How to tell over/under steer in stack software
12-26-2006, 04:48 AM
#1
Rennlist Member
Thread Starter
How to tell over/under steer in stack software
The threqd below on data acq got me really curious but since i didn't want to hi jack the thread I thought I'd ask the question here. How can you tell over or under steer in regular data acq software? i have a Porsche gt3 rs zith sequential gearbox and stack data, including PLT.
Many thanks in advance
JM
Many thanks in advance
JM
12-26-2006, 09:04 AM
#2
Lifetime Rennlist Member
There is a very simple answer to your question, assuming you lack tire slip angle sensors. You can't measure under/oversteer, regardless of whether your system is entry level, mid level or high end. Those terms are defined by relative slip angles and unless you are measuring slip angles, you can not make a correct determination of under/oversteer.
That said, you can get an indication if you run 2 GPS units - one at each end of the car - to see the different lines and then see if you are tail out or pushing.
With steering angle, along with speed & g's, you can make an educated guess with some math channels. What you do is compare the speed corrected steering with inverse corner radius so see if you have more or less steering angle than what you expect for a neutral drive through the corner. After years of using this math, I can tell you it is a reasonable measure of under/oversteer, but not perfect. It rarely tells me anything that I can not already feel. BTW - This 'opinion' is backed by two professional data engineers (with 40+ years of combined experience as full time paid engineers who live and breath this stuff 24/7. If someone want to dispute this, I would sure like to know their qualifications for such a stance.)
The concept that you can measure it by using a GPS trace, vs g's would ONLY make sense if you have one measure at the front of the car and the other at the rear. Even so, it is going to take some serious math to sort that out.
That said, you can get an indication if you run 2 GPS units - one at each end of the car - to see the different lines and then see if you are tail out or pushing.
With steering angle, along with speed & g's, you can make an educated guess with some math channels. What you do is compare the speed corrected steering with inverse corner radius so see if you have more or less steering angle than what you expect for a neutral drive through the corner. After years of using this math, I can tell you it is a reasonable measure of under/oversteer, but not perfect. It rarely tells me anything that I can not already feel. BTW - This 'opinion' is backed by two professional data engineers (with 40+ years of combined experience as full time paid engineers who live and breath this stuff 24/7. If someone want to dispute this, I would sure like to know their qualifications for such a stance.)
The concept that you can measure it by using a GPS trace, vs g's would ONLY make sense if you have one measure at the front of the car and the other at the rear. Even so, it is going to take some serious math to sort that out.
12-26-2006, 11:08 AM
#3
Addict
Rennlist Member
Rennlist Member
OK, enough - what is this with the tire slip angle sensors??? What are you talking? I'm dying to know? We use Correvit S400's all the time to calculate vehicle body slip angle, which can be directly interpreted to give OS/US condition and measurement... and there's another, I think it's the RS200 or something like that, GPS-based and inertial unit... but these are all vehicle-level, not tire-specific, and about $20k each.
Are you measuring tire slip angle directly, and how, or are you calculating it from other measures?
As to the topic, I'd be looking to compare measured yaw rate vs. driver's intent - based on steering angle and speed, Ackerman bicycle model stuff. Unfortunately, I haven't had time to drag my car out to the skidpad and baseline it, so it's easier said than done.
But fundamentally, by definition, oversteer/understeer (OS/US) are the vehicle seeking a smaller/larger radius than the driver wants/is steering for. You can back-calculate corner radii, as indicated, from lateral g's, from steering input and vehicle geometry, from yaw rate... but keep in mind that the measured corner radius and the desired are only ever going to agree when the car is either cornering well below the limit of the tires (0.4g or so, maybe slightly higher for race tires), or if you're fully loaded up and the car's drifting perfectly at the limit. I don't know about you guys, but my car rarely drifts perfectly around every corner on a lap! Usually I'm pushing and cursing pretty hard to get it through those corners at top speed!
As mentioned on the other thread, when you have some pretty blatant OS, it can be pretty obvious by seeing the loss of speed in the trace. However when it's more subtle, or when you've got anywhere from mild to serious US - gonna be hard to find that without looking at speed, steering input, and yaw rate IMO.
