Possible bug in AN3109

[diego239955_st]

Posted on March 20, 2010 at 09:52

Possible bug in AN3109

[damh]

Posted on May 17, 2011 at 13:44

This code is risky.

I don't see the initialization of the counters and DMA. DMA transfer size/counter should be below or equal 250. Is RxBuffer2_SW used for more than one DMA packet? If not, why are there 2 counter variables RxCounter1 and RxCounter? Why do you use copy instead of changing the DMA buffer? Changing DMA buffer is easier and faster ;)

[Tesla DeLorean]

Posted on May 17, 2011 at 13:44

http://www.st.com/stonline/products/support/micro/files/an3109.zip

The problem is that the code in the application note is ILLUSTRATIVE, it is not a general solution which can be dropped into your application. For example it disables DMA once it interrupts instead to reinitializing it for the next burst of data. I generally agree that the DMA Rx data should be double buffered, and set up as a circular transfer, too address a couple of latency issues. -Clive

[marbel]

Posted on May 17, 2011 at 13:44

Hi,

I have also been looking at the AN3109.  My goal is simply to remove as many interrupts as possible from my system. When I run the system, it generates interrupts periodically and is not at all depending on transfers on my UART TX line. I would have wished for a solution where the rising edge starts and resets the timer. Am I missing something here? // Martin