No update here, it seems to be related to special GPS / OBD update rate constellations and is on the list.
- Harry
RPM data is very "noisy"
-
- 20 or more Posts ★★★
- Posts: 87
- Joined: Fri Mar 13, 2015 9:31 pm
Re: RPM data is very "noisy"
A month ago, I began fiddling with the OBD sensor delay value in the "expert settings"... since like others I was having very jittery/noisy data from the OBDLink MX... this includes throttle, wheel speed and RPM. This was most noticeable when accelerating (i.e. foot down, increasing speed).
I exported lap data to CSV and started messing around in excel. In addition to widely fluctuating values, I also noticed the my OBD wheel speed seemed to be ahead of GPS speed... I could see that while slowing down (decelerating). I created some formulas in attempt to find wheel speeds that matched OBD speeds, and ultimately derived a timing difference of about 0.7 to 0.8 seconds.
I then set the OBD sensor delay to 1.8 instead of the default of 1.0, and I noticed a much smoother, consistent and accurate recording of OBD data (wheel speed, RPM). I've only tested this locally at my business campus, so I can't push too hard (like on track). I've tested on some "straight roads" and I now have near perfect match of OBD and GPS speed. I set cruise control for 60 MPH... and Laptimer data is showing both GPS and OBD at 59.9 to 60.0....
My "guess" is that LapTimer has some algorithm or "OBD data adjustment" where it is trying to match OBD to GPS data. When I got them close via the sensors delay values, the OBD data started looking much smoother. So I don't know if LapTimer is saying something like"Well GPS speed is 50, but OBD say 40, so I'll adjust OBD".
I have a track weekend coming up, where I can really hammer it to find out... and I hope it works, since I want to use OBD wheel speed, since GPS is not all the accurate when going thru a hairpin turn (even with 10Mhz GPS).
Harry, any thoughts?
Regards,
Ed
I exported lap data to CSV and started messing around in excel. In addition to widely fluctuating values, I also noticed the my OBD wheel speed seemed to be ahead of GPS speed... I could see that while slowing down (decelerating). I created some formulas in attempt to find wheel speeds that matched OBD speeds, and ultimately derived a timing difference of about 0.7 to 0.8 seconds.
I then set the OBD sensor delay to 1.8 instead of the default of 1.0, and I noticed a much smoother, consistent and accurate recording of OBD data (wheel speed, RPM). I've only tested this locally at my business campus, so I can't push too hard (like on track). I've tested on some "straight roads" and I now have near perfect match of OBD and GPS speed. I set cruise control for 60 MPH... and Laptimer data is showing both GPS and OBD at 59.9 to 60.0....
My "guess" is that LapTimer has some algorithm or "OBD data adjustment" where it is trying to match OBD to GPS data. When I got them close via the sensors delay values, the OBD data started looking much smoother. So I don't know if LapTimer is saying something like"Well GPS speed is 50, but OBD say 40, so I'll adjust OBD".
I have a track weekend coming up, where I can really hammer it to find out... and I hope it works, since I want to use OBD wheel speed, since GPS is not all the accurate when going thru a hairpin turn (even with 10Mhz GPS).
Harry, any thoughts?
Regards,
Ed
Re: RPM data is very "noisy"
Sorry for the delay in updating the thread. I recently switched phones (to a Droid Razr M with very few apps) and corrected my issue of not getting 10 Hz on QStarz. I now get a better OBD update rate with this new setup as well (around 5 Hz) and I turned off all other PIDs except for RPM (see below for my exclusion list) so I could maximize my data points. The jumpiness does not seem as bad as before but definitely still there. I will upload a video from my latest trackday within the next day so you can see for yourself.
I am not getting OBD wheel speed so I'm not 100% about your theory in the post above. It seems like some interpolation issue to me but I'm sure Harry knows a lot more about it.
I'm currently on HLT v19.0.17. This phone is not connected to the internet so I forget to update regularly
I am exluding the following PIDs:
0101
0105
010B
010D
010F
0110
0111
011C
EDIT: Here is an overlaid video from my last track day
[youtube]http://www.youtube.com/watch?v=R_P07ZdDNFw[/youtube]
I am not getting OBD wheel speed so I'm not 100% about your theory in the post above. It seems like some interpolation issue to me but I'm sure Harry knows a lot more about it.
I'm currently on HLT v19.0.17. This phone is not connected to the internet so I forget to update regularly
I am exluding the following PIDs:
0101
0105
010B
010D
010F
0110
0111
011C
EDIT: Here is an overlaid video from my last track day
[youtube]http://www.youtube.com/watch?v=R_P07ZdDNFw[/youtube]