Wilmington Data Crunch – Part 2: Showcase Turn

The showcase turn was an increasing radius on Day 1, the way most people drove it .  I lost of lot of time in it on my third run. The data can tell us how much.

If you recall from Part 1 we are looking at 20Hz GPS data. I don’t have engine parameters, steering angle, brake pressure… just position, velocity and acceleration and what can be derived from those.  For Part 2 I’ve slightly redefined the segments as compared to Part 1. They look like this, with the start being the lower left green dot:

Full 1 Mile GPS Track In 7 Segments

Full 1 Mile GPS Track In 7 Segments

The cross-mark is in the middle of the showcase turn, which is Segment 5. Here’s the segment time data, with Run 3 set as the baseline:

Segment Times- Part 2

Segment Times- Part 2

The Run 1 S5 delta is -00.40, meaning 4 tenths of a second faster (less time) than in Run 3. What happened to lose so much time the third time through this corner?

Continue reading

Wilmington Data Crunch – Part 1: The Big Sweeper

This is part 1 of a series that will look at basic autocross data crunching, using data from my car at the Wilmington Champ Tour last weekend. First up: the big sweeper on Day 1.

The Mile-Long Wilmington Day 1 Course

The Mile-Long Wilmington Day 1 Course

The picture above is the whole track from start (lower left) to finish as recorded by a Vbox Sport with 20Hz GPS. Two runs are shown, Run 1 in blue and Run 3 in red. Run 3 was only 0.135 seconds faster than Run 1 overall, but that was enough to move me from 2nd to 1st for the day. What I found by looking at the data was that there were huge swings in time between the two runs. In other words, mistakes in Run 1 were being offset by better driving in Run 3, but then a mistake in Run 3 would knock it back down again. (Run 2 was mostly a running disaster. I don’t want to talk about it!)

This post will be fairly technical. That’s just the way it is. I’m convinced that most folks can’t get fast in autocross fast without learning from data. It’s really not that hard. It just takes some effort and the money for the electronics, of course. In my case, being Apple people at home, I had to purchase a PC because the V-box software, even though free, only runs on Windows. (I got the cheapest laptop money could buy.)

Continue reading

B-Street At Wilmington: Most Exciting Class In Autocross?

I’m recovering at home today from an exhausting but exhilarating weekend at the Wilmington Champ Tour where B-Street, with 17 competitors, was a back-and-forth dog-fight that had the racers and the spectators going crazy.

dawn day 2 wilm. champ tour

Dawn on Day 2 at the Wilmington Championship Tour

Dawn on Day 2 broke with yours truly holding onto 1st place by a thread. I’d started  Day 1 Round 1 with a run fast enough for an eight tenths lead. Greg Vincent had coned his first run but came back in the 2nd round to go in front by .087 seconds. Kenny Tsang, Allen Chen, Nathan Young and Michael Bombard were all struggling through the first two rounds with the extremely long and fast course, but that wasn’t to last.

After third runs were complete the top six were separated by less than 0.7 seconds. Allen came back from coning his first two runs to stand 6th with a 74.948, Nathan was in 5th at 74.827, Kenny was 4th at 74.751, Michael, after a cone and then a slowish run to get his bearings, rushed into 3rd with a 74.523, Greg stood on his 2nd run with a 74.319 and I improved slightly to 74.271 to end the day at the top by .048. So, it was Corvette, S2000, S2000, Corvette, S2000, Corvette in the top six.

I need to say a special thank-you to Mark Pilson. Mark had been my main instructor at EVO school and has always taken an interest in my progress. After his runs on Day 1 I asked him for any words of wisdom. I expected a point or two about a key corner. What I got from Mark was a complete, corner by corner evaluation of the course that really helped a lot. Thanks Mark!

The Day 2 course was reversed, with significant changes and about ten seconds shorter. It didn’t flow as well as day 1 and was much tighter with a showcase decreasing radius turn that was throwing a lot of people way off the pace, including me.

After round 1, Allen jumped from 6th to 1st with a sizzling 65.253 for a total of 140.211! Nathan was also in the 65’s, moving from 5th to 2nd at 140.397. Kenny, Mike and Greg were all in the 66’s and stood 3rd (140.478), 4th (140.814) and 5th (140.989), respectively. I had center-punched two cones at the finish and was way down the list. I don’t want to even figure out where I was… too depressing!

Mike went out first for round 2 and slowed down. Then Allen threw down a 64.736, taking over the lead, and I’m thinking “64.7, are you kidding me? I can’t do a 64.7.” Sure enough I only manage a 65.311, which is just enough take the 2-day lead from Allen. Greg then does a 65.043 taking the lead from me. Nathan then goes Allen one better with a 64.622 which moves him to 2nd in the two-day totals. Not to be outdone, Kenny does the fastest time so far with a 64.604 and just like that he is in the lead! It seems like all the fast drivers are doing 64.7 or better except for me and I’m dropping down the order quickly.

