zachr

I'm working on my data logging setup for autocross (and track) with SoloStorm and a RaceCapture GPS+data logging box. Unfortunately the data provided over OBD is insufficient so this week I'm going to get taps set up on the CAN wires behind the passenger footwell fuse panel. Unfortunately RaceCapture doesn't have predefined CAN mappings for the 992 as far as I can tell (although some data may work with the 991 profile they have).

Is this something I'd be able to find in PIWIS, perhaps? Or if anyone happens to have resources for this info I'd be very appreciative!

ProCoach

Glad to send you to some folks who can sniff and identify or if they have them, sell them to you. It's time consuming, so fair to use a professional. DM me and I'll send their contact info.

Matt Romanowski

Mitch Minton is the man a lot of folks use, but I know he's pretty busy right now. (MoTeC or vBox Joe depending how says it) Joe Hullett https://www.digitalcompetitionsystems.com/staff has done a lot of Porsche work as has Chris Brown http://squigglylines.com/

zachr

Ehh… I’m pretty savvy, I think I’ll give reverse engineering the few things I care about myself before I go that route. Information wants to be free, after all, so if I learn anything I’ll be sure to share here.

ProCoach

Looking forward to it.

If you go to the 992 GT3 Cup service manual, I believe the .dbc information or what you're looking for is listed in a section. Should be the same as the street car.

zachr

Thanks for the tip! For posterity I found this:

Unfortunately it was a dead end. Maybe the GT3 street car populate these channels but my Carrera GTS does not

I did find channels for everything I want, but the format and formulas will take more work. It seems like Porsche doesn't consistently start and end streams along 8-bit boundaries which certainly complicates things. I'll be out of town for the weekend so won't get much chance to experiment until next week at least, but here's what I've got so far:

• throttle_position: The only one I think I got completely. Channel 289, bit 12, 8-bit data. Ranges 0-250 (and weirdly, the last few mm of pedal travel do nothing), so divide by 2.5 to get a nice 0-100% scale.
• rpm: Ran out of time working on this one, but found something that correlates in Channel 168, byte 7: 13 at idle (~900rpm), smoothly increasing to 50 at the ~4000rpm soft rev limiter. This is definitely another case where the actual data length isn't along 8-bit boundaries, but I didn't have time to explore how the adjacent bits behaved. This one shouldn't take long once I get back to it.
• brake_pressure. Channel 262, bytes 2 and 3. Byte 2 rests at 100, byte 3 at 0.
  • Byte 2 increases smoothly with brake pressure, eventually reaching 255 and cycling over as brake pressure continues to increase.
  • Byte 3: as pressure increases it jumps from 0 to 4 instantly when Byte 2 hits 140 (a quirk: as pressure decreases it goes back to 0, but not until 133 &#129300. Aside from the 0-4 jump, byte 3 generally increments by 1 (to 5, then 6 was the highest I ever got with some pretty serious legwork) as byte 2 cycles over from 255 back to 0 and vice versa.
  • This was before I realized the data wasn't necessarily in even 8-bit increments... with that info I might be able to go back and figure out the formula.
• steering_angle: Channel 160, bytes 4 and 5. Both are 0 at dead center. Byte 4 increases when turning in either direction, up to 255 before cycling back to 0 and starting over. Byte 5...
  • Turning left: increases 0-15 linearly, incrementing when byte 4 cycles over from 255 back to 0.
  • Turning right: immediately jumps to 16, then increases linearly up to 31.
  • Fun side note: byte 3 is interesting. It rests at 40 no matter what the current wheel angle is, but while actively turning it moves down as low as 32 (turning left) and as high as 50 (turning right). The further from dead center, the quicker the number jumps relative to how quickly you turn the wheel, but it always falls back to 40 once the wheel is at rest. Maybe something related to power steering? ¯\_(ツ)_/¯

Hopefully someone finds this helpful, and even more hopefully someone has some ideas about the encoding and formulas I should test once I'm back!

zachr

I spent a little more time on this tonight and got data that I'm pretty happy with for almost everything! Photo of the current config attached. Throttle and RPM are both nice and easy.

