F730 flash PSIZE vs. Supply voltage

[RMcCa]

I am a bit confused by the language in the F7 reference manual, section 3.3.5 (RM0431) that describes the flash program/erase parallelism. Does the power supply limit the maximum value of the parallelism or does it limit one to a particular size?

The caption for table 6 is "Table 6 provides the correct PSIZE​ values". Should it say "correct maximum PSIZE values"?

I am asking because i have been writing bytes to the flash with Vcc=3.3V and PSIZE = 0 and its been working fine for a while​, but i wonder if that's only luck and i should switch to 32bit writes. Speed is not an issue in this case.

[Imen.D]

Hello All,

Thank you for all reported issues.

This is raised internally for correction : even for incorrect comment in the stm32f7xx_hal_flash.c and the wrong max value for 32-bit program operation.

Best Regards,

Imen

[Pavel A.]

Pardon me for jumping in.

Does “max parallelism size�? mean “max parallelism size for given voltage"?

That's, lesser parallelisms also work at the same voltage?

And for the voltages - there is a minimal voltage for a given parallelism size?

So answer of the OP question is, it is always OK to write at PSIZE=0 ?

-- pa

[RMcCa]

I believe so.

It would probably clearer if they listed the minimum voltage for each size.​

[waclawek.jan]

+1

+1

[RMcCa]

Studying table 52 (pg 126) of the data sheet, it appears to take approx 10mW/byte to write to flash​, so there must be a limit to width based on the power available. This means that shorter writes than the max are OK.

[waclawek.jan]

That's not necessarily the only criterium, FLASH transistors are picky, the HV source and other required circuitry can be designed in a million ways, trading between dozens of requirements; and I've seen various weird and seemingly inexplicable requirements for FLASH (and EPROM) programming in the past.

JW

[RMcCa]

Thanks. This is all a result of unemployment and quarantine. ​the F7s are complicated and i can't imagine the task of ensuring the accuracy of the documentation, which could be a more important product than the chip itself.

[Piranha]

This is a very good question/issue. I was going to ask it some day, but @RMcCa​ did it already. So thank you and thanks to @Imen DAHMEN for clearing it up and fixing documentation.

I can add that I also have been using byte (8-bit) writes at VDD = 3,3 V on F76x for some time in different scenarios on Discovery, Nucleo and custom boards and haven't run into any problems with that.