DamonB

This plot is from an event yesterday. The layout was interesting and a lot of fun to drive. The first turn is a tight 180 degree immediately off the start (bottom right corner), hitting second gear midway down the short chute before making a hard left into a three cone optional slalom. A smooth 180 degree turn to the left brings you down what can be a straightaway if your exit from the previous turn is tight enough. Then hard into another 180 degree turn to the right that has a decreasing radius at the exit. There is room to drive this turn as a constant arc but that was the wrong way to go IMO because it pulls you out farther to the left and gives you more distance to travel along before reaching the next kink and it also doesn't allow as wide an entry into the next long, flowing 90 degree righthand turn which was the fastest part of the track. According to the usage plots on the map I still took it a little too easy through that kink.

What I thought was interesting was the big 180 that is highlighted in green. That area's lateral and accel g's are plotted on the charts at left between the green bars. I drove the first half of the turn with a wide smooth arc and just past halfway I stabbed the brakes for a moment to slow the car and allow me to turn it in tighter for the exit. This set me up better for the next portion of the course and even though it meant slowing here I was able to get on the gas sooner than most other drivers since I finished my turning earlier. The usage plot on the map and the chart on the left shows I was at maximum lateral g the entire time, even when I stabbed the brakes hard for that brief moment in the middle of the turn (The red bar on the charts is the bullseye on the map. Notice the big spike in braking there) I stayed at max cornering g. In order to do this that means when I stabbed the brakes I would have to open up the steering when I braked and then turn in hard immediately after the car slowed. I didn't even realize I was conciously opening the steering up for the instant I braked but in order to plot such numbers I would have had to. If I didn't the front would have pushed out wide since the tires would have no braking grip to offer as they were purely cornering at max grip through that entire corner.

Reviewing the in car video showed that I really did open the steering there, but I don't recall doing it! I guess these things are just becoming innate after enough thinking about it. I was able to back up the last event's scores as well although these were not quite as exceptional, they were very close.

Umrswimr

Excellent plots, Damon. I can't wait to get mine in the mail. I won a Palm M100 for $35 on eBay last night, so there's no going back now!

HedgeHog

D,

Are u logging w/ a Palm or did u strap a laptop in ur car?

Thx.

DamonB

Palm M100; I don't own a laptop. The Palm is so small I just put hook velcro on the back of it and stick it to the passenger side carpet on race days. Velcro sticks to the carpet like crazy.

If using a Palm there is no need to get the newest, fastest most excellentist version because Geez only uses the Palm as the storage device for data from the accelerometers. All you need is memory capacity, there is no need for speed. My M100 is capable of logging 40 continuous minutes I think it is.

Umrswimr

Damon- where did you mount your Geez!? Any specific requirements? I was thinking about mounting it in one of the three rear storage bins.
How big is the cube?

DamonB

The "g-cube" sensor is about a 2.5" cube with a cord coming out of it. If you really want to go nuts and split hairs you would put the cube at the center of gravity of the car but it's honestly not that important. Other than the cube there is a small interface box that the cube and Palm plug into. That box is about 1/2 the size of a pack of cigarettes.

Most people velcro the sensor in a cup holder or to the bottom of a storage bin etc. You can mount it with velcro and take it out easily if you wish. I use mine so often I leave the cube and interface box in the car all the time. I have the cube hidden under the center console and the interface box mounted neatly to the tranny tunnel on the passenger side.

DamonB

Some results from this past weekend. This particular surface I run on a couple times a year and it without a doubt offers the least grip of anywhere I have ever run. The site is nice and smooth but is sealed asphalt and quite slippery, especially if it's hot out. I didn't remember my lesson from last time that since the grip is so low here you really have to soften the car up to make it work. My first run the car felt absolutely terrible, the front would push horribly and the car just would not point into the corners. It was a real dog. We only made three runs that day so I knew right off the bat I needed to change the car a LOT and then be very aggressive on the second run to find out what else I needed to do with the car and my attack. For the second run I took nearly all the shock out of the front of the car and dropped tire pressure 2 pounds at all four corners. This along with drifting the car slightly around the corners resulted in over a 2 second improvement. One run left and I decided against softening the front sway bar as the car was workable and I only had one run left; I felt I could still improve with just tire pressures and didn't want to add the swaybar into the mix and end up with an entirely different beast with only one run to go. For the third run I dropped tire pressure in the front another pound and decided to attack as I did on my second: drive the car hard into the corner and then turn it in hard enough just before mid corner to purposely make the rear end break loose and help rotate the car. The car wasn't sliding tail out, I just carried enough throttle for the rear end to stay a little loose and help rotate the car. Unfortunately on my third and final run I got a little over zealous in one of the first corners and got the car sideways requiring me to back out of the throttle and save the car. I drove the rest of the course better but that mistake made my third run a tenth of a second slower than my second.

DamonB

Here's the first run.

The start is at the far right and leads into a very tight left hander that leads into a right hander that gradually opens up before turning back to the left at the top right corner of the pic. The car pushed so badly that I could not speed it up very aggressively. My usage rating of an 84% for this run is very poor as I expect those numbers to be in the mid 90's or better. All in all a poorly handling car that was poorly driven. The second half of the run was better but I had no confidence in the car at all. Time was a 77.4 I believe.

