Getting a unique identifier for IMXRT1062

[nsi]

I'm trying to get a unique serial number for each device and was hoping to rely on the IMXRT's.

In the Reference Manual, I found a Unique ID in the fusemap.

Based on the answer from camelshoko in this thread, I created a code that read the data from the fusemap.
However, after trying it on several devices, it seems that the "unique id" is common to several devices and the part that changes is at offset 0x430 which is marked as "Reserved" in the RM.

Is reading data at offset 0x430 a reliable way to get a unique value for the chip ? If not is there another "proper" way ?

[Habib_MS · Accepted solution]

Hello @nsi ,

The reason that the unique ID is the same in some devices is because the unique ID only changes for different family parts. As mentioned in this post.

At the same time, you can read the offset 0x430 if it helps you to recognize the devices, but the reason of the value is reserved for internal use. As exception of the bits [19:16] that these are the Silicon Revision number as mentioned in the RM.

Also, if you experience any issue, do not hesitate to let me know.

BR

Habib.

[nsi]

Hello, thank you for your reply !

Just to confirm :

- Reading at offset 0x430 is not a reliable way to do it since the actual meaning of the value won't be communicated

- There are no identifier unique to the device available ?

[MultipleMonomials]

We spoke to our NXP FAE about this question, and it appears that @Habib_MS 's answer is wrong.  Our FAE Ramdas said:

Short answer – you are correct in that using OCOTP_CFG0 and OCOTP_CFG1 will give you a unique ID for every single MCU.

[...]

Please note that the offsets are 0x410 and 0x420. I believe the confusion in the thread referenced in your email below was referring to offset 0x430 of the OCOTP reg

[Habib_MS]

Hello @MultipleMonomials,
I apologize for the confusion. What I meant to say is that offset 0x430 indicates the silicon revision number, not the Unique ID. Sorry for the inconvenience. Also, you can detect the unique ID via CFG0 and CFG1 to identify devices, as says Diego Charles in his post.

BR

Habib. 

[Habib_MS]

Hello again @nsi, 

Define whether it is viable or not depends totally on your application, if the Silicon Revision number bits help you to have a difference, you can use these bits.

Also, unfortunately we have not a unique identifier to each device available.

Sorry for the inconvenience this may cause.

BR,
Habib.

[MultipleMonomials]

Wait... so, you're saying that the value described as a UNIQUE_ID in the datasheet is not, in fact, a unique ID?  That contradicts both this post from Diego Charles and this text from AN12419:

This Device ID is unique for each IC and can be read from the OCOTP registers HW_ OCOTP_CFG0 and HW_ OCOTP_CFG1. The eFuses are programmed during manufacturing.

We currently have implemented a system that relies on the unique ID via CFG0 and CFG1 to identify devices, and it has seemed to work fine on the devices we have tested.  However, we haven't started using it en masse yet, so I suppose we might not have noticed if there were occasional duplicates.  If we need to change how this system works, it's better to do it now rather than later.

Could you please double check whether this is the case and, if the ID is not, in fact, unique, could we please update AN12419 and the MIMXRT datasheet and/or errata sheets in order to indicate this?  Thank you!

[nsi]

Thank you for the information, I will look for another way to device that identifier.