TMC is an independent, primarily volunteer organization that relies on ad revenue to cover its operating costs. Please consider whitelisting TMC on your ad blocker and becoming a Supporting Member. For more info: Support TMC

Let the hacking begin... (Model S parts on the bench)

Discussion in 'Model S' started by wk057, Dec 14, 2015.

Tags:
  1. Andyw2100

    Andyw2100 Well-Known Member

    Joined:
    Oct 22, 2014
    Messages:
    6,542
    Location:
    Ithaca, NY
    So do I understand correctly that the data you would be able to capture from your P85D and then feed to your bench set up for performance monitoring would be more accurate data than what is available through the REST API combined with VBox monitoring?
     
  2. tom66

    tom66 Member

    Joined:
    Dec 17, 2013
    Messages:
    625
    Location:
    United Kingdom
    #82 tom66, Dec 16, 2015
    Last edited: Dec 16, 2015
    Holy...A drop of 100V under full throttle?
    I thought you were off on your 50V estimate but...wow. Yea. Tesla are getting every last drop out of the pack.


    This makes sense, as the gateway is integrated into the MCU. The gateway probably determines error codes, not the applications processor. (I believe there is a Freescale automotive processor, and an Altera FPGA on the MCU main board. The automotive processor probably handles the gateway interaction and the FPGA probably handles something to do with the display or camera...it's unclear..tied into a few large DDR memory ICs.)
     
  3. wonko

    wonko Member

    Joined:
    Mar 14, 2014
    Messages:
    58
    Location:
    Germany
    Could someone please get me the part number of the camera itself (the Tesla module number may help, too). I remember that in the very early days, some manufacturer was priding himself with having the Tesla contract, but I can't find that article anymore. It contained part numbers, and I was able to find the tech sheets including much of the digital camera format.
     
  4. smac

    smac Active Member

    Joined:
    Aug 4, 2013
    Messages:
    1,745
    Location:
    Nottinghamshire
    Well done on this project!

    Out of interest I note none of your throttle positions are 100% (seems capped at 98%) I wonder what an S85 would look like?

    We certainly know that throttle maps can be adjusted OTA, I wonder if there is something here to investigate (if performance tuning were the goal).
     
  5. wk057

    wk057 Senior Tinkerer

    Joined:
    Feb 23, 2014
    Messages:
    5,599
    Location:
    Hickory, NC, USA
    Yep, looks that way. Definitely faster than the REST API. The voltage and amperage are literally reported 100 times per second, and speed with 0.1 MPH resolution is reported 10 times per second. That's pretty awesome data.

    Yep, pretty much matches my estimates from a few months back. http://skie.net/uploads/packampsvshp.jpg

    I don't have this info handy, but I'll dig it out later if no one else steps up with it before I get a chance.

    Please excuse my throttle position part. The throttle position shown is kind of an estimate. I'm not 100% sure the range of that value so I just set the max value I saw as 100% in my parser. It's definitely throttle position, though.
     
  6. artsci

    artsci Sponsor

    Joined:
    May 10, 2012
    Messages:
    6,190
    Location:
    Timonium, Maryland
    There is no part number on the camera. Tesla has a part number but that will not help trace its origin. As I recall from open it up its made in Japan.
     
  7. PRJIM

    PRJIM Member

    Joined:
    Jul 30, 2012
    Messages:
    322
    Location:
    USA
    Omnivision makes the IC not sure who the Tier-1 is that does the integration (maybe Tesla themselves?)
     
  8. Shortmanz

    Shortmanz Member

    Joined:
    Jun 9, 2014
    Messages:
    58
    Location:
    Plano, TX
    Is this what you are looking for? http://www.ovt.com/release.php?release=263&lang=english
    It is a press release dated October 16, 2012 from OmniVision about Tesla using their OV10630 image sensor SOC running at 720p.

    I found the link on this page: http://teslatap.com/undocumented/
     
  9. CHG-ON

    CHG-ON Still in love after all these miles

    Joined:
    Jun 24, 2014
    Messages:
    3,079
    Location:
    Santa Cruz Mountains, USA
    You are WAY too smart for me! But how fun to watch you go through this.

    Thanks!
     
  10. wk057

    wk057 Senior Tinkerer

    Joined:
    Feb 23, 2014
    Messages:
    5,599
    Location:
    Hickory, NC, USA
    #90 wk057, Dec 16, 2015
    Last edited: Dec 16, 2015
    OK, so for kicks I went ahead and CAN logged a bit of P85D fun and plotted some graphs. This definitely verifies the ~1300A ceiling on the P85D power draw.

    launch1-CANgraph.jpg

    (parsed CAN data: http://skie.net/uploads/launch1-parsed.txt )

    Launch from 0 to ~67 or so. (No launch mode, still on v7.0 2.7.77). Interesting that the throttle position and power data are ahead of the speedometer data by ~0.3s or so. This was a hair up hill, pack was pretty cold. I think my "Pack temp?" variable is related to the pack temp, since it rises after launches... but I don't think it's actually correct since it's 48F or so in my garage. (Hence the question mark ;) ) Also appears that it either took about 0.6s for me to mash the accelerator (not likely), or that variable is feathered.

    launch2-CANgraph.jpg

    (parsed CAN data: http://skie.net/uploads/launch2-parsed.txt )

    Tires skipped a little on this one, and you can see it in the high resolution power output graph. Interestingly, time to 60 MPH was exactly the same in both launches and I got a reading of 60+ MPH for the first time 3.881 seconds after throttle position was > 0, or 3.600 seconds after the first non-zero MPH read. (Obviously no roll out nonsense here).

