Robert Henriksen

Amen to that. (on all counts - me, too!)

TD in DC

This is the line of reasoning that led me to start a thread about "Why G-Sum?" I think that G-Sum could be very useful in comparing two drivers on the same track at the same time in the same cars, but I am not entirely certain about its usefulness with respect to a single driver trying to determine how to drive differently (e.g., how close you really are to the limit) based solely on the G-Sum. I see the following problems, among others. First, you are comparing your calculated G-Sum against some theoretical max, which, for the reasons discussed above, just doesn't seem that accurate to me. Second, I can imagine that you could have two different drivers maximizing G-Sums through the same track segment but on different lines, and one line would be faster than the other. Of course, you can look at G-Sum lines to see issues with "non-smoothness."

Robert Henriksen

My name is Color
data charts beat a Playboy
Please pass vaseline

Mark in Baltimore

I completely agree. Great to have you back, Mark.

SundayDriver

You are very welcome. That is the part of Rennlist I like and will try to stay focused on.

SundayDriver

Here is what I do and why...
Who cares about g's? We care about lap times or finishing positions - I will stick to lap times. If I overlay a faster driver's data with mine, I can see that they went deeper into the corner. Or they didn't and actually entered slower but were able to get on the throttle sooner and pick up an extra 2 mph exit speed. Or I can see that they are just cornering 1 mph faster than I am. 1 mph is often a couple of seconds on most tracks. I can see, from the track map and time stamp that they were, in fact, a full second faster through the corner than I was. Time means something.

OTOH, I can do g data and see that they were closer to the max g-sum than I was. But which direction? Maybe they drove at the max and were on a crappy line. They win the g-sum race but I win the time trial. IMO, g-sum is, AT BEST, a poor proxy for time and speed. I have yet to see CC make any claim about what g-sum tells him that you can not also see in speed and distance. The opposite is not true.

All of the above is, of course, my OPINION and I do not claim that to be fact. My opinion is based on a fair amount of experience. I am not the fastest guy out there, but I don't embarass myself, nor make excuses for poor performances. My ideas about data are supported with my experiences.

924RACR

Thanks - this does help put it in perspective... I've just got my racecar closer to, but still below, the 1:1 ratio - as stiff as possible with just t-bars - and definitely not finding it too stiff (though it's roughly 10x stock rates)... the shock adjustment comparo is interesting too - though mine are not, it's all good data, thanks.

(I like the sig too!)

TD in DC

Thanks to all for the responses. Mark, your description of the usefulness of the data is exactly what I have been thinking for a while. I am going to the track tomorrow, and I can't wait . . . ha ha

Skip Wolfe

Forgive my ignorance here, but I have a couple of questions. I understand the basic "gross" tuning, but when you get to the fine tuning of dialing in only the back or the front. What are looking to change with the adjustment to only one end - understeer/oversteer characteristics? If so do you loosen the loose end or tighten the opposite end? Also on a sedan would you be able to just tune using adjustable sway bars?

Sorry again for the rookie questions. I'm looking to enter the wonderful world of double adjustables this winter and want to get a handle on what to do with them.

RedlineMan

Hmmm...

If it were a discussion of G-spot, I'd have a clue... and some "data." As it pertains to cars, I can see the forrest inspite of the trees. It's about driving, folks.

Yes indeed, the stiffer the spring rate - or the more the spring rate approaches the car's mass - the more critical adjustments become. Tiny little adjustments of 1500 spring collars on an 1800lb open-wheeler make huge differences in the corner weights. 100 lbs off a cross on a 2500lb car with 600lb springs? Probably wouldn't notice. 1/16th turn to get hundreds of pounds of change for the former car -vs- 3-4 turns to get 10 on the latter.

My car weighs 25-&-change, running 500F/450R, and with single adjust shocks (bump/rebound together) I can indeed tell the difference in a one click adjustment. It is subtle, but this is most often all it takes to make the car just slightly tighter or looser. I have not touched my sways since I've had the shocks.

Sporting manufacturers have often employed stiff damping and softer springs to get the best of both worlds - firmness when rapid motion is encountered, and a bit of suppleness when things are at a slower pace. GT3's included, apparently.

