Ruling out fake STM32G474s

abond · ‎2025-08-15

Hi,

I'm facing some issues with some STM32G474RET6s, and want to rule out counterfeit chips, even if that seems unlikely.

After googling for some solutions, I confirmed the AP IDR register and the DBGMCU_IDCODE both hold a revision ID of 3, which I expected from the X code on the label. However, I noticed that the markings on the chip match the LQFP80 footprint, rather than the LQFP64 we purchased.

I have read the qualifications about how device markings can differ, but it doesn't hurt to ask: is this outdated documentation, or is there a possible issue?

Thanks.

waclawek.jan · ‎2025-08-16

Unfortunately, TN1433 has not been updated with the new marking style containing the 2D code. For some information, though, have a look e.g. here.

As @AScha.3 wrote, tell us why do you think it's a fake, there may be some other reason for the problems you are encountering. So far, I haven't heard about fake 'G4xx.

JW

PS. Just out of curiosity, did you take that photo using some AI enhancement?

View solution in original post

AScha.3 · ‎2025-08-16

Hi,

why you have the idea the G474 could be a fake ?

Is it not working ?

Wrong ID when connecting with CubeProgrammer ?

Where did you buy it ? On Chinese night market, at 20 ct /pcs ? Or from distributor ?

If you feel a post has answered your question, please click "Accept as Solution".

waclawek.jan · ‎2025-08-16

Unfortunately, TN1433 has not been updated with the new marking style containing the 2D code. For some information, though, have a look e.g. here.

As @AScha.3 wrote, tell us why do you think it's a fake, there may be some other reason for the problems you are encountering. So far, I haven't heard about fake 'G4xx.

JW

PS. Just out of curiosity, did you take that photo using some AI enhancement?

abond · ‎2025-08-16

I don't think it's a fake, but I'm trying to rule it out if I can. Everything looked legit except for the discrepancy in the markings, so I figured I'd ask.

Thanks for the change report. I'll reach out to my distributor to see if they have something similar for the RET6 that explains the discrepancy.

As for the photo, a coworker of mine took it. I suspect he used his phone's "magnifier", which may or may not have AI enhancements. I'll ask him on Monday and let you know here.

Thanks again.

allenfjordan · ‎2025-08-16

I took the picture with my phone then just cropped it to show the chip, no other enhancements or AI.

We're seeing some odd issues with the CAN bus, ADC noise, and a few chips dying unexpectedly with normal GPIO usage. Just trying to rule it out.

waclawek.jan · ‎2025-08-17

I see.

The reason why I asked is the black "Q", and a couple of recent experiences with weirdly autoprocessed phone-taken photos.

The symptoms you mention are quite varied. I can't comment on CAN. ADC in the 'G4 is probably less than optimal, despite this line being marketed as mixed-signal; the pressure to have minimal supply pins appears to be one of the reasons, see the 'G4-specific ADC appnote (AN5346) and there may be clues also in some posts at this forum.

JW

allenfjordan · ‎2025-08-18

I think that was just some debris on the chip that happened to look like a Q, hah. Is there anything else we can try to test for fakes?

waclawek.jan · ‎2025-08-19

> Is there anything else we can try to test for fakes?

I don't know of anything.

If you're still convinced this is the source of the issues, you can contact ST directly through their web support form, but they probably will just tell the obvious answer @AScha.3 gave above, namely that if it's from a reputable source then it's OK.

JW