Author [NL] [FR] [ES] [DE] [SE] [IT] TPS on CAN Bus polling slowly  (Read 13557 times)

0 Members and 1 Guest are viewing this topic.

  • Offline chuckxc   au

    • Tiger Jedi  ‐    1504
    • *****
      #10

    Offline chuckxc

    • Tiger Jedi
    • *****
    • Posts: 1504
    • Bike/Model: 12XC, 18XCX
    • Country: au
    Re: TPS on CAN Bus polling slowly
    Reply #10 on: Dec 10, 2023, 10.57 am
    Dec 10, 2023, 10.57 am
    Thanks HMC, that's interesting.
    I thought it could have been some kind of redundancy because of the critical importance of the throttle. Not sure why one sensor is half the voltage of the other. Could that half voltage sensor #2  be the source of throttle signal for the 'limp home' failure mode?

    Laterally unstable unless moving.

    My third Triple - 1976 Laverda 3CL Jota
    My 4cyl grunt - 2005 Honda CB1300F

  • Online T800XC   england

    • Premier Member
    • Tiger Jedi  ‐    3053
    • *
      #11

    Online T800XC

    • Premier Member
    • Tiger Jedi
    • *
    • Posts: 3053
    • Bike/Model: Tiger 900 RP
    • Country: england
    Re: TPS on CAN Bus polling slowly
    Reply #11 on: Dec 10, 2023, 02.21 pm
    Dec 10, 2023, 02.21 pm
    There's a bit more to this than it may first appear.

    Yes there are often 2 sensor readings and Triumph have some DTCs to cover certain sensor correlation errors, but on the 900 (at least from my current testing) the Twistgrip #1 sensor reads 1.00V to 4.49V, TPS #1 sensor reads 0.62V to 3.89V, and Twistgrip #2 & TPS#2 sensors both report the same value range of 1.49 to 4.12V. To add to this mix, not all of these are linear.

    ...happiness isn't happiness without a violin-playing goat...

  • Offline HMC

    • Tiger Member  ‐    18
    • **
      #12

    Offline HMC

    • Tiger Member
    • **
    • Posts: 18
    Re: TPS on CAN Bus polling slowly
    Reply #12 on: Dec 10, 2023, 07.44 pm
    Dec 10, 2023, 07.44 pm
    "Not sure why one sensor is half the voltage of the other. Could that half voltage sensor #2  be the source of throttle signal for the 'limp home' failure mode?"
    I don't believe so, I suspect the half voltage allows checking for a short between the 2 sensor wires, and the limp function would use the signal it saw as plausible albeit limiting performance and revs.
    Regards HMC

  • Offline chuckxc   au

    • Tiger Jedi  ‐    1504
    • *****
      #13

    Offline chuckxc

    • Tiger Jedi
    • *****
    • Posts: 1504
    • Bike/Model: 12XC, 18XCX
    • Country: au
    Re: TPS on CAN Bus polling slowly
    Reply #13 on: Dec 10, 2023, 08.48 pm
    Dec 10, 2023, 08.48 pm
    Yes, good point, if they were the same voltage you couldn't distinguish between a failure and a short, as you point out HMC.

    So I guess we can say, when it comes to safety, two throttle sensor inputs are better than one! Certainly a good way to sense a fault. I feel safer already.

    And as you point out Rich, sensor # 2 and TPS #2 readings track the same value. But that could just be the ECU using only the sensor input and duplicating it on a TPS#2 output reading because, as far as I can determine, there is no physical TPS #2 device or any signal input to the ECU.
    Last Edit: Dec 10, 2023, 08.58 pm by chuckxc
    Laterally unstable unless moving.

    My third Triple - 1976 Laverda 3CL Jota
    My 4cyl grunt - 2005 Honda CB1300F

  • Online T800XC   england

    • Premier Member
    • Tiger Jedi  ‐    3053
    • *
      #14

    Online T800XC

    • Premier Member
    • Tiger Jedi
    • *
    • Posts: 3053
    • Bike/Model: Tiger 900 RP
    • Country: england
    Re: TPS on CAN Bus polling slowly
    Reply #14 on: Dec 11, 2023, 02.45 am
    Dec 11, 2023, 02.45 am
    *Originally Posted by chuckxc [+]
    And as you point out Rich, sensor # 2 and TPS #2 readings track the same value. But that could just be the ECU using only the sensor input and duplicating it on a TPS#2 output reading because, as far as I can determine, there is no physical TPS #2 device or any signal input to the ECU.