All that calculating makes my head hurt. One shouldn't forget about driving in there somewhere, eh? That being quite the point.

"When the time comes to close the books and turn the key, theory goes out the tailpipe. The only thing that tells the plain truth is the clock." - Me.

RedlineMan

Skip;

Gross adjustments can be done with sways quite effectively, that being their real purpose on a more stiffly sprung car. If you do not have adjustable damping, you can do it with sways, but it gets a little time consuming sometimes trying to make tiny adjustments searching for minute improvements. Sways can kind of be a bit crude for the fine stuff.

Think in relative terms. You will be looking to effect one end relative to the other. Assuming you have your spring rates pretty well chosen, and damping in the ballpark, you first have to decide what the problem is.

Understeer - The simple explanation is that the front won't stick. It is more appropriate to say that the front is sticking less than the back. In general, if you want an end to stick more than the other, you soften it. That now means you have softened the front relative to the rear, which should give you more front grip.

This is not always the way to go, however. You need to realize that there are many tasks the suspension is called upon to perform, and dilineate them from eachother.

If you did a critical analysis of how the front was working in ALL aspects, you may decide that it was handling the bumps very well, keeping the car dynamically composed, and giving you good brake performance. You might decide that the damping rate was just where you wanted it... for everything but a few corners where it would not turn. To change it for one characteristic - understeer - would mess up two or three others that are fine.

Remembering the "relativity" of balance, you could effect the same end result - more front grip - by stiffening the rear, making it grip less than before, relative to the front. This would likely preserve your braking performance and bump handling, yet also allow you to reduce the understeer.

You could break it down even further if necessary. You could have the understeer in only one direction, say a left-hander. Everything else was fine. To tune out this left turn understeer, you could increase the damping of the right rear shock to stiffen only it, only effecting the amount of rotation you got in left turns without messing up everything else.

Does any of that make sense or answer your questions?

Skip Wolfe

Thanks John. Didn't really think of the sways as gross tuning and the shocks as fine tuning but that make sense.

Nordschleife

I worked my way through (a very good) university teaching Operations Research, Statistical Methods and Applied Mathmatics. I have designed, written and used data analysis systems for variously Grand Prix Yachts, the America's Cup, Le Mans and Supercup. So I have what might be described as an informed opinion.

The seat of the pants is one of the most delicate and sophisticated instruments ever invented. I know of no logging and analysis system that comes within 2 or 3 per cent of it.

In the heat of competition, any system that relies on FFT is probably too complicated.

Off line I use a lot of LP and analyse the Dual very carefully. This gives me hints about where to go looking for improvements.

Cars are easier than yachts - I bet that surprised most listers - air is 'curved', yachts exist and operate in two fluid bodies, this can get difficult.

What yachting teaches most clearly is:

proper calibration is vital.
two boat/car teams are four times as efficient as one boat teams, yet the costs are only 1.5 times as much.
change one thing at a time.
test, test and test.
have a development program that runs through the race campaign.
having the best squad and being better prepared than everybody else is a great way to be.
the best organised team usually wins

R+C

RedlineMan

Very interesting as always, Nordy;

Let's see you convince ColorChange of that!

FFT? LP? Dual? Please explain. I don't want to miss any kernels!

Nordschleife

FFT - Fast Fourier Transformation, commonly used in Digital Signal Processing. From the dark ages of computing, yet stillo relevant. Was a break through technique because it allowed data to be broken out into its constituent elements without overwhelming the limited available processing power of pretty basic computer systems. To get any meaningful answers, you usually have to do a great deal of data scrubbing in the real world, which is costly in terms of development time and sucks MIPS. The technique probably works better in other applications where the environment is better defined and controlled.

LP - Linear Programming - a technique for working on optimisation problems, where there are a number of constraints. The Dual tells you something about the various sensitivities of the constraints and how the optimal solution would change under different conditions.

One of my problems was predicting the time and direction of wind shifts. The algorithm was fairly straight forward, getting clean data to make it work was horrendous.

I have seen a lot of people coming up with unusual theories based upon extensive data analysis. However, time after time the underlying data, or assumptions were faulty.

Claude Rouelle is worth listening to on Data Analysis

R+C