STM32F401RBT6 128KB Flash or 256KB

[srdjan]

Posted on January 13, 2017 at 13:50

Hello,

I have a funny question. I have STM32F401RBT6 and my firmware is ~140KB in size. I can download it without any issues, but the datasheet for this chip says that it should have only 128KB of Flash?

Also when I query the ID_FLASH_ADDRESS register it returns 128.

The code downloads and verifies properly.

Also when I use the ST Link utility, and modify bytes in Flash between 128KB and 256KB they are stored correclty. I'm confused by this. Does this chip has 256KB of Flash or not?

Thank you,

[Oliver Beirne]

Posted on January 13, 2017 at 14:08

Hello

Stokic.Srdjan

I have moved your thread to the

https://community.st.com/s/topic/0TO0X000000BSqSWAW/

where someone should be able to assist you.

Many thanks

Oli

[srdjan]

Posted on January 13, 2017 at 14:13

Thank you,

I'm still getting used to the new ST forums, and I posted the question in a

bad place!

Srdjan Stokic

Mobile: +389-78-835-505

On Fri, Jan 13, 2017 at 2:09 PM, Oliver Beirne <

[Tesla DeLorean]

Posted on January 13, 2017 at 15:37

The 128KB and 256KB part share a common die, the cheaper part only has 128KB tested.

[srdjan]

Posted on January 13, 2017 at 15:53

Does this mean that if I test each chip once before using it, it would operate properly?

I have a stock of 128KB chips, and our firmware has grown to be bigger than this size, so it would be convinient to continue using this part, even if it involves testing each chip on a socket before soldering.

[Tesla DeLorean]

Posted on January 13, 2017 at 17:58

Potentially, but you'd have no redress against ST if it failed then or subsequently.

Tester time is quite a significant cost (parts/hour), and can be reflected in the price stratification.

Your testing should at least catch gross failures in the array if you take a couple of passes with reasonable test patterns.

All the 205/405 parts have historically shared a 1MB die. If you exceed the physical array size the processor will Hard Fault.

[Amel NASRI · Accepted solution]

Posted on January 13, 2017 at 17:59

Hi

Stokic.Srdjan

‌,

As already said by

Turvey.Clive.002

‌, only the 128 KB are tested by ST. This means that there is no guarantee on the behavior of the other 128 KB.

-Amel

[T J]

Posted on January 13, 2017 at 21:30

Hi,

actually all chips and processors , transistors and diodes have a testing sequence before the part number is assigned.

the extended 128k on your processor is likely to be fully functional except at extreme temperatures.

but still, it is best that you test for the odd 'stuck bit' in flash.( but it is unlikely to be faulty.)

Over many years I have used this extended memory on another processor without issue.

I have used extended RAM without issue also.

[Tesla DeLorean]

Posted on January 13, 2017 at 22:07

>>

have a testing sequence before the part number is assigned

Yes, but they may not be fully tested, especially with internal FLASH where the erase/program cycles push into multiple seconds. It is a function of testing 2X 128KB vs 256KB parts in a day/hour, not finding bad 256K parts that are shippable as 128K ones.

I don't buy the temperature thing either, they don't test over temperature, the wafer has parametric test sites used quantify the speed and performance of the transistors, and the metal/polysilicon layers, the temperature testing is done while validating the IC design over the process window prior to mass production.

Within the CPU/PERIPHERALS the scan chains may have been fully exercised to get a very high degree of coverage, but these are for the logic, not banks of FLASH. There may also be a number of self-test circuits.

There is wafer/die level testing to eliminate the grossly failing devices before the die are removed and packaged. This is usually a quick and less thorough test.