    Without looking back at the 800's service manual I can't confirm for that model, but on the 900 there are 2 throttle position sensors in the one TPS housing, each providing their own signal to the ECU. There are also correlation error DTCs linked to the TPS signals so faults can be detected.

    The outputs from the two sensors have different min, max & range voltages but the output vs position curve shapes track one another.

    ...happiness isn't happiness without a violin-playing goat...

  • Offline chuckxc   au

    • Tiger Jedi  ‐    1504
    • *****
      #15

    Offline chuckxc

    • Tiger Jedi
    • *****
    • Posts: 1504
    • Bike/Model: 12XC, 18XCX
    • Country: au
    Re: TPS on CAN Bus polling slowly
    Reply #15 on: Dec 11, 2023, 05.34 am
    Dec 11, 2023, 05.34 am
    Yes, now that I have a closer look at the 800 TPS it appears there are possibly two outputs so, like the 900, two TPS sensors are processed.
    And the same behaviour as you observed on the 900, the position curve shapes track.
    Laterally unstable unless moving.

    My third Triple - 1976 Laverda 3CL Jota
    My 4cyl grunt - 2005 Honda CB1300F

  • Offline de_fault   ch

    • Tiger Cub  ‐    3
    • *
    • Topic Author
    • #16

    Offline de_fault

    • Tiger Cub
    • *
    • Topic Author
    • Posts: 3
    • Country: ch
    Re: TPS on CAN Bus polling slowly
    Reply #16 on: Dec 12, 2023, 10.03 am
    Dec 12, 2023, 10.03 am
    *Originally Posted by T800XC [+]
    The 530 data can quite easily be converted to twistgrip percentage (more on that later), although the maximum value I've been able to record is 84.4% for the full throttle position. I'm not sure why Triumph have limited the maximum value in this way, but it is linear from zero-to-full throttle.

    The Service ID 22 data shows the twistgrip percentage (closely matching the CAN 530 data) but it also shows the TPS percentage. For the same twistgrip position, the TPS values vary depending on the riding mode that's selected, but the twistgrip percentage values remain constant across all modes. This makes sense because the throttle opening (as reported by the TPS) is defined by the engine map being used for a given scenario. I've plotted some of this data and will try to post it in a reply later.

    The advantage of using CAN 530 data to read the twistgrip position (or percentage) is that it's constantly streaming on the CAN-bus so you only need to read the values. The Service ID 22 data on the other hand has to be polled, so is less convenient.

    I was hoping that was the actual TPS and not the twistgrip, perhaps I was too optimistic. One of the data points I need is the true TPS as I'm also planning on flashing the ECU with a new map with TuneECU. I already have the RPM (ID 0x518) and I welded an AFR sensor to the header so I'm only missing the throttle. Next up would be doing a few runs to gather data.

    I've read from your other posts roughly how to poll Service 22 with an ELM327, very useful, however I'm using a MCP2518 with a transceiver from the previous KTM project so I have to do everything at a much lower level.

    If I understood the other post correctly, to get the VIN (example) I would have to build an extended remote frame like this:


    based on https://en.wikipedia.org/wiki/CAN_bus#Extended_frame_format




    If this is correct the next step is figuring out:
    • what DA and D5 mean in the header (are they always the same or do they change based on the requested data?)
      *Originally Posted by T800XC
      "... In this case, 'DA' is the type byte, 'D5' is the target byte and 'F1' is the source byte and is the generic one for OBD scan tools."
    • the correct data identifiers for what's available through service 22 (how should I even go about this). For now I'll stick to looking for the TPS without getting overwhelmed
    • the CRC algorithm. This shouldn't be too difficult by either snooping around and analyzing the data I already have or checking the internet for info. Do let me know if it's a gargantuan task tho.