Data Logging Metrics

[Belatti]

I should install a camera to actually watch driver steering wheel input... but I dont think he moves the wheel 90° in rather plain and constant radio curves.

If you've got a free or loose racecar I sure wouldn't be surprised to see that kind of steering input.

I've seen MUCH larger steering inputs on big sweeping oval tracks.. about as constant radius as you can get.

Thanks Tom.

Thats a small formula Renault 1.6 with almost no DF.

[speedsense]

Dave, Speed, thanks both!

Check this data:



Where:
Red id Long G
Orange is Lat G
Black is Combined G

Yep! There is room for better braking there. The combined G (in black) begins an upward slope after the decreasing slope of the Long G, in red.

and then:
Green is a calculated steering angle with this formula:

Black is the steering wheel sensor

I think the is something wrong with this sensor. Maybe I should recalibrate it. Cant it be that the o-ring that acts like a belt between the potentiometer and steering column is slipping a bit?




Here I plotted the adjusted steer formula in blue, on top of the Lateral G, in orange:

[Steering]*[Speed]*pow([Speed],0.5) according to Pi software.
What do you think of it?

An indication if the belt was slipping would be if the steering data begins a lap and ends a lap in different places.....
A suggestion on belts (or o-rings) used on steering shafts, use a longer o-ring and double loop the o-ring on the shaft and half twist the end that is between the sensor and the shaft.. A bit of traction tape on the shaft helps as well.
Strings pots work quite a bit better that belts, though more expensive... though if you have an old VCR hanging about, you can sometimes get a cog belt and/or a set of gears out of them, and modify it to fit the steer shaft

What sampling rate are you recording? If it's 10hz, you need to boost that up to 20hz for the lat, long, steering, throttle, speed, Engine R....

Are all the steering corrections during throttle inputs? Would be interested to see how the throttle input is effecting the steering (whether off or on, or combination of both) ...And.. is the steering going beyond zero and in the opposite direction?

[speedsense]

Dave, Speed, thanks both!

Check this data:



Where:
Red id Long G
Orange is Lat G
Black is Combined G

In an "ideal" braking event, the comb G should come out of the peak of the braking. The gap after the peak, generally is considered a coast. What can effect this is braking uphill,downhill as gravity intervenes or a crest in the braking area.
A trail brake will put the peak inside the upward slope of the combG.

In regards to the handling graph, are the last two rights faster that the first one? If they are, and are throttle on situation this car is either aero imbalanced or stiffer in roll in the rear or too soft in roll in the front. By appearance the first right has an understeer condition leading to a correction of O/S, probably due to a lift while understeering and rotating the car into oversteer.
The driver of this car also needs to slow his hands down, as witnessing the entry steering input. The driver has a lot of very small steering inputs which would indicate A) a nervous car B) a nervous driver (at least as far as his hands are concerned)

[Belatti]

I did a couple of tests and a recalibration and now I know for sure the sensor is OK. That cog belt advise is a good one, speed, Ill remember that

What sampling rate are you recording? If it's 10hz, you need to boost that up to 20hz for the lat, long, steering, throttle, speed, Engine R....

Are all the steering corrections during throttle inputs? Would be interested to see how the throttle input is effecting the steering (whether off or on, or combination of both) ...And.. is the steering going beyond zero and in the opposite direction?

I use a 25Hz sampling rate, except in dampers (200Hz) wich is the max I can get and in engine temp and pressures where I use 5Hz.

Here I graph the throttle and brake inputs
(green front brake pressure) (blue rear brake pressure)



Picture yourself an extrapolation line from when steering is in 0 (straight) and youll notice the it goes beyond 0 and to the opposite direction.

[Belatti]

In regards to the handling graph, are the last two rights faster that the first one? If they are, and are throttle on situation this car is either aero imbalanced or stiffer in roll in the rear or too soft in roll in the front. By appearance the first right has an understeer condition leading to a correction of O/S, probably due to a lift while understeering and rotating the car into oversteer.
The driver of this car also needs to slow his hands down, as witnessing the entry steering input. The driver has a lot of very small steering inputs which would indicate A) a nervous car B) a nervous driver (at least as far as his hands are concerned)

The last right is the faster curve (130kmh). The first right and the only left are equally slow (95kmh). Then in the middle there is a medium speed right (115kmh). Im talking about apex speed here. The track is not banked and has no height change. In the first corner entrance there is a big bump from a changing circuit layout.

The first thing I said when I saw and measured the car was that the ARBs are wrong, the rear its too stiff and the front too soft, considering its RWD.
The downforce its not big, arround 70Kg-80Kg at 200Kmh and mostly in the rear.

