Yuku free message boards

Copied from an earlier post I made:
To gain answers to your question you need a minimum of six sensors running on your car, those sensors are, Eng Rpm, Wheel rpm (or GPS signal) for speed, distance (track mapping), Lateral G and Long G (actual sensor,Not calculated), steering position sensors, throttle position.
Through these six sensors you can gain insight to every movement you make as a driver. All of these sensors should be calculated/calibrated properly and accurately, otherwise you will experience a "garbage in, garbage out" answer. All of these sensors need to be sampled at least 20 samples per second. 10 samples per second is OK, but you will miss a lot of actions/decisions you are making as a driver.
To determine US/OS there are several ways to achieve this. You will need to have the ability to combine graphs (in the same XY plot) such as throttle/long G together and steering and Lateral G. (Some systems don't allow for this and you didn't state what system you have).
You will need both graphs to help determine the handling.
If you have math channels add these two graphs, Speed Adjusted steering= you need Steering angle (can calculated from a steering wheel calibration or from caster plates, the one from the steering wheel is easier to use) the math channel is as follows:
Steering angle X MPH x Square root of MPH.
This will generate a signal that is adjusted for speed... you can pull 1.5 g's in a 40 mph corner and 1.5 g's in 120 mph corner but the steering input will be vastly different between these two corners. This math channel makes the steering signal a constant unrelated to speed.
Ploting this signal combined with Lateral G's shows the handling relationship. (the standard is Lateral G as a postive number when turning Left, negative: turning to the Right.. the same goes for steering.)
O/S is the easier of the two to find. You simply look for the steering correction of the O/S, the steering heads in an opposite direction of the lateral G's.(typically the steering needs to go over center to be considered O/S, not all the time but most of the time) Once you find this, you need to search the events just before the correction, by plotting other availiable sensors, such as throttle/long g plotted together (can be in a seperate XY..if that is possible in your software, ideal would be in the same XY in a different location within the graph) Was there throttle input just before? Off throttle? Braking? Entry...mid...exit? Was the speed higher than laps before?(adding a speed signal to this) You can answer all of these and more through those plots.
U/S is the much harder one to find. Understeer is not the same number between drivers. What you think is understeer may not be understeer to another driver so the analysis of this has to include previous experience of known understeer conditions, according to your perception. If you know a location on the track that you feel an understeering condition, look at the relationship of the SPD adjusted steering to the lateral G, it will decrease it's gap to the lateral G as you approach understeer. Relate that to a corner where the car is working well and the gap there. If the understeer is really bad the steering signal may intersect the lateral or even pass beyond the lateral. The smaller the gap, the greater the instance of Understeer.
The ability to overplot, other laps on top can help determine if this a consistant problem and can show through careful analysis if this is a track induced (keeps happening in the same spot, with different driver inputs, is one way to look at it)
This is a relatively simple way of determining understeer/ oversteer. There are many more...

IF the software has the ability to plot a signal vs. signal, you could put steering on one axis, and lateral on the other. And a second plot with SPD ADJ steering/ lateral. This will show the relationship of G's to steering and show the steering inputs according to the level of G force. An ideal car would pretty much plot all the steering in a straight line,(this never happens!.. at least on a car that's being driven hard).
A third way is to use the inborn handling math that some of the software programs have.
Motec I2, Competition Data Trackmaster, and PI Tool Box have handling channels that calculate handling based on LF, RF turning angles, corner radius, car path, ideal path and so forth. All of these are useful in determining the handling of the car.
The above programs also have the ability to plot data signals onto a track map making locating the event much easier to see. Motec through Rainbow colors, Pi and CDS through actual plotting of the signal(s). (Much easier to see than using rainbow charts, especially when more than one signal is involved or suspension signals).
Each also has a steering wheel graph, with , that is another souce helpful in analysis
PI and CDS have a graphical interface that through the use of suspension signals, will show the movement of the plane of the vehicle via four suspension vs. the plane of the setup table (extremely useful tool in answering the car.. the driver or the track question and suspension/shock/ARB adjustments)
CDS is the only one that I know of that plots the O/S and U/S directly on the map, showing the locations of each making the process of analysis a very rapid experience. In my mind, (I've been a Rep for CDS now for 17 years) CDS has the best graphical interface.

Ted, I have been doing data analysis now for 17 years full time, and I'm still learning new things and ways to look at data, in the data business you never know all. The U/S, O/S question still remains the toughest single answer to extract accurately, because we still can't sense the drivers brain and the actions they take because of their thinking,...or in some cases lack of thinking, LOL.

One more thing. When I got into this business I knew absolutely nothing about computers, data and a limited knowledge about race craft. The team I was with fortunely hired a very experienced consultant for several races. By watching how he analyzed data, greatly increased my knowledge and ability 100 fold (and improved the team/driver the same amount) and set me on the path and put me in the business I am today. That experience can greatly enhance your and teams learning curve.
Hope this helps.

George Main
DataAnalyst/ChassisEngineering/DriverCoaching consultant
speedsense@sbcglobal.net

        New Post Re: Math def. THanks george that did help. I found a good book actually to understand the fundementals of data analysis. If i remember correctly it said that when appraching a corner oversteer happens when lateral acceleration peaks at this point the steering is full lock.

Actually O/S is the easiest to figure out, because the reaction of the steering wheel to Lateral G signal is in the opposite direction, and when the steering reaches 0 and heads in the opposite direction (opposite steering correction or worse case, full opposite lock. Also there is a significant drop in lateral G, as the car slides with the steering reaction right after. It is defining O/S where the driver "catches" the car before steering beyond 0, that becomes difficult, as this can also be defined as a neutral condiition.



Understeer occurs when lateral acceleration peaks, this is then followed by opposite lock of the steering wheel. I hope that made sense.
Looking at my data it looks to be understeer, but Im really not sure if it could be atributed to driver tehinique as well.

Determining U/S with just a lateral G is very difficult except in extreme understeer condiitions, where the Lateral G signals "flat lines" and doesn't gain or lose G force. This is very severe understeer. It is the combination of the two graphs, steering and Lateral G together in the same plot, that will determine understeer. However because the steering amount will vary according to speed, you will need to eliminate the speed variable by this formula: [(Steering Angle)x(MPH)x(Square Root of MPH)] call this signal, spd steer or SpdAdjustedsteering. It is the "relationship" of the signal to the lateral G, that the signal that will help determine understeer. When the car feels very good, stable and fast note the relationship or distance from the signal to the Lat G graph. IF the signal start's to get closer from this observed point, to the lat G Graph the car is moving towards understeer. When this becomes understeer, is pretty much up to your driver, each driver perceives understeer, or the amount of understeer it takes to become understeer for THAT driver. There isn't a number or level in the world that can determine this, as it varies from driver to driver. It is very subjective at best.

Coming back to steering angle, i have a channel called steering pot which measures the movement is degrees. Steering angle measures movement by voltage changes. Does this make a difference when applying the formula you mentioned above?

Aneglo, every channel you have is originally measured in voltage. A data logger is a recording volt meter, and it is also a software calculator, it takes that voltage recorded and turns it into an engineering number that is of some value, such as steering angle, provided you gave it the right numbers to make the change from voltage to engineering units. Kinda like punching the right numbers on the calculator, punch the wrong ones you'll get the wrong answer!!