DamonB

reza, go to the 16th post in this thread. It's an email from the creator of Geez explaining this to me.

reza

hmmm....so if in one turn you have high G, then next turn does not have as high G, that means you are not pushing it right?

How does it get this max if based on the maximum given by your run.
I think the biggest G would be just before it lost all traction then spin. Should you test the limit first that way you know what the max is?

DamonB

First off "usage" means tire usage. The Friction Circle of a tire says we can get 100% accel, 100% brake or 100% cornering, but in order to do two of those at once we must trade one for the other. If we want to use 75% of the tire's grip in cornering we only have 25% left to accelerate or brake for instance. If we try to do more than that the tire will loose grip and slide.

The driver's goal is to always use 100% of the tires' grip. If he is not then the car will not go around the track as fast as it is capable of.

Geez determines tire usage just as Byron said. The software will look for the highest peak G value recorded on a given lap; this value will change every single lap. It will then compare the entire lap to that peak value and determine how close you stayed to peak at every location on the track.

As a for instance my sustained left and right G's are both 1.21 (That's a very good thing. Means the car is "straight" and I'm driving with equal vigor in each direction). So for that run Geez knows that given that surface and these tires I am capable of doing 1.21 g's. The software then just looks at the entire run and everywhere I am doing less than 1.21 g's it determines I am not getting maximum performance out of the grip I have available. In a long sweeping corner I must be cornering at 1.21 g's in order to score high. On a straightaway I must be accelerating at the verge of wheelspin in order to score high. Under braking I must be on the verge of tire lockup to score high. Trailbraking into a corner I must be balancing my steering input perfectly with my braking in order to score high etc.

Even the most expensive Pi Research or Stack systems work this same way. No system is capable of reaching out, touching the track surface and measuring the coefficient of friction of every single pebble you are to ride over. The software has to assume that at least for some instant on that lap you were at the limit of the tire. Maybe it was in a corner. Maybe it was accelerating just before spinning the tires. Maybe it was braking just before locking up the tires. Maybe it was storming into a turn and perfectly trading your braking for cornering. That instant then determines for the software what the tires can do on that surface.

This is not fool proof. Even the most expensive data system on an F1 car cannot tell you that the reason the car generated half as much G in a given corner and yet slid off the track was because the car in front blew an engine and oiled the track. As anyone will tell you the key is not getting the data, it's analyzing it. Analysis requires some human interpretation.

If you look at my map between 45 and 50 seconds there is some red in it again showing less than 70% usage. This section does not concern me much because that section was dirty all day, I HAD to go slower there. Geez doesn't know that one spot on the course tended to be gravely so it scored me lower. That's why you need the input from the driver: Me. I know what was on the course and I do not use the data as gospel, I use it as an aid. The numbers are not false, but only the driver can truly understand the condition of the track and therefore his point of view is very important in interpreting the data correctly.

maxpesce

And from this we learn that GOOD BRAKES = FAST LAPS

DamonB

One of my Sunday runs from the SCCA Houston NT.





One of my gf's runs from the same day.

Umrswimr

She's a lot smoother than you are.

Looks like she was a lot closer to he "theoretical" limit throughout the course (judging by the red)... How close were her times to yours (since we all know she didn't beat you in YOUR car, did she?)

reza

I think red means you are not using the full potential of the available grip.
light blue/blue is over 90% usage.

DamonB

Yeah, reza is right. I have both maps drawn the same:

90%+= light blue
80%-89%= dark blue
70%-79%=violet (pink)
anything under 70%= red

You want to see lots of light blue and very little or no red. FWIW you can chart anything you want on the maps, I just use these colors and these scales. You can make the colors and what the represent on the map whatever the hell you feel like.

Remember the colors have nothing to do with speed, they represent how much of the available tire grip was being used. Light blue does not mean the car was traveling faster than in the violet section, it just means the car was closer to the limit of adhesion.

On Day 2 I ran a 52.7 to her 57.1
On Day 1 she had to run in the wet and I didn't so her times were much slower than mine. She drove the best I have ever seen her and is beginning to realize how cornering the car hard isn't enough, you have to be very aggressive with throttle and brake to go fast between all the corners. The NT courses can be intimidating if you're not used to big, fast courses.

Umrswimr

Ahhh, that explains it... I thought the color scheme was the OTHER way around.

reza

Damon,

Do you think the Geez Software would work with G-tech data?

Reza

DamonB

Absolutely not; they are two different applications. It would be like asking Corel Draw to read AutoCAD files for instance. Their native language is completely different.

reza

Can Geez use data from CSV format?
e.g. I could get export from Gtech in Excel, then format it for Geez, and let Geez do its analysis...

DamonB

No. Geez is a self contained system and is only designed to work with itself. The accelerometers are proprietary, the calibration of them is proprietary, the Geez applications loaded to the Palm to allow the recording and processing of data are proprietary and finally the download and analysis into the Geez desktop software is proprietary.

Byron Short invented and wrote Geez; his background is in software. I'm sure someone could write an ap that could translate anything to anything but I don't see the point in using G-Tech data in the Geez software. The entire Geez package is only $400. If I ever upgrade from Geez I will go this route .

reza

yea...i was trying to be cost effective here since g-tech is only $250 retail, and it does have two accelerometers. and measure both the forward and lateral accelerations.

InsaneGideon

Looks like a nice setup there, Damon!

I was wondering, of the pics you've shown us, why don't the laps connect continuously?

Are you displaying only a portion of the lap, or is it due to error accumulation from the double integration used to calculate distance & yaw angle?

Just a minor curiosity.

DamonB

Everything posted here so far are from autox runs so the start and finish are in fact two different locations. I have not made it out to the roadcourse yet to do any datalogging but Geez! is capable of recording multilap sessions as well. On a roadcourse the start/finish would connect or at least be very close.