Are you measuring tire slip angle directly, and how, or are you calculating it from other measures?
As to the topic, I'd be looking to compare measured yaw rate vs. driver's intent - based on steering angle and speed, Ackerman bicycle model stuff. Unfortunately, I haven't had time to drag my car out to the skidpad and baseline it, so it's easier said than done.
But fundamentally, by definition, oversteer/understeer (OS/US) are the vehicle seeking a smaller/larger radius than the driver wants/is steering for. You can back-calculate corner radii, as indicated, from lateral g's, from steering input and vehicle geometry, from yaw rate... but keep in mind that the measured corner radius and the desired are only ever going to agree when the car is either cornering well below the limit of the tires (0.4g or so, maybe slightly higher for race tires), or if you're fully loaded up and the car's drifting perfectly at the limit. I don't know about you guys, but my car rarely drifts perfectly around every corner on a lap! Usually I'm pushing and cursing pretty hard to get it through those corners at top speed!
As mentioned on the other thread, when you have some pretty blatant OS, it can be pretty obvious by seeing the loss of speed in the trace. However when it's more subtle, or when you've got anywhere from mild to serious US - gonna be hard to find that without looking at speed, steering input, and yaw rate IMO.
12-26-2006, 11:09 AM
#4
Addict
Rennlist Member
Rennlist Member
PS - do you have access to such sensors on the bus in your car - steering angle, speed, yaw rate? Or does it not have all those? Sorry, I'm somewhat less familiar with all the options on the shinier variations of the P-cars... 
12-26-2006, 02:47 PM
#5
Lifetime Rennlist Member
Originally Posted by 924RACR
OK, enough - what is this with the tire slip angle sensors??? What are you talking? I'm dying to know? We use Correvit S400's all the time to calculate vehicle body slip angle, which can be directly interpreted to give OS/US condition and measurement... and there's another, I think it's the RS200 or something like that, GPS-based and inertial unit... but these are all vehicle-level, not tire-specific, and about $20k each.
Are you measuring tire slip angle directly, and how, or are you calculating it from other measures?
As to the topic, I'd be looking to compare measured yaw rate vs. driver's intent - based on steering angle and speed, Ackerman bicycle model stuff. Unfortunately, I haven't had time to drag my car out to the skidpad and baseline it, so it's easier said than done.
But fundamentally, by definition, oversteer/understeer (OS/US) are the vehicle seeking a smaller/larger radius than the driver wants/is steering for. You can back-calculate corner radii, as indicated, from lateral g's, from steering input and vehicle geometry, from yaw rate... but keep in mind that the measured corner radius and the desired are only ever going to agree when the car is either cornering well below the limit of the tires (0.4g or so, maybe slightly higher for race tires), or if you're fully loaded up and the car's drifting perfectly at the limit. I don't know about you guys, but my car rarely drifts perfectly around every corner on a lap! Usually I'm pushing and cursing pretty hard to get it through those corners at top speed!
As mentioned on the other thread, when you have some pretty blatant OS, it can be pretty obvious by seeing the loss of speed in the trace. However when it's more subtle, or when you've got anywhere from mild to serious US - gonna be hard to find that without looking at speed, steering input, and yaw rate IMO.
Are you measuring tire slip angle directly, and how, or are you calculating it from other measures?
As to the topic, I'd be looking to compare measured yaw rate vs. driver's intent - based on steering angle and speed, Ackerman bicycle model stuff. Unfortunately, I haven't had time to drag my car out to the skidpad and baseline it, so it's easier said than done.
But fundamentally, by definition, oversteer/understeer (OS/US) are the vehicle seeking a smaller/larger radius than the driver wants/is steering for. You can back-calculate corner radii, as indicated, from lateral g's, from steering input and vehicle geometry, from yaw rate... but keep in mind that the measured corner radius and the desired are only ever going to agree when the car is either cornering well below the limit of the tires (0.4g or so, maybe slightly higher for race tires), or if you're fully loaded up and the car's drifting perfectly at the limit. I don't know about you guys, but my car rarely drifts perfectly around every corner on a lap! Usually I'm pushing and cursing pretty hard to get it through those corners at top speed!
