Announcement

Collapse
No announcement yet.

Frustrated

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Frustrated

    It is hard for me to understand why detailed information is not available for the Core Control Modules. Since these are in production, the specifications and theory of operation documents should be developed and available for release. Dataport Systems, Inc., the lead organization for HiTechnic and I believe Modern Robotics, Inc., Put out reasonable documents on the original FIRST HiTechnic motor and servo controllers. The specifications at, www.modernroboticsinc.com, provides limited information if you are willing to search through the titles and pull all of the pieces together. There is no information concerning the memory map and programming interface to the modules.


    I would like to say that the NXT had served its purpose well. I am not convinced that speed of the processor was a problem, however, lack of memory and restrictive I/O capability were both serious problems that needed to be addressed....and now have been.

    Let's look at some approximate numbers.

    A direct drive robot with 4" wheels will travel at a speed of approximately 31.415 in/sec

    We have been told that the system will complete a read/write update with the peripheral devices every 50ms. Therefore, the robot may move 1.57 inches between updates. Remember that the robot is controlled by the motor controller, not the robot controller. If there is a programmed encoder value in the controller, the robot will stop accordingly. Yes, I do know that this was a crap shoot under LabVIEW... but we do not have to concern ourselves with that now. The challenge remains the same that we must create algorithms that account for the system communications and signaling delays. We now have different delays that we need to experience based on the different programming styles we choose to use.

    Rounding down our quad core 1.2 Gig processor down to 1 Gig for system overhead gives us a usable 4Gig processing speed which equates to a possible 200,000,000 instructions being processed between peripheral memory map updates. Plenty of power there. This is of course considering the robot controller only since the driver station is only a game pad input and a HMI/communications device.

    Concerning some other data about the old system. Assuming that all 256 memory locations were downloaded by I2C to the NXT, it would take minimum of 200ms to transfer the data in one direction. I note that the NEW legacy module is listed as only transferring 27 bytes of data per channel which equates to about a 50ms transfer time for a 2 channel motor controller. This also gives us a reason why it is suggested that we do not daisy chain motor or servo controllers on the legacy modules. It will double the time to transfer the data. Adding legacy controllers is the best method to keep data throughput high.

    Something that is currently left in the ether is what kind of bottleneck is at the FTDI controller. Each of the Core Control Modules, (except for the Power Distribution Module, I assume), has a microcontroller that arbitrates the signals in that module. These microcontrollers may use a variety of serial communications protocols. The protocol used is converted by the FTDI chip to a USB protocol. I was assured at world by a programmer from Modern Robotics that the speed was fast enough not to cause any delays. He also related that they could upgrade the modules if necessary, (I assume buy flashing the controller). Being a system technician who hates black boxes, I would like to know what is really inside. It does not mean I will not treat it like a black box. I do my best teaching and troubleshooting when I can break down the internal signals and not work with just the module inputs and outputs. Discrete components to me are tubes and transistors. I have taught the maintenance of a computer system transistor by transistor.

    I am sorry if I have offended anyone... but I feel a little better now.

    I am looking forward to a great season! GO TEAMS!!

  • #2
    I like your way of thinking.

    Cheers, Ollie

    Comment


    • #3
      Rank on

      Originally posted by ChiefCT View Post
      ...I am sorry if I have offended anyone... but I feel a little better now.

      I am looking forward to a great season! GO TEAMS!!
      I thought the forums were where people go TO offend people, or is that only on Chief Delphi? (j/k, sort of)

      Anyway... All valid points, and clearly many share your frustrations over some portion of the roll-out, although (in my experience) very few teams have your level of advanced technical knowledge. i.e. my frustration is limited to lack of an SDK, and retail pricing data.

      As a Rookie team moving up from FLL, I doubt we'll be advanced enough to encounter problems at your level, but these kids have surprised me before, so we'll see how it goes.

      Thanks for sharing, rant on (if it helps), and lets all discuss it at our last year in STL.

      Michael
      Michael P Clark
      Founding Mentor, FTC 9958
      http://www.redfishrobotics.com
      "We're Hooked on FIRST"

      Comment


      • #4
        Originally posted by RedfishRobotics View Post
        I thought the forums were where people go TO offend people, or is that only on Chief Delphi?...l
        Now I'm offended.

        Comment

        Working...
        X