924RACR

OK, cool. Yes, I agree WRT the steering input, and that's why I've been saying again and again that it doesn't seem trustworthy. However it is interesting how that one channel DOES seem to match up on the very last corner (11 and 12, which are Swamp Turn part 1 and 2) and in 3 aka Skeethouse (at least after turn-in.

As mentioned, agreeing with your observations, the car still does have substantial understeer set up in it at the moment; mainly noticeable on turn-in to mid-corner, sometimes can get a little loose putting down the power on corner exit. I intend to raise the rear/lower the front to get the car closer to level to try to improve this for next season, as mentioned.

Specific to the Turn 1 (Gulch turn on the map) - your observations sound correct, as the track drops off at the end of the braking zone, then rises back up hard mid-corner, with a crest at the second apex (it's run as a single late-apex, actually), at the 2nd flag station marked on the map. So your description matches the car response pretty well.

To see the track with elevation changes, it might be helpful to watch a video or two:
http://vaughanscott.com/videos.htm

924RACR

Any further words of wisdom from anyone? Bueller?

SundayDriver

I will add one more thing, but I think the guys trying to learn this stuff should jump in with guesses and questions. This is a good opportunity to learn for Kary and others - this stuff should make you think, guess and learn. I have seeded this with some comments and cbracer has added a lot (for those that do not know him, Chris is an engineer at MoTeC so he really knows his stuff).

I did observe the loss of speed and radius along with steering changes in T1. I was going to add that you could either carrry more speed there, or it was showing a reduction of grip (drop off in track, fluid on track, etc). This is a case where shock sensors and some math would show more of what was going on (of course, in this case it is obvious if you know what the track does, but sometimes we are not even aware of little elevation changes). Shock pots can be run through math channels accounting for spring rates and geometry, yielding downforce - front, reat and center of pressure. Now, this downforce reading does not differentiate between aero downforce and elevation changes. It would show a loss of downforce in this corner which would tell you that there was a drop off in the track. Now I have not attempted the next step, but you could estimate available grip from that downforce then compare cornering speed, g's or corner radius to that.

mikew968

A humble question:

How does one even begin to get to this level of data understanding?

Mike

924RACR

Excellent point I've never thought about - using the shock pots to determine available normal force at each corner! Yeah, you can then really do a lot, including comparing that (and the resultant available longitudinal and lateral forces) vs. what the driver's asking for (based on brake apply, throttle apply, and steering apply)...

I agree, this should be a good opportunity to learn - perhaps I'll post another, slower lap from that race (that was my fastest lap as well as my personal best, 0.7s off the class lap record).

Mark (and Chris, and any other experienced hands) - if you have any more direct comments/observations you wouldn't like to let slip out of the bag - would you consider PM'ing them?

Mike - I'd answer by gathering a lot of data (from your own driving), comparing how the car felt with what you're seeing, comparing faster laps with slower laps, and slowly developing a pattern of behaviour. It's best treated as a scientific experiment. The icing later on is to have another, more experienced driver drove your car and compare where the speed is and isn't. Takes time. It's just like working with tire temps and pressures - you need a decent volume of data before you can start to draw any conclusions.

924RACR

Doh... I just realized that the very last pic I posted in this thread was a different lap, inadvertantly. The first two pics, along with the lap below, are my 1:19.619:



This was the 3rd pic posted, was lap 9, at 1:21.110:


So this should be a nice chance for the rookies to compare. Even better for comparison purposes is the following comparison of the two laps, with only lateral g's and speed (the purple speed trace corresponds to the yellow lat trace; the second lap, which IS the slower, has both in light blue):

924RACR

This might be a better lap to compare for instructional purposes (for reasons that are obvious when I compare the two above); the following trace shows the fast lap (again, in purple/yellow) with the light blue lap only a half a second back, at 1:20.309:

SundayDriver

I agree with Vaughan. I started with a Stack system and began to see what happened with data knowing what happened on the track. I also ran video, which helps connect the data to real world. I think the first eye openers were seeing where I was not flat, and I thought I was. Then I had a small off - just slid off the outside of T7 at Mid Ohio and really did not know why. Looking at data I saw that I kept going faster and faster until I hit the point where I was going too fast for the same throttle application point - unloaded the front which was already at the limit and the car pushed off the track. I did that by overlaying two laps.

Go to seminars. All the data vendors will tend to do 1 hr sessions at trade shows and major races.

Spend time crawling through data and try to see what it tells you.

I hired a data engineer for some data coaching. That helped a lot in understanding. Later I attended the Claude Roulle 3 day seminar. Good stuff. You do, however, need to mix the data with track awareness (do you notice that the car is simply oversteering or that it is only oversteering in mid corner). Some understanding of suspension geometry also helps put it all together.

SundayDriver

I never thought of that available grip/normal force idea until I was typing that message. That is why this stuff is so cool to post - we all learn.

I don't have any more observations without spending some more time and I will share as I think about this. But, I would really like to see others jump in and take some shots at understanding it.

Here is a fun one you can try. Create a channel that smooths the steering - maybe filtered with 2 seconds of data. Then overlay with actual steering. Another math channel to calculate the difference and you have steering aggressiveness.