As mentioned on the other thread, when you have some pretty blatant OS, it can be pretty obvious by seeing the loss of speed in the trace. However when it's more subtle, or when you've got anywhere from mild to serious US - gonna be hard to find that without looking at speed, steering input, and yaw rate IMO.
What you are describing, math wise, seems to be pretty much what is supplied with my data system to calculate OS/US. I agree with all you said about it's limited value. May seat of the pants is that it pretty much agrees with what I can feel. Nice calculation but what is the point of that data if I can already feel it.
12-26-2006, 06:22 PM
#6
Rennlist Member
Thread Starter
Originally Posted by SundayDriver
There is a very simple answer to your question, assuming you lack tire slip angle sensors. You can't measure under/oversteer, regardless of whether your system is entry level, mid level or high end. Those terms are defined by relative slip angles and unless you are measuring slip angles, you can not make a correct determination of under/oversteer.
That said, you can get an indication if you run 2 GPS units - one at each end of the car - to see the different lines and then see if you are tail out or pushing.
With steering angle, along with speed & g's, you can make an educated guess with some math channels. What you do is compare the speed corrected steering with inverse corner radius so see if you have more or less steering angle than what you expect for a neutral drive through the corner. After years of using this math, I can tell you it is a reasonable measure of under/oversteer, but not perfect. It rarely tells me anything that I can not already feel. BTW - This 'opinion' is backed by two professional data engineers (with 40+ years of combined experience as full time paid engineers who live and breath this stuff 24/7. If someone want to dispute this, I would sure like to know their qualifications for such a stance.)
The concept that you can measure it by using a GPS trace, vs g's would ONLY make sense if you have one measure at the front of the car and the other at the rear. Even so, it is going to take some serious math to sort that out.
That said, you can get an indication if you run 2 GPS units - one at each end of the car - to see the different lines and then see if you are tail out or pushing.
With steering angle, along with speed & g's, you can make an educated guess with some math channels. What you do is compare the speed corrected steering with inverse corner radius so see if you have more or less steering angle than what you expect for a neutral drive through the corner. After years of using this math, I can tell you it is a reasonable measure of under/oversteer, but not perfect. It rarely tells me anything that I can not already feel. BTW - This 'opinion' is backed by two professional data engineers (with 40+ years of combined experience as full time paid engineers who live and breath this stuff 24/7. If someone want to dispute this, I would sure like to know their qualifications for such a stance.)
The concept that you can measure it by using a GPS trace, vs g's would ONLY make sense if you have one measure at the front of the car and the other at the rear. Even so, it is going to take some serious math to sort that out.
JM
12-26-2006, 06:42 PM
#7
Addict
Rennlist Member
Rennlist Member
Originally Posted by SundayDriver
These sensors were discussed in the seminar I recently attended. There was not much detail as the pricetag was described as being around $25k each with two needed. Probably the same things you are working with professionally.
What you are describing, math wise, seems to be pretty much what is supplied with my data system to calculate OS/US. I agree with all you said about it's limited value. May seat of the pants is that it pretty much agrees with what I can feel. Nice calculation but what is the point of that data if I can already feel it.
What you are describing, math wise, seems to be pretty much what is supplied with my data system to calculate OS/US. I agree with all you said about it's limited value. May seat of the pants is that it pretty much agrees with what I can feel. Nice calculation but what is the point of that data if I can already feel it.
I'd be very interested in getting your take on the data I posted in the other thread. As mentioned, I think I've got a half-decent vehicle model for OS/US evaluation in there, if maybe not perfect. Certainly all the steer angle and yaw rate data is there for evaluating vehicle response...