The thing is, the driver feels very comfortable with this setup. We tried to use a softer rear and stiffer front ARB setup and he complained about understeering everywhere. With the setup we have got now (the one you can see at the graphs) he was able to put 18 laps in a 0.3 secs gap. He made a hat trick, BTW

[Belatti]

Back on topic, Im rather new to the Race/Data Engineer world and till now Im a bit reluctant to Data Logging Metrics, so I have not established any of my own.

I prefer to break down the data in "events" for key parts of the track to analyze them.

For example, a key part of the track would be corner N1.
The events here would be:
1. Straight line braking
2. Braking and steering input begining (corner entry)
3. Steering and throttle begining or coasting (apex)
4. Any steering/throttle/brake disruption (apex)
5. Steering return and full throttle back (corner exit)

Thats no analyze a particular driver and help him out.

To make a fast comparison of drivers of course I´ll have to design my own metrics.

[speedsense]

Back on topic, Im rather new to the Race/Data Engineer world and till now Im a bit reluctant to Data Logging Metrics, so I have not established any of my own.

I prefer to break down the data in "events" for key parts of the track to analyze them.

For example, a key part of the track would be corner N1.
The events here would be:
1. Straight line braking
2. Braking and steering input begining (corner entry)
3. Steering and throttle begining or coasting (apex)
4. Any steering/throttle/brake disruption (apex)
5. Steering return and full throttle back (corner exit)

Thats no analyze a particular driver and help him out.

To make a fast comparison of drivers of course I´ll have to design my own metrics.

While there are many things that make up a driver's fastest lap, there are two things that stand out that if improved the drivers will go even faster. Those are A) time spent at full throttle B) using the steering wheel as little as possible.

To track these... a histogram of throttle involving three or four bins, so that you end up with a bin of throttle >80% to 100%
Steering use... Dev*Steering.... will put the steering wheel into a velocity. The lower the velocity of the steering, the better the car is, especially steering use at entry and when full throttle is struck.

In your driver's case, the majority of the problems at this track are somewhere around half throttle where most of the steering input (velocities) are occurring. Even a slight reduction of this will account for a faster lap speed.

Driver's technique can be ground down into basic types. 1)Those driver's who need the car to rotate for them (tend to like an oversteering car...and.. 2)those driver's who rotate the car around any problems (read understeer) (this driver will tend towards a slight understeer for confidence in fast corners)and utilize his ability to rotate around it.
The direction of setup will differ between these two groups, almost in an opposite direction.

An excellent question for these two groups of drivers is as follows...Which is faster to them (pointed at the driving) through a fast corner..slight oversteer or slight understeer? If the answer is O/S ...ask the following question...
If your going through the fast corner, and you experience O/S, will you attempt to go even faster on the next lap?.... unless he's nuts and has a lot of money for crash damage..most will answer no.
Replace O/S with slight U/S, and ask the same question again? 9 times out of ten, the answer will be yes, they will push it the following lap.
Which condition is more confident inspiring? Which one is actually a faster sub-par handling condition?

Just maybe you may be able to convince the driver to try a different mentality approach to the handling of the car...understeer, can become a friend rather than the FOE that it is now....

Most of the time, if your driver has come from a career in karting, understeer..even the slightest bit.. will be a very bad thing to the driver. They tend to like a bit of O/S. They "believe" it's faster. Especially if the driver is new to cars. Even roll will bother them, as roll in a kart means the end of the kart that's rolling is going to give up. ...Just the opposite "effect" in a car.

One technique I have found to "cure" this "karting" problem, is force the driver to spend a day driving the car with A) varying amounts of U/S B) disconnect the sway bars, both and then just the rear...

Just remember that a driver's ego is fragile when it comes to his driving and your rear end is not in the car. So when describing "problems" that you see in data, DON"T ever suggest or recommend a driving technique change but rather blame the car, ask about the car, not him and leave it on his shoulders to figure out it's him and your life/career in racing data analysis will be a calmer one and hopefully a long lasting one....
want a short career in analysis? ..start teaching your driver how to drive from your laptop.... you will be the first one he'll want to get rid of... [-X

as always, IMHO

[ubrben]

Just remember that a driver's ego is fragile when it comes to his driving and your rear end is not in the car. So when describing "problems" that you see in data, DON"T ever suggest or recommend a driving technique change but rather blame the car, ask about the car, not him and leave it on his shoulders to figure out it's him and your life/career in racing data analysis will be a calmer one and hopefully a long lasting one....
want a short career in analysis? ..start teaching your driver how to drive from your laptop.... you will be the first one he'll want to get rid of... [-X

as always, IMHO

So true.

I think the key here is looking at logging metrics and using it to understand driving style. If you then decide a setup direction to help th particular driver have a better car for his style.

On the other hand, if you have a gentlemen/pro combination the best gents are always open to comparisons and realise they need to follow what the pro is doing to reduce setup compromises as much as possible. Slightly off-topic - but semi-relevant I think.

Ben