This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Question about analog encoder solution for higher resolution

Other Parts Discussed in Thread: TLV2772

Normal digital encoder (quadrature encoder) will generate about x4 pulse in motor control. While analog encoder would be make use of sin/cos waveform to do interpolation and high resolution. Is TI has any solution to interface with analog encoder, to generate higher resolution?

Thanks very much

Michael

  • Michael,

    The C2000 group has largely dealt with TTL incremental encoders, however we have had some interest in sincos incremental encoders.

    At this moment, we do have an appnote for how to interface a sincos encoder with an older (now obsolete) C2000 device, the F240x family. 
    http://www.ti.com/lit/spra496

    However, because of interest we are relooking at this interface with some of our newer devices.  I will outline some ideas that we are considering below in the hopes that it gives you some ideas:

    1) Having both the sin and cos differential signals go through individual gain+offset filters to make the waveforms 0-3.3V and single-ended.
    2a) Bring each of these single-ended inputs to a comparator + ADC input.  The comparators' output can then be used to count how many periods have passed (either the comparator outputs could be directly tied to QEP inputs or potentially done via software).  In parallel, the core or CLA can do some atan postprocessing to extract higher resolution position information.
    2b) Same as above only use external comparators to generate the QEP inputs.

    (again note that the above has not been proven out yet)


    Thank you,
    Brett

  • Hi, Brett

    Thanks a lot for your very reply. The analog encoder has very high resolution over incremental encoder which use in some area where very high accuracy and control is needed. I did noticed that application note, which is the only one available.

    TI does atan postprocessing on Atan B/A (need both two channel inputs). In theory, atan processing will have high accuracy for a true SinCos signal. 

    What I understand a common practical approach is do sampling and inverting either A or B at different  phase, which is simpler demand on processing needed. This also valid as long as signal is more linear instead of sin/cos, from quite some analog supplier.

     I would like to use practicle approach. Would TI able to support it?

    Regards,

    Michael

     

  • Hi Michael

    Resolution depend on a lot of issues.

    First is the type of algorithm you use to perform R2D conversion (sin/cos -> angle). Obviously atan method is straightforward, but it is highly susceptible to noise. And on analog signals there will always be noise. Even more so if we are talking about power electronics. I find tracking method to give much more smoother signals, and you get speed information for free.

    Second is the sensor output.

    If sensor has an absolute output sin and cos signal have a period of one per one mechanical revolution. Here it will be quite a challenge to get above 14 bit resolution with integrated 12 bit ADC. It is doable, but with heavy filtering, which complicates control loop stability and limits speed range.

    If sensor has an incremental output so that sin and cos signal have a period of 2^n per one mechanical revolution it is somewhat easier to get a high resolution signal. SW becomes more complex (you do need to use QEP to get coarse value and tracking sin/cos gives you fine resoulution)

    As for the software you are more or less on your own.

    Regards, Mitja

  • HI,

    I had a technical query regarding the interfacing of the linear encoders for numerically controlled machine tools with TI C2000 microcontrollers.

    This is for interfacing electronics of 1VPP encoder, as mentioned in Heidenhain's technical literature (Linear Encoders for Numerically Controlled Machine Tools.pdf  Page -37). It is mentioned that +/-Ub shall be dual supply configuration +/-15V.
    http://www.heidenhain.in/fileadmin/pdb/media/img/571470-28_Linear_Encoders_For_Numerically_Controlled_Machine_Tools.pdf


    However, if you refer to TI provided solution for Heidehain encoders literature (SPRA496.pdf page 11) in figure no-5, a single supply configuration of 0-5V is used for interfacing with the micro-controllers.

    TI has used single supply TLV2772 operating at 0 to 5V.
    Can you please suggest:
    a. If the proposed solution by TI of single supply will work?
    b. What is the exact voltage level of A+ and A- signals coming from encoders? Is signal level for  A+ as [0.5Vpp + 2.5VDC] & A- as [-(0.5Vpp + 2.5VDC)]? This understanding is coming from U0 as 2.5V

     
    Regards,

    Prashant

  • Hi Prashant,

    Where in the Heidenhain manual are you seeing that +/-15V shall be used?  I tried doing a search, but I'm not seeing it.  Most of these encoder seem to be powered by a +5V supply.

    ---

    According to some of the Heidenhain docs:

    A+ = 2.5V +/- 0.5V
    A- = 2.5V +/- 0.5V

    if A- > A+, the signal is in one half of the sine wave whereas if A- < A+, the signal is in the other half.


    Thank you,
    Brett

  • Hi Brett,


    Thank you very much for your reply.

    I have attached a document from Heidenhain. Please refer page no-5 "Input circuitry of subsequent electronics". The subsequent interface electronics have a differential op-amp configuration with operating with dual supply UB = +/-15V.

    So, if I understand it right from your explanation , are A+ & A- both riding on+2.5V DC having AC component of 500mv with 180deg phase shift? Under same section in  above document referring to encoder electronics, A- = inv(U0 i.e. 2.5V+/-0.5V) = -(2.5V+/-0.5V)? Is this the reason for having dual supply op-amp?

    The power supply to the encoders is +5V as you rightly said.

    Sorry if I my questions are invalid. But would be grateful if you can help me out.


    Regards

    Prashant

    3683.Heidenhain_Interfaces.pdf

  • Hi Prashant,

    For what it's worth, I sympathize.  Heidenhain's specs can be tricky.

    3464333 said:

    I have attached a document from Heidenhain. Please refer page no-5 "Input circuitry of subsequent electronics". The subsequent interface electronics have a differential op-amp configuration with operating with dual supply UB = +/-15V.

    So, if I understand it right from your explanation , are A+ & A- both riding on+2.5V DC having AC component of 500mv with 180deg phase shift? Under same section in  above document referring to encoder electronics, A- = inv(U0 i.e. 2.5V+/-0.5V) = -(2.5V+/-0.5V)? Is this the reason for having dual supply op-amp?

    I see.  I assume these rails are as they are to guarantee good linearity.  However, really you just need to make sure there is margin on either side of your sine-wave so that clipping does not occur.  2.5 +/- 0.6 (for margin) = 1.9 to 3.1V.  Because of this, I don't see the dual supply as being strictly necessary.  Powering the opamp off of 0 and 5V is likely adequate.  As you're interfacing to a C2K please keep in mind that the input for the C2000's ADC can be 0-3.0V (for most older chips) or 0-3.3V (on most newer devices) so the output of you opamp needs to scale into this range.

    I don't see the A- = inv (U0) thing stated in the specific doc you've referenced.  What I've described above is my assumption, I haven't specifically used Heidenhain's 1Vpp interface but have used other 1Vpp (SinCos) interfaces and they behave like I've described.


    Good luck,
    Brett

  • Hi Brett,

    Thanks again for your explanation,

    Brett Larimore said:


    I don't see the A- = inv (U0) thing stated in the specific doc you've referenced.  What I've described above is my assumption, I haven't specifically used Heidenhain's 1Vpp interface but have used other 1Vpp (SinCos) interfaces and they behave like I've described.



    I have re-attached the same above document, refer to the sticky note on page-5 in reference diagram of encoder electronics. U0 is the input to the transmitter and A- is inverted signal of U0.

    Although, I presume, Heidenhain encoders would be no different from industry standard encoders you may have used and the understanding/representation in the specs can be tricky.

    Thanks again for your support.

    Regards,
    Prashant

    8233.Heidenhain_Interfaces.pdf