    The speedometer reading is 10 Hz (0.1s resolution) data and the power is 100 Hz (0.01s resolution / 10ms resolution).

    These were starting at about 90% SoC.

    Anyway, that's enough fun for today. :)
     
  11. Sogorman

    Sogorman Member

    Joined:
    Nov 8, 2014
    Messages:
    203
    Location:
    Phoenix, AZ
    @wk057, I appreciate all that you bring to this community! This thread is fantastic.
     
  12. Benjamin Brooks

    Joined:
    Jan 21, 2013
    Messages:
    583
    Location:
    San Jose, California, United States
    Very interested to see if pack 'true' SoC and cell / string voltages will be decodeable on the bus traffic
     
  13. smac

    smac Active Member

    Joined:
    Aug 4, 2013
    Messages:
    1,745
    Location:
    Nottinghamshire
    Out of interest (and sorry if I missed it) which CAN Bus logger are you using, and where are you tapping into the data line on your working car ?
     
  14. lolachampcar

    lolachampcar Well-Known Member

    Joined:
    Nov 26, 2012
    Messages:
    5,147
    Location:
    WPB Florida
    I've got CAN bus product from previous efforts along with some spare time. I am wiling to provide both to a forum member support effort (non-profit similar to the key fob project) should the group want to develop a Tesla CAN data gathering and analysis tool set. The hardware design has long since paid for itself :)
     
  15. wk057

    wk057 Senior Tinkerer

    Joined:
    Feb 23, 2014
    Messages:
    5,599
    Location:
    Hickory, NC, USA
    I haven't caught any CAN data that looks like it could be string voltages... but then again, the car is sending 2000+ frames per second on the power train CAN bus, so... lots to look through. Still holding out hope since the diagnostic screen shots we've seen suggests this data is there, just not sure if it's something always sent or not.

    I'm using EVTV's SavvyCAN box along with some custom hardware I whipped up previously. Most of the CAN buses are accessible on Tesla's diagnostic connector that is hidden above the little pouch that's under the 17" screen. The pouch is just held in with a couple of clips and pull right down.

    Sounds like a plan. Does your CAN bus stuff work at 500 kbit? That was an issue I was having with some of my cheap CAN hardware...
     
  16. lolachampcar

    lolachampcar Well-Known Member

    Joined:
    Nov 26, 2012
    Messages:
    5,147
    Location:
    WPB Florida
    My bits are based on a Freescale 32 bit device (running at over 50 MHz IIRC). The CAN engine is integrated but autonomous including address screening. It can handle CAN bus rates up to 1M and would have zero problem filtering, parsing and logging traffic from a 1/2M bus. I'll have to check but I seem to remember 8 Mbyte of on board flash. One of my older designs used an SD card but these particular boards were meant to be potted in epoxy and had onboard flash parts.

    Here is some diagnostics stuff from a man in the middle voltage (potentiometer) based suspension module. The wheel speeds are CAN data (from a 1/2M bus) while the wheel heights are local a/d and d/a data.
    New Page 1

    I normally use IXXAT for my logging and have found my hardware often catches stuff that the IXXAT does not. It is the benefit of a dedicated processor and no Windows overhead.

    I think I have a pile of boards still laying around. I'd put them into a project at cost if we could come up with a plug and play Model S harness and some good data viewing software.
     
  17. pgiralt

    pgiralt Supporting Member

    Joined:
    Jun 16, 2013
    Messages:
    1,520
    Location:
    Cary, NC
    Earlier in the thread you had mentioned that the speed was a 1-byte value (0 - 255) but in the latest graphs you mention 0.1 MPH precision for the speed readings. Are these just coming from different CAN busses?
     
  18. artsci

    artsci Sponsor

    Joined:
    May 10, 2012
    Messages:
    6,190
    Location:
    Timonium, Maryland
    That's the sensor not the camera itself. This was a dead end WhiteP85 and I went down several years ago when he was designing the camera switch PCB. When he did the front camera switch design hw found that there was a way to invert the front camera image on the horizontal plane but it was a software solution that required access to the touchscreen controls.
     
  19. wk057

    wk057 Senior Tinkerer

    Joined:
    Feb 23, 2014
    Messages:
    5,599
    Location:
    Hickory, NC, USA
    Oh, right. I must have forgotten to mention I deciphered a better speed read out.

    The *numeric* speed displayed on the IC is one byte. The speed used by the analog meter (now missing from the autopilot cars on the IC but still in the raw data) is actually 12-bits and is speed_in_mph/10.
     
  20. pgiralt

    pgiralt Supporting Member

    Joined:
    Jun 16, 2013
    Messages:
    1,520
    Location:
    Cary, NC
    ahh - thanks for the clarification.
     

Share This Page

  • About Us

    Formed in 2006, Tesla Motors Club (TMC) was the first independent online Tesla community. Today it remains the largest and most dynamic community of Tesla enthusiasts. Learn more.
  • Do you value your experience at TMC? Consider becoming a Supporting Member of Tesla Motors Club. As a thank you for your contribution, you'll get nearly no ads in the Community and Groups sections. Additional perks are available depending on the level of contribution. Please visit the Account Upgrades page for more details.


    SUPPORT TMC