TPIC8101: integrator output

Hello,

    I'm still working on a project utilising TPIC8101 but have come across something that I cannot interpret without trial-and-error.

"D9 to D8 of digital integrator output followed by six zeros"

1) How is D9:D8 laid out? Left-to-right? right-to-left? right-justified? left-justified?

And another question or two:

2) what is the bandwidth of the bandpass filter?

3) if microcontroller clock frequency is 16 MHz, set OSCin frequency to 16 MHz?

  • And to answer #1 myself:

    "Byte 2" is formatted as: [ D9 D8 0 0 0 0 0 0 ]

    And of course, "Byte 1" is formatted as: [ D7 D6 D5 D4 D3 D2 D1 D0 ]

    I did end up finding and resolving the problem. I'm logging the data via serial to the PC via an custom graphing application I wrote. C++ uses serial 8-bit communications by default... where as C# uses 7-bit... which would explain the strange clipping I was seeing as I was graphing the knock signal/noise as an 8-bit value.

  • In reply to David Sobon:

    David,

    Sorry we dropped the ball on this one. Here is my feedback, this device is new to me and I had to dig up the information. 

    1) How is D9:D8 laid out? Left-to-right? right-to-left? right-justified? left-justified?

    You are correct.

    Byte 2: [ D9 D8 0 0 0 0 0 0 ]

    Byte 1: [ D7 D6 D5 D4 D3 D2 D1 D0 ]

    2) what is the bandwidth of the bandpass filter?

    The Q-Factor for this filter is 2.30. Therefore from the center frequency that you set, and the Q-factor I just gave you, you can calculate the bandwidth of the filter. 

    I used the following website for the above information: http://www.sengpielaudio.com/calculator-cutoffFrequencies.htm

    3) if microcontroller clock frequency is 16 MHz, set OSCin frequency to 16 MHz?

    You are not calculating anything for this. You are just sending a clock signal into the device between 4MHz and 24MHz and then setting the prescalar with the value you are sending in. Page 13 of the datasheet. You basically are telling the device what the input frequency you are sending in and it will set the internal divider accordingly. 

    Hope this helps,

    John G


  • In reply to John Griffith:

    This post is helpful however, is it possilbe to model the band pass filter somehow...maybe in Matlab?  Can TI disclose more about the design and architecture of the bandpass?  For example, is it IIR?, FIR?, what are the tap coefficients?

  • In reply to Tom Mathews:

    Hi Tom,

    The filter is a biquadratic (2nd order) IIR filter. Unfortunately, this was originally a customer device and most of all the modeling information was removed from the datasheet when it became a catalog product.

    Not sure if you will be able to simulate the filter with just the q factor, the type of filter and the fact that is has a gain of 2 in the bandpass region, but if you need more information I would have to loop in the current team that supports the device. I no longer work in that group.

    Let me know what you think.

    Thanks,

    John

  • In reply to John Griffith:

    It may be possible to reconstruct a satisfactory filter model from what we do know: (IIR, second order, known center frequency, Q=2.3, Passband gain = 2). 

  • In reply to Tom Mathews:

    Tom,

    I'm glad John was able to help.  If you need more details beyond what John has shared, please start a new thread and I'll look into it.

    Thanks,

    -Clancy