DamonB

For the second run I softened up the front shocks considerably and dropped all the tire pressures. This course really had no fast transients and it lacked grip so my plan was to just let it mush around. I hate mushy but that was the fast way. Since the front end would now stick I could use the throttle to rotate the rear of the car. On the first run if the wheel was turned and you tried to accelerate the car would just push; couldn't do it. On the second run I steered the car with the throttle in every corner longer than 90 degrees or so and made over a 2 second improvement in time.

I purposely didn't correct the track maps much. If you compare the map of the second run to the first you can see that many of the corners on the second run look "bigger" and they don't show as much of a turn. This is because of the fact I was sliding the rear of the car slightly. Geez only measure acceleration in the x and y axis'; it has no yaw sensor. Because of this when the car is loose the sensor records (rightly so) a decreased angular path compared to the true path traveled. This makes the corners look not as tight on the track map. Again this can be easily corrected but I left it alone as an example of how the map looks uncorrected with the rear a little loose. The first map is about perfect, the second is uncorrected with the rear sliding in long corners.

Umrswimr

Interesting... I can't wait until we both run next month so I can compare the results directly and see where I can improve.

The biggest things I notice when I comapre your graphs to mine: you are a LOT more aggressive with the pedals. Look at how your throttle/brake graph looks compared to mine:

Mine:


Yours:



Granted, your course is a LOT more technical than mine was, but you definately appear to be more aggressive with the gad/brake pedal. I guess that explains the 14% difference in agressiveness rating, huh?

Umrswimr

Damon:
One more thing: how do you have Geez! set up to to compute the max utilization percentages? Mine's set to 5 second sustained which seems to give me a LOT of utilization over 100% which seems a little extreme. I understand that there will be instantaneous spikes above 100%, but I'm getting a LOT of it...

DamonB

I have the sustained setting set to record over .5 seconds. Anything in the sustained column ocurred over half a second or longer. I find that gives me a more representative reading of what the car is doing in sweeping corners.

I've always been very aggressive with the brakes before turn in and always heavily trailbrake as my friction circles show. Geez has taught me some instances where it's better to brake earlier and accelerate earlier rather than maximizing my mph around the corner. This means I spend more time with the throttle down and have the ability to carry more speed through the next section.

DamonB

It's not as bad as you think. The slope of the lines in the graph are what really count as if the graph is over a longer time distance it will automatically look "softer", if it has more manuevers then it will look more aggressive. What you want to key in on is the slope of the line when you're hard on the throttle or the brakes. The closer to horizontal the harder it was performed, the closer to vertical the more gently it was performed.

Umrswimr

Ahh, you're right... It's 0.5s, not 5 seconds.

Even so, I have a LOT of utilization over 100% which seems a little weird.

DamonB

Usually I see that when I'm carrying slight wheelspin out of corners. Not outright tire spinning, but overspeeding the tire compared to actual groundspeed.

Umrswimr

Mine happened on the heavy braking spots, too. Take a look at the start- it's a straight drag launch through 2 gears (you can see the gear change in the middle), then HEAVY braking to slow for the slalom cones. I exceeded 100% on the braking. Notice the green line as well- I'm over 100% in that braking zone, too.
I'm going to try changing the sustained utilization time and see if that corrects things.

Umrswimr

Alright, I dunno if that was a good idea. It lowered the max percentages down to move resonable levels, but it also killed my rating.

Hmmmm....

DamonB

Yesterday was interesting. It went from raining, to drizzling, to misting, to drying, to misting again to drying. Conditions were different for the first few runs and then stayed at mostly dry but cool. Anyways, thought you guys might like to see what it looks like when you blow it big time There was a large carousel turn that I attempted to drive a bit sideways during a run. Spectators said the first 80% of it looked really great but then it got bad fast

First pic is of a "good" run that day. Second pic has some issues. Can you spot the difference?

DamonB

^
I spun the car in the second pic. Basically I drifted the car around the carousel and it got away from me. Car was in a left hand turn and the car's yaw rate kept increasing in the counterclockwise direction as it got more and more sideways. I finally stopped the yaw but the car snapped back the other way faster than I could correct; I then spun it clockwise. It's amusing to watch the in car video because I knew I had "caught" the car and start unwinding the wheel before the car begins to snap the other way. When I knew it was coming I shuffled that steering wheel for all I was worth but had to go from nearly full right lock to full left. It snapped back around faster than I could turn the wheel and I knew I was going for a ride

Geez has no yaw sensor, so the map looks screwy if you don't know how to interperate it. The car enters the left turn and then it suddenly turns very sharply to the left and makes a tightening loop that crosses back over itself. The car didn't really do that but because the sensor saw the car turning left at a certain speed for a certain length of time it thinks the car made a circle, even though it was in fact merely sliding slideways at great speed. There is then a pause in the yaw as I try to catch the car which shows up as the first loop beginning to straighten back out again. Then immediately the map shows an increasing turn to the right when the car snapped the opposite direction. As soon as I hit the opposite steering lock without saving the car I knew I was toast so I locked the brakes and dipped the clutch and waited for it to stop.