For brake pressure I want a reasonable 0-100 scale, so I might adjust the divisor. I saw a peak of 630 before setting the formula, but getting to that number was a pretty insane amount of leg force. I think I'll divide by 5.0 instead and allow the brake value to exceed 100% under insane braking, unless when I get out and actually drive the data doesn't match up well enough with my gut feel for brake pressure. I'd rather have the higher fidelity in medium/low pressure situations.

The only thing I still couldn't quite figure out is steering angle. I'm pretty confident in the divisor of 8.72 (anyone know the significance of this number?): that produces 360 when the wheel is one full rotation left; 470 when the wheel is just a hair to the right; ~829 when the wheel is one full rotation right (which, using 470-1 as an offset, is 360°).

Maybe it's just that I'm brain dead, but I can't figure out how to tweak this formula to turn it into a useful value for logging. Setting -469 as an offset makes turning right look reasonable, but then turning left is -468 at 1° left and -29 at full lock left...not useful. Maybe I need to just brute force this with a lua script that applies an offset only when the raw value is positive or something?

zachr

Had a eureka moment last night and finished up steering angle before I started my commute this morning. It's another case where the data isn't an even byte length - from above, I switched to bit mode with offset 32 and length 13, in sign magnitude. Switching to negative 8.72 for the divisor gave me negative numbers to the left and positive to the right, and what i think is a perfect mapping for steering degrees with 880° of total wheel rotation.

Hope someone finds this useful! I'll upload the CAN mapping profile to the RaceCapture library when I get a chance, though I assume no one else here uses their capture box.

ProCoach

Yep, called signed or unsigned messages...

jhenson29

@zachr I find it interesting, if not immediately useful. I find I have plenty to go through without CAN data at the moment. Maybe next year.

zachr

Now for the result!
Data logging worked pretty much perfectly: no more lag/drift over the course of a run, and the update rate is now fast enough to appear accurate all the time.

stellman

Integration with SoloStorm looks great. Care to share where you've installed the RaceCapture and routed everything? Still on the fence about buying one with my 991.1 but the bluetooth OBD2 sensor isn't the fastest thing on the planet, heh.

zachr

Unfortunately I didn't get any install photos when I did the CAN tap, but found this thread from a fellow forum member. I used the same (or at least similar) positap contraptions. Since my car is primarily a daily, I wanted to be able to install and remove the RaceCapture quickly, and have no visible trace when it's not installed. Taking up glovebox space permanently would've been a last resort.

So I trimmed the autosport labs M8 4-pin CAN pigtail cable to about 8" (leaving ground about 18" to reach the screw in the far front of the passenger footwell), and grabbed a 2m M8-4P extension. I can fish the pigtail out of the fuse panel, attach the extension, and run it under the passenger floor mat into my center console where, under the rubbery mat, I've affixed some command strip velcro pieces. I have the same velcro pieces on the bottom of the RaceCapture. The mat sits on top of those command strips as if they aren't even there when I pull out the RaceCapture. As for the GPS... right now I've been suction cupping it to my windshield, but I may move that to the exterior of the car for a better signal -- tbd.

I was feeling a little lazy but here are a few easy photos:

Tiny positap!

Running the pigtail under the passenger floor mat.

Fuse tap to a switched, previously-unused fuse: C12 if I remember right.

Center console with command strip velcro. The RaceCapture fits nicely there when needed, and when not used the rubber mat covers them up without a trace.

zachr

Also, I think I forgot to share the final configuration. All of these are little-endian and only steering angle is signed -- by default left is positive but I preferred the other way around, hence the negative multiplier. I'm not sure what unit the raw brake pressure is in, but I opted for an almost 0-100 scale instead of raw units. Technically I can get this up to ~115 "percent" with immense braking force, but it's good enough for government work.

ShakeNBake

Racelogic has tools and instruction for sniffing codes. The challenge is isolating things within a chatty stream of data. Contact racelogic support, they can help. I used it to find the active areo comms in the 992RS.