Entering the 3rd round, it is now Kenny 1st, Greg 2nd, Nathan 3rd, I’m 4th, Allen is 5th, Paul Kolatorowicz and his Solstice has snuck into 6th and there have been so many lead changes we’ve all lost count.

Allen cones his 3rd run, so he can do no better than 5th. I’ve messed up the showcase turn both runs and know that unless I can fix it, I’m done for. (Scott Hurley had counseled me on how difficult the visuals were, so I really felt like I had been forewarned, but to no avail.) Finally, on the third try, I remember what Scott had told me, do it half-way decent and get my 64, a 64.422 in turns out, which slings me from 4th back to 1st. (You can see the video of that last run here.) Greg goes out and gets his 64, a 64.469, which solidifies him in 2nd, only .095 back of me over two days. Nathan, pushing for all he’s worth, spins right at the finish line, the back of the car comes close but does not cut the timing light, relegating him to 4th. But, what’s this? Nathan is doing a 3-point turn to get back through the finish, the corner workers are eyeing Kenny (carrying 2 cones) as he rounds the last sweeper, getting the red-flag ready, and I’m screaming “Get outta there Nathan… don’t you dare give Kenny a re-run!” Nathan clears the finish, Kenny does not get a re-run and ends up in 3rd place.

Man, was I drained. Top six were Corvette, S2000, Corvette, S2000, Corvette, Solstice, with the top five separated by a total of 1.001 seconds over two days.

Here are the top 5 trophy results:



The weekend proved one thing: there’s nothing exciting like a big class with a several folks all closely bunched at the top, trading the lead back and forth. B-Street may not have the name talent that SSR does right now, but it’s just as exciting, has 55 people registered for Nats, and the class is definitely getting faster.

Deceptive Cones & Late-Apex Revisited

Many course designers will include a sucker cone. Reader DeWitt had some really good comments on an earlier post and submitted a photo of the finish of an autocross he did (last year with ETRSCCA?) that’s a perfect example.

The figure below shows the section leading to the finish. The orange dots represent the approximate position of the cones and the white line is DeWitt’s path from his data. (Remember that everything in the following figures is approximate… don’t get hung up on preciseness.) I added the black lines to show how I think of the “shape” of the course boundaries. I try not to reduce autocross courses to road-courses, but we will cover that in a minute. The inside corner cone conveniently circled in orange near the bottom is the “sucker” cone.

basic course.001

Dewitt, in my estimation, did well to ignore the “sucker” cone. Folks that took the bait drove a line more like the red one in the figure below. By doing that, some got into a problem at the end of the straight line of cones where they had to slow severely to negotiate what turned out to be a severely decreasing radius. Reader Dewitt thinks the red path was slower than the one he took and he is probably right, especially if they took too fast and direct a route along-side the straight line of four cones on the inside of the corner.

tight path.001Now, if you look at the way I drew the red line, ask yourself if the minimum radius is really that much tighter than the white line? It is tighter, but not by that much. Notice also that on the approach the red line is shorter. My guess is that the red line, properly executed, would not have been slower than the white line.

What do I think I would have done, if I had been as smart as I am sitting at this desk, all cloaked in warm hindsight? Something like the figure below, I think, which has two major differences from the white path actually taken by DeWitt.

best path.001

I would have departed from the white line at A, continuing to accelerate longer, taking a more direct and faster route to B. (This is almost, but not quite the same as the red path in the earlier figure. It also depends on there not being other limiting cones, not shown.) Before B I would have braked very hard to get down to the speed necessary to rejoin the white path from B onward. So, I think I get to B in less time and then go around at the same radius and speed thereafter.

This is actually a really good example of open vs. closed course design. Many clubs would have mandated the white path with more cones, forcing everyone to take the same path. By leaving it open, this club encouraged the competitors to think about what they wanted to do and what they thought would be the faster line.

The other difference is that I would have been tight on the cone at C, which means my arc past C is bigger and therefore faster, on my way to an accelerating finish at the same spot as the white path, but at a different angle. This assumes my car accelerates strongly. If my car is very low-power, I might take a tighter arc from C, ending right of center at the finish, cutting the distance to the finish as much as possible.

One other point: reader DeWitt characterizes the white path as a “late-apex” path. I disagreed. This is a good opportunity to explore this question.

To figure this out, I have drawn a more-or-less constant width track through the cones. I then draw in what I think would be a road-race late-apex path through the course, as shown in green.

late apex?.001

If I were doing a track day, faced with the track boundaries shown in black, and wanted to late-apex this corner, I would actually double-apex it, first making an apex at A and then another at B. Beyond A the car is still going very fast, at some point trail-braking into C. To form a very late-apex, one drives a path that goes down to C, which allows an early acceleration point and increases the length of the “straight” beyond B. Now, no one should do this because there is no significant straight beyond B, just another connected curve, so it would waste time to take the path at C. If there were a long straight beyond B, this is exactly the path to take as it will decrease lap time overall.

So, I leave it to you guys to decide whether the white path is a late-apex path or not, or whether both are late-apex, with the green path just very late, having made the apex (the closest spot to the inside of the track) beyond the last cone in the wall of four cones. I call the white path the “momentum maintenance” path. It makes an apex before the last cone of four in the wall. I think it is basically correct except for the revisions I mentioned.