The pink path shows the true direction the car went as it was sliding. Since Geez has no yaw sensor all it saw was the car cornering to the left at x g's for a length of time before straightening up and cornering to the right at x g's for a length of time and then coming to a stop. The track map Geez drew is correct in that sense but this shows you what the lack of a yaw sensor will make the interpretation of the track map look like if not corrected for.

WorldPax

Fascinating thread. Looks like I may have to pick up one of these. Should be an invaluable tool in setting up the FB STS2 car- I think these are going to be great in this class and everyone else is looking at Miata's and CRX's. I think Mark Daddio was or is using this software back in 98-99, that was a plot that would make you sick, 1.3left to 1.3 right in less than a 1/4 of a second, this was in the slalom at Nats in a DS Neon ACR.

Barwick

how does it handle off-camber and other corners like them?

DamonB

Somewhat the same issue as yaw. The only thing the off camber corners really throw off are the scores on your performance. Geez grades you by figuring out what the maximum performance of the car was by looking at your max accelerations in every direction for each run. This establishes the performance envelope of the car for that particular run. Afterwards Geez then looks at every point along the run and determines how close to the limit you were throughout the run based on your own "high" numbers. Because of this it is not completely fair to directly compare scores from different tracks.

For instance the site this last weekend is relatively flat but is angled at one end for drainage. There were two different turns on that end of the course and the camber of the ground worked for you in that case, increasing cornering grip. Naturally I could corner harder in those two corners than I could in the corners located on the flat part of the course. Geez says "in this long corner you pulled 1.3 g and in this long corner you pulled 1.27 g and in this other one you only pulled 1.15 g. You weren't using all of you're performance in the 1.15 g corner so I'm going to grade you slightly lower there". Now this is not right or wrong, it's merely what is possible. No datalogger on earth is going to be able to tell you what the coefficient of friction is everywhere on the course along with knowing its inclination or anything else. It has to suppose the surface is regular. This is why you don't put all your self evaluation in the scores when comparing two different tracks as they can be off some points because of things like this. If I had to guess though from all the sites I run on the scores are within a window about 5 points wide.

It is absolutely and completely fair to directly compare my 6 runs that day to eachother because they all ocurred on the exact same course, even though it was raining for some of them. My rain runs were naturally slower in lap time but that doesn't mean my scores were lower. As long as you use all the available grip of the tire in each run, no matter the conditions, you score high.

The first page or two of this thread has more explaining on how exactly Geez works.

Umrswimr

Hehe.... Nice work, Damon! I have yet to loop it on my Geez! logger. It actualy looks quite a bit different that I would have expected. Though, without a yaw sensor, I don't know what I *did* expect it to look like.
See ya in 2 weeks out at TMS- this should be fun!

Umrswimr

Well, here's the results from the event at TMS. I intentionally did not correct the course because Damon and I are going to do a comparison to see how close our two cubes measured.
The course was extremely technical- almost to the point of being annoying. There were three spots on course that they intentionally threw in offset gates in close proximity to control the speeds. Unfortunately, I felt that it screwed up the rhythm of the course and, in the case of the last one before the finish, actually made the course MORE dangerous. By forcing hard, transitional cornering at high speed, you cause instability. Sure, the cars are going slower, but they're also less in control. Overall, I think the course could've been configured for a little more "flow" without causing any safety problems. In fact, it probably would've been safer. As it as, we had two cars end up in the dirt on the finish...
I also broke my golden rule: walk the course until you can do it in your head. Unfortunately, the course was extremely long and we ran out of time. I only got two walks and, even after running it three times, I still couldn't draw it for you. Sightreading a racetrack is NOT the way to go fast. Stupid mistake and I knew better...
Well, on with the analysis:




In a word- I sucked. Badly. Which I knew.
Here's the course with utilization % measured. The first thing you'll notice is the "overlap" which doesn't actually exist. Because the two big 180 degree sweepers were heavily on camber, is screwed up the map. The G measurements are correct, just that the cube doesn't know if you're corning hard or driving the car up a hill.
This also causes one other problem: since the cube compares your intantaneous usage to the max sustained during the run, an "on" camber corner will register very high and become the reference point for max G's. All the other corners are then normalized to this max G force and a Utilization % is assigned. However, we all know that an on camber corner is not a fair comparison to a "flat" corner and, as expected, my utilixation numbers are quite low comparatively. I usually register in the high 80's. This run was a dismal 70%. Blech.
Unfortunately, there's no way to correct this besides deleting the offending heavily cambered corners and re-normalizing the graph. Of course, that also destroys the course and its entire relevance to the race. So try to ignore the total utilization numbers and focus on the spots where I obviously screwed the pooch... Like the entire first half of the course.

Umrswimr

Here's a map that Damon found of the actual course layout. All I did was correct the finish speed- no other adjustments were made to the GEEZ plots.