Setting up Quadrature Encoder Mode in STM32Cube

Hi,

http://www.st.com/content/ccc/resource/technical/document/application_note/54/0f/67/eb/47/34/45/40/DM00042534.pdf/files/DM00042534.pdf/jcr:content/translations/en.DM00042534.pdf

You can find several working encode mode examples on the forum.

Thanks for the link to the Application Note. I am using ST Cube application to configure my timers. I am using TIM4 in Encoder mode. I set the counter period to 1000. TIM4->CNT does give the current value of the counter. Which is the correct callback function that is called once the counter counts up to 1000?

As I've said I don't CubeMX and use an old-fashioned pencil-and-printed-pin-table-from-DS method. For me this avoids the 'non-intuitive' part, but apparently this does not work for everybody either. Actually, my colleague who does the HW, uses CubeMX as a successor of the former pin-selector javoid (I forgot the name, it animated ellipses instead of cube-wireframes in About - I still see this as the most remarkable feature of these softwares ;) ). For new designs we assign pins and features in parallel, he clicking me playing the pen-and-highlighter game, these processes appear to complement themselves. He spends quite a lot of time clicking back and forth each time he encounters a grey; he acquired quite a knack for this during the years; but it must be said that he also does a lot (and by 'a lot' a mean 'a huge lot') of DS reading in the preliminary phases of design, so that's not just some blind clicking.

2 eurocents

JW

Hi All ,

   I am interfacing a quadrature encoder to TIM4 in the Encoder Mode using STM Cube with STM32F103. I set up the timer and configure the counter to counter till 1000. The function HAL_TIM_IC_CaptureCallback() is called on every interrupt. I am excepting a call back on when the counter overflow happen ie, when the count reaches to 1000 pulses in this case.

I my application i know for 1mm travel i will be getting suppose 1000 pulses, so i will set 1000 as the pre load count value and if the encoder count reaches to 1000 count i should get an interrupt.

Please help me with a solution.

Thanks and Regards,

Vineeth

Posted on September 01, 2017 at 00:25

First, I made a mistake of posting my question as comment, which will not be seen as a new reply with question. So here it goes.

Hi JW,

Board: NucleoF722 with STM32F722

I'm using TIM5 for quadrature encoder counter. Following you suggestion to configure the timer, I have question with reference to capture below. Since the encoder channel A and B are mapped to TI1 and TI2 and the encoder Mode is TI12, and the counter is counting on both rising and falling edges of TI1 and TI2 inputs.

Question: Why selecting Rising Edge instead of Both Edge in the polarity selection? (In fact the chip User Manual says 'This configuration (Both edge) must not be used for encoder.' and I don't understand why not?) 

Anyone know why the polarity must be set to Rising Edge (instead of Both edge)?

Best,

Brian

I don't Cube so I don't know what exactly is meant by 'Both edge', but if it's setting of  TIMx_CCER.CC1P (which you've read) then it's because that is a setting for the 'edge detector circuit', in the channel input, which in encoder mode (and a few other modes named therein) does not work as an edge detector but as a selectable invertor, thus has only two settings (and the 'both edges' is the third). The encoder decoding circuit in the slave module does take levels from the input channels (TI1FP1 and TI2FP2 signals, see TIM block schematics) rather than edges - it then detects the edges in itself.

JW

In other words the CubeMX is wrong under its Parameter Setting\Encoder\Polarity to have the selection of Rising Edge, Falling Edge, Both Edges. Because in encoder mode, regardless of TI1 Mode, TI2 Mode, or both TI1 and TI2 Mode, the counter should be triggered on both edges of each channel; it is not a choice. Therefore, the wording should be Inverted or non-inverted under Polarity selection.

Brian

JW,

'

[rant mode on]This is why I am not interested in cuboids and kin. At the end of the day, if you really want to use the chip and not only a subset of its features you need to go down and understand the working of it, so cuboids are only an unnecessary intermediary. Sparkling and fuzzing and technicolor, yes, I understand the selling points. I'm just sad ST spends money/time/people on this fluff rather than the right stuff. [rant mode off]'

I'm a newbie to STM32 tool sets and so I found Cube is helpful in understanding how the registers are acessed, and a quick way to see generated codes that initializes the registers -- in a strict sequential order sometime -- for the peripherals to work properly. I total agree about reading the registers and their bits meaning. In fact I write my own code since the Cube generated code is too long; i.e. Cube uses over 50 lines of C code to send SPI data, while I need only two lines of code:

while((SPI2->SR)& 0x0080); // if busy

As about the Cube wording, there is nothing wrong about using a proper wording when the timer is in Encoder  Mode. The GUI screen should be adapted to each mode of the timer usage. Why the encoder mode has to share the same description with other modes?

Best,

Brian