do operate to a certain degree even with the newer batch chips, flashing LEDs and generating some beep tones, so the chips do have some operation.  We are still trying to get some more information from the original designer of the board, but he no longer works with that company so may be difficult.
Has anyone experienced similar problems with these STM32L151RET6 chips.  Is there known fake or faulty devices on the market or something.  Or is there some difference with the 1.1 / 1.3 that could be the cause?
One thing I thought was a bit strange, the barcode on the original packaging from the supplier has been blanked out.

Thanks in advance if anyone can shed some light on what the problem may be or where to investigate further.  I will try and find what the 1.1 and 1.3. signifies.

[tom66]

If the chip appears to operate 'as normal' I would say it is vanishingly unlikely to be a fake.  However, it is possible that they are recovered and cleaned up from older PCBs (though they look pretty good for this) or it's possible there's a pin compatible variant e.g. without an ADC peripheral that has been remarked.  That could create issues as software would appear to work but as soon as a missing peripheral was accessed it is quite likely to experience 'undefined behaviour'.   

If the STM32's have a device ID, it's worth trying to extract that and comparing it.

It's also worth getting the customer to write some test firmware that exercises the chip, rather than the full suite of software.  Hopefully they have the source code from the original software before the engineer left them, otherwise they are up a certain creek without a paddle.

It's not uncommon to blank out supplier barcodes on greymarket products, as these parts are often sold without authorisation and the supplier doesn't want to lose any deal with ST.  We've had Zynq FPGA's with their 2D barcodes erased, on each device, in the past.

[DeanA]

Ok, maybe nothing to do with the marking in bottom right hand corner (11 or 13), looks like it's just date code as shown in datasheet

[pcprogrammer]

Take a look at the chips in your photo. The one on the left is Y 0 W 51 and the one on the right is Y 7 W 34

A bit of a problem with the year is the single digit, so hard to say how old it really is.

No idea about the 13 and 11 though.

[HwAoRrDk]

Year '0' is probably 2020, and year '7' probably 2017. Especially when you consider that 2020 matches the date on the label shown.

Is it me, or is the style of the legs different between the two chips? The one on the right appears to have legs that are wider at the top where they meet the plastic.

[dietert1]

The right one in the image is probably fake. I just checked several STM32 in 64 flatpack and the pin 1 marking on all of them has a curved bottom, like the left one in the image.
Also i remember at one of the big distributors L151 was marked as "not recommended for new products".

Regards, Dieter

[DeanA]

The right one in the image is probably fake. I just checked several STM32 in 64 flatpack and the pin 1 marking on all of them has a curved bottom, like the left one in the image.
Also i remember at one of the big distributors L151 was marked as "not recommended for new products".

Regards, Dieter

Ok, thanks,
Would it be normal for fake chips to have some functionality but not work 100% as normal, seems a bit strange to me, unless they are ST factory reject on something.
I will see if I can try reading device signature and things with the programming software but I would need to do some more homework on that.  I have ST-link hardware so I could probably use that to dig deeper,  I did not recognise the programming software and hardware the customer was using, I'll have a closer look.

[dietert1]

Don't know whether a tool exists to verify a STM32 MCU according to reference manual, at least i am not aware of it. If the application had proper error reporting, it should tell you what is wrong/missing.
Maybe the developer did not consider the error sheet. There are always subtle differences between silicon revisions.

Regards, Dieter

[HwAoRrDk]

There is this, by forum member techman-001:

https://mecrisp-stellaris-folkdoc.sourceforge.io/bluepill-diagnostics-v1.6.html

Although it is aimed at F103 chips. May or may not work for an L151. Perhaps some of the same checks apply.

[jmelson]

The right one in the image is probably fake. I just checked several STM32 in 64 flatpack and the pin 1 marking on all of them has a curved bottom, like the left one in the image.
Also i remember at one of the big distributors L151 was marked as "not recommended for new products".

Regards, Dieter

But, the OP said the one on the RIGHT works fine, it is the LEFT chip he is asking about!
Jon

[voltsandjolts]

Newark has a few hundred of these now, if you want to grab some genuine ones for testing/comparison.
https://www.newark.com/stmicroelectronics/stm32l151ret6/mcu-32bit-32mhz-lqfp-64/dp/42Y1069

[tom66]

An STM32 that's in stock?  Say it ain't so!

[DeanA]

Newark has a few hundred of these now, if you want to grab some genuine ones for testing/comparison.
https://www.newark.com/stmicroelectronics/stm32l151ret6/mcu-32bit-32mhz-lqfp-64/dp/42Y1069

Yes, thanks, we plan to get some of those for comparison.

[DeanA]

Year '0' is probably 2020, and year '7' probably 2017. Especially when you consider that 2020 matches the date on the label shown.

Is it me, or is the style of the legs different between the two chips? The one on the right appears to have legs that are wider at the top where they meet the plastic.

Yes, I had a look under the binocular scope and the 734 datcode chips (the working ones) have wider leads at the top.

[dietert1]

The right one in the image is probably fake. I just checked several STM32 in 64 flatpack and the pin 1 marking on all of them has a curved bottom, like the left one in the image.
..

But, the OP said the one on the RIGHT works fine, it is the LEFT chip he is asking about!
Jon

So, then it's not about fakes but about silicon revisions and it's worth looking at the error sheet. STM won't hesitate to apply changes for resolving known hardware errors.

Regards, Dieter

[tom66]

Both chips show a few signs that they have been removed from PCBs, something I didn't really notice too much the first time I looked.

For the right hand chip from older board this might be expected (maybe desoldered to inspect?), for the left hand chip did it come off the tape like this?  Or is that also removed? 

Leads bent, and a glossy look to the leads in places that looks like it could be excess solder.  Hard to say for sure but I would certainly expect excellent coplanarity for a brand new chip.    Worth checking a few parts for signs, also look for dirt/flux stains/IPA on the chips.