InsaneGideon

I shoulda known from the cones in your sig!

If they do connect, that's a real testament to the accuracy of the system. Interesting.

DamonB

I wouldn't expect them to connect; just be close. The system is very accurate but it is "ignorant" in some regards. The system only measures accelerations in the two horizontal axis along with time. Integrating these can give you speed, direction and distance. The problem is the system doesn't know the orientation of the car; only the sensor and this throws off the track map. The system cannot differentiate between going downhill and the car pitching under braking for instance. It cannot know the difference between the car rolling during cornering or being in an off camber turn. It cannot know the difference between a slow in acceleration caused by the driver or a slow in acceleration because the car is climbing a hill. It cannot understand the car is yawing at 15 degrees with the tail out in a big power slide around a corner, it only knows the path the sensor made through space. These problems do not bring errors into the recorded data but they do bring errors into the representation of the track map, therefore there are simple adjustments that can be made to the map to correct for these.

More advanced systems use a 3 axis accelerometer, yaw sensor, wheel speed sensors, GPS etc to be able to correct for all these errors.

Umrswimr

And cost a TON more.

DamonB

Yesterday was a good day. Nice concrete surface and an interesting course. After walking I expected it to be a little tight and that it would be overly painful to me due to boost lag. I just tried to be as aggressive as I could and the car felt very well. Yesterday produced the best results I have yet acheived; I am used to seeing usage grades in the mid 90's but yesterday I pegged the scale in damn near everything on all 6 runs

Results and scores from 6 runs:


Can I ever repeat that? That will be the challange.

Here is a map showing the course:

Umrswimr

Ok, WOW... That's impressive. You're getting pretty damn good with that car of yours.
I didn't have such great luck today in a less compatitive field:
http://forums.corvetteforum.com/showthread.php?t=865160
I just can't seem to do what you're doing: be consistent. How much of your improvement do you think is a result of Geez!? Basically, do you think you could have done that without the added help of Geez!? You put down a nearly perfect run the first time through- how the hell'd you do THAT?

And turn your PM's on... They're off by default.

DamonB

Geez gives me feedback. Right or not my feelings have been I know how to drive and control the car, I just didn't feel I always drove the course in a manner that would yield the shortest time. I know how to control the car but I didn't always get the strategy right in other words. I feel I am capable of adapting and if am told "Do it this way" I can do it, but you have to know what to try. Geez lets me try a couple different approaches or plans and then compare them to eachother so I can get some answers to questions like what kinds of tradeoffs to make for certain portions of the course. After using it for quite a few events and studying it I am coming up with new thigns to try. Once I try them I can look at Geez results from the next event and see if they worked or not.


I don't see any data from Geez until I go home at night and download it. During the day it doesn't help me; all critiquing is after the fact. That said I have noticed some things I tend to not do as well as I could and so I have worked on fixing those problems. Geez is sort of lke having a coach who watches your performance and then talks to you after the game. During the game he can't help you (unless you have a laptop and download you runs one at a time right then and there. Then you can see right away what you are doing and prepare for the next run or session. I don't have a laptop so I don't see anything until I get home).


I've done about 60-70 events over the past few years and made hundreds and hundreds of runs. I've walked a lot of courses, tried a lot of adjustments, reviewed a lot of in car video and become familier with the idiosyncricies of the different sites and weather. Experience helps.

Remember that high scores don't necessarily mean fast laps. For instance I could drive the course "at the limit" but take an improper line that therefore covers more distance and leave me with a slower time than one that wasn't as "perfect" but covered less distance and therefore had a better time. I think these high scores are somewhat of a fluke. I'm used to scoring in the mid 90's but when you really dive into it and figure out how Geez computes the scores this particular course was a little different than usual. The majority of the turns were at about the same speeds and so by constantly maintaining smooth cornering loads and quick transitional manuevers in between I spent a great deal of time with the car at the limit as opposed to a course with some long straight out of a very tight corner were my boost lag would show on the chart as being traction limited and thus score me lower (even though I can't help the boost lag). My runs tend to be pretty consistent but this particular weekend was exceptionally so. 6 runs were all within tenths of seconds; only .3 seconds seperated my fastest and slowest runs. I just drove very consistently and I have found that just sort of happens by itself once you have the course, car and attack plan figured out. There are plenty of other times (usually large, fast courses) where I try something different that may help or hurt; you just try it and see. These results didn't involve that as I was certain how to attack and I just kept trying. I feel I drove a little better later in the day but as the course heated up the grip fell off slightly.

I run again at the same site next weekend. We'll see what happens.

Umrswimr

Thanks for the update, Damon... I wasn't sure how much constructive feedback Geez! gave you in actually improving your times. Like you mentioned- it lets you try different lines, analyze them, and improve that in the event that you see a similar corner in the future. Sounds exceptionally helpful, provided you understand the limitations (turbo lag, higher # != fastest time, etc)

DamonB

You can go to www.extremegeez.com and download the demo version for free. Everything in the software is functional with the demo version except for the ability to download runs into it; you have to use sample runs that are provided. Other than that it's fully functional and will show you what you can do.

Jacqueline doesn't really understand all the vehicle dynamics or what the datalogger can really do but she claims it still helps her because just by looking at the track map with usage plotted on it she can see where she did "well" and "not so well". She can compare that to what she felt in the car and by watching for those same "poor" situations the next time she can do better.

Umrswimr

Wooo-hoo! Damon, thanks for the link earlier- I'm the proud owner of a semi-used GEEZ! now!

I'm gonna have a LOT of questions for you! Better get ready for a few hundred PM's!