RM0091 documentation error (stm32f0x1/0x2/0x8)

[frankmeyer9]

Posted on November 07, 2014 at 13:03

Can you spot the (copy&paste) error in the following snippet from the stm32f0x1 reference manual ?

If these MCUs are really capable of storing two bits in one, ST should probably file a patent application immediately ... Is it otherwise safe to assume that bit 29 of USARTx_CR1 is not properly documented ? worst-forum-software-ever reference-manual-error

[waclawek.jan]

Posted on November 07, 2014 at 13:30

> Can you spot the (copy&paste) error in the following snippet from the stm32f0x1 reference manual ?

No, I see only a white area below the question - probably forum software problem. Can you please post RM0091 version and page? Jan

[frankmeyer9]

Posted on November 07, 2014 at 17:43

DocID018940 Rev 6 (latest version available on ST webpage),

page 702, the very first lines, description of USART_CR1, bit 28, ''M1 : word length''; How can a single bit be ''[10]''... Such things could happen. But my sympathies for this forum software are very limited, to express it politely...

[zzdz2]

Posted on November 07, 2014 at 18:15

I would say these are two bits: M0(bit 12) and M1(bit 28).

[frankmeyer9]

Posted on November 08, 2014 at 10:38

You are right.

But then I consider it a design bug. Why spreading related bits over several words ? May it be someone had limited foresight when making the first design ?

[zzdz2]

Posted on November 08, 2014 at 11:08

It looks like ''compatibility'' design.

Older STM32s have a single M bit(12). They renamed existing M bit to M0 and added new M1 bit(28).I agree it's suboptimal but there is some reason behind it.

[frankmeyer9]

Posted on November 08, 2014 at 17:17

It looks like ''compatibility'' design.

... I agree it's suboptimal but there is some reason behind it. As I said, IMHO the reason behind is limited foresight when doing the first design. Actually, foresight is not the right word, since there had been 7, 8 and 9 bit serial protocols around years before the first ARM. Similiar to the infamous ''A20 gate'' in intel's x86 CPU designs.