[Ovmsdev] OVMS TPMS on Tesla Roadster - New Jersey area

[HONDA S-2000]

What about recording only the unknown CAN bus data? Assuming that  
most of the CAN bus messages on the Tesla Roadster are known, those  
would be discarded without using any EEPROM space. I realize that the  
unfortunate truth may be that there is still a lot of unknown data.

Another possibility might be to compress the data, if there is enough  
processing power for that.

Finally, if there is still two much data, a two-stage process could  
be used. First, OVMS firmware could build a histogram of CAN bus  
message data types during a time-span when the TPMS tool is known to  
be sending data over the CAN bus. Then, based on information in the  
histogram, it might be possible to determine which data to save and  
which data to discard. A second pass of OVMS firmware could then be  
fine-tuned to record just the pertinent data. GPRS communications, if  
two-way, could be used to switch between modes and/or select the  
types of data to stream and/or record.

I'm assuming that the TPMS tool sends either 1 complete message for  
the 4 tires, or perhaps 4 messages of 1 message per tire. There might  
be a fixed number of repetitions of the data, just for error  
checking. I sort of doubt that the data is sent continuously,  
although that is not impossible. But it does seem most likely that  
the amount of data that needs to be recorded should be small, if only  
it is possible to determine in advance how to identify the data that  
should be retained.

Brian


On Apr 29, 2013, at 21:56, Mark Webb-Johnson wrote:
> Problem is memory in the PIC18F. It is tiny (a couple of KB of RAM  
> and EEPROM). Nowhere near enough to store a stream of can bus data.
>
> We could conceivably stream it over a GPRS data connection - so  
> long as the connection was stable that might be workable, but I  
> doubt whether it could stream at more than 9600baud or so so we're  
> gonna lose a lot of data.
>
> From the wiring point of view, it is three / four pins, so should  
> be possible to just tap those.
>
> On 30 Apr, 2013, at 12:46 PM, HONDA S-2000 wrote:
>> I am very interested in this bit of hackery, but I am in Seattle.
>>
>> Would it be possible to have the OVMS passively record CAN bus  
>> activity? My plan was to (some day) look at the OVMS source,  
>> figure out a way to turn it into a very short-term CAN bus  
>> recorder, and set it to record for the duration of the TPMS tool  
>> usage. I'm not really sure how feasible that would be, but it  
>> seems possible.
>>
>> I assume that the PVMS might need to be wired in differently than  
>> it currently is, such that the TPMS tool would still be able to be  
>> attached at the same time.
>>
>> I don't even know which CAN bus the TPMS tool attaches to, or  
>> whether the OVMS can access that bus, so there's quite a bit of  
>> hand-waving on my part at this juncture.
>>
>> Brian
>>
>>
>> On Apr 29, 2013, at 21:32, Mark Webb-Johnson wrote:
>>> We've got some users interested in looking at the CAN bus  
>>> activity for the TPMS tool. The work will be in the New Jersey,  
>>> USA, area.
>>>
>>> The users doing the work are experienced with Tesla and OVMS, but  
>>> not developers and don't have any CAN-USB tools.
>>>
>>> Is there anyone in New Jersey with a CAN-USB adaptor, knowledge  
>>> to tap the car bus, and time?