Get Your Transients In Order

Most people rank three qualities in order of importance for an autocross car: 1) peak lateral grip, 2) transient response, and 3) power-to-weight ratio. Let’s talk now about transient response.

What we mean by transient response is how fast a car can change direction, that is, how quickly can turning be initiated or reversed. All forms of automobile racing value transient response, but probably none as highly as autocross. Does any other form of motor-racing regularly negotiate slaloms? The closest are probably the chicanes incorporated into road-race circuits for the purpose of slowing the cars. Here’s some data from my car in a long slalom at TAC/TVR #6 this past Sunday.


The top lin dog data is speed. Notice that it is constant at about 45 mph until I accelerate at the end. (You may also notice that I start accelerating even before the peak lat-g is reached. I “steal” a little cornering to do this, but it may also mean I was under-driving this slalom.) The lower data line is lateral G’s. They alternate plus and minus at about .58 G’s with a little over 1s peak to peak. The slope of the line between the peaks is a measure of how fast the car is transitioning from left to right… how fast it takes to get from max lateral G right to max lateral G left. The steeper the slope, the faster the transition.

What we’d really like would be perfectly vertical lines separating broad plateaus of max lateral G, as shown below.

instant transient.001

How do we get closer to perfectly vertical lines? Also, if we could get to max lateral G faster, we’d get to both a higher maximum value and higher average velocity. As it was, the car did not have time to get to even 0.6 Gs lateral in the slalom before having to reverse, but the car regularly reached 1.1Gs on that lowish-grip surface on that day on longer corners. [Update: I now know that the data device I was using does not capture the lateral-G peaks in a slalom. The car was getting to more than 0.6G, but not all the way to 1.0G. EF 2020]

The first thing to remember is that if you want the car to transition quickly you must ask it. How do you do ask it? You turn the steering wheel fast! Fast hands shorten the transition from a larger radius to a smaller radius, or from turning left to turning right, by quickly establishing useful slip angles at both front tires, thus creating maximum lateral Gs as soon as possible. In theory, a slalom can be taken the fastest with alternating steady-state maximum lateral G turns with instantaneous reversals. The faster you turn the wheel, the more you mimic an instantaneous reversal.

However, the more over-steery the car, and the faster the steering ratio, the more you may have to limit your hand speed. It may also be valid to do like many top drivers: turn the wheel too fast and too far initially, intentionally causing excess rotation, then turn back to catch the oversteer, applying throttle to shift the weight to the rear and increase rear grip.

Besides fast hands, high roll stiffness is necessary to achieve a quick change of direction, which is one of the major advantages of stiff springs in autocross cars where the class rules allow it. How long it takes for the sprung mass to roll to its new attitude directly affects how fast a car can change direction. In classes where you cannot change the springs to increase roll stiffness some cars can use an extra-stiff front anti-sway bar allied with shocks valved to produce high levels of compression damping with a knee at low (1 to 2 in/s) shaft speeds and digress (blow off) at higher velocity. (The digression is necessary to keep the car from becoming unstable over sharp bumps at high speed.) The figure below is a dyno plot of my present shocks, showing average Force vs. Velocity. Notice the knee at 1 in/s in the compression data, which are the top half of the chart. (The bottom half is rebound. I have linear, not digressive behavior in rebound.)


The compression damping resists and slows the compression of the outside shock, thus limiting roll and speeding up weight transfer. This produces higher transient roll stiffness than otherwise. It has no effect on the maximum roll angle reached after the car takes a set because shocks only creating damping force when the shaft is moving. In a  slalom the car may not have time to even reach the full roll angle, especially if the bump damping is high.

Rebound damping also helps turn-in by slowing the extension of the inside shock, thus again resisting roll and speeding weight transfer across the axle. However, the autocross car is limited in the usable level of rebound damping because, in excess, it hurts grip by binding up wheel motion. The effect is that the tire does not stay fully in contact with the pavement. Even on a substantially flat and smooth surface too much rebound can “pull” weight off the contact patch and reduce grip. For one thing, you are rockin’ & rollin’ out there, creating your own pitch and roll dynamics. Every pound of force a shock produces in rebound is obtained from the contact patch of that tire.

Let me repeat that, and put it in italics, because I’m not completely sure you really got it the first time and I haven’t seen it put quite this way before: All rebound forces developed by a shock absorber are achieved by reducing the load at the contact patch of that tire, which in turn reduces the grip available from that tire.

To an extent this is what we want when turning, because that pound of force also acts on the mass of the car to limit it’s upward motion at the shock for good transient response. When rebound damping is used to excessively control body motions we may regret it. For instance, when we try to turn after a ripple in the pavement we may find a momentary decrease in grip has exchanged our normally sharp turn-in for a half second of understeer.

Toe-out in the front can also speed turn-in by quickly creating slip-angle at the inside tire. This pulls the front into the corner even before any weight has shifted to the outside tire.