Trending Topics
12-27-2006, 01:51 AM
#8
Drifting
I don't have a ton of hands on DA experience, Mark has a lot more and is probably correct in saying that you need the slip angle sensors to directly pinpoint the under/oversteer but I believe there are ways to interpret the lateral g curve and steering sensor to look for an under/oversteer situation.
I'll quote a recent RaceTech article on a GPS based system (1 accelerometer):
Oversteer will generally make the trace (lat g) ragged as the car grips, loses grip with opposite lock correction and grips again. The effects of oversteer on the lateral g curve is to put rounded spikes into the curve at a rate that reflects how quickly the car reacts. A purpose built racecar will react more quickly than a modified production car and the frequency of the spikes will be shorter. If a steering trace has been logged, the symptoms are unmistakable and a trottle trace can oftern tie the oversteer in to the driver's right foot.
.....
Understeer can be more subtle. You might see the lateral g forces fall away as the driver applies the throttle and this could also be tied in with an increase in steering angle. On the other hand understeer is often shown up as a flat area on the g trace - often towards the end of the corner - or diminishing g levels through the turn.
It's hard to believe that all of your g, steering, brake, and throttle data can't spot the symptoms of over/understeer. The $25k slip sensors may be ideal but I think reading the lateral g trace can be helpful as long as you combine it with steering trace, driver feedback, dividing the corner into segments, and other traditional methods.
What do you experienced DAS guys think?
I'll quote a recent RaceTech article on a GPS based system (1 accelerometer):
Oversteer will generally make the trace (lat g) ragged as the car grips, loses grip with opposite lock correction and grips again. The effects of oversteer on the lateral g curve is to put rounded spikes into the curve at a rate that reflects how quickly the car reacts. A purpose built racecar will react more quickly than a modified production car and the frequency of the spikes will be shorter. If a steering trace has been logged, the symptoms are unmistakable and a trottle trace can oftern tie the oversteer in to the driver's right foot.
.....
Understeer can be more subtle. You might see the lateral g forces fall away as the driver applies the throttle and this could also be tied in with an increase in steering angle. On the other hand understeer is often shown up as a flat area on the g trace - often towards the end of the corner - or diminishing g levels through the turn.
It's hard to believe that all of your g, steering, brake, and throttle data can't spot the symptoms of over/understeer. The $25k slip sensors may be ideal but I think reading the lateral g trace can be helpful as long as you combine it with steering trace, driver feedback, dividing the corner into segments, and other traditional methods.
What do you experienced DAS guys think?
12-27-2006, 03:42 AM
#9
Intermediate
Join Date: Aug 2003
Location: Corona Del Mar, CA
Posts: 49
Likes: 0
Received 0 Likes
on
0 Posts
"Oversteer will generally make the trace (lat g) ragged as the car grips, loses grip with opposite lock correction and grips again. The effects of oversteer on the lateral g curve is to put rounded spikes into the curve...."
If this is based off a GPS calculated lateral G, and the driver can catch a car that oversteers that number of spikes, then the quote would be correct and you would have one truely amazing driver.
The rounded spiking that many see in lateral G is a function of the tire gripping and ungripping as a grains across the pavement. (5Hz to 10Hz) (20Hz+ rate is just chassis vibration) It does not represent a car that understeers or oversteers. The biggest advantage to calculating oversteer/understeer is to help the novice driver identify understeer. The second quote also has a few problems..... the lateral G will always decrease as the driver applies throttle since he is driving out of a corner on the maximum envelope of the traction circle. There's no easy answer. To quote my favorite stock market analyst, "you must do your research" on all the data that you can collect.
If this is based off a GPS calculated lateral G, and the driver can catch a car that oversteers that number of spikes, then the quote would be correct and you would have one truely amazing driver.
The rounded spiking that many see in lateral G is a function of the tire gripping and ungripping as a grains across the pavement. (5Hz to 10Hz) (20Hz+ rate is just chassis vibration) It does not represent a car that understeers or oversteers. The biggest advantage to calculating oversteer/understeer is to help the novice driver identify understeer. The second quote also has a few problems..... the lateral G will always decrease as the driver applies throttle since he is driving out of a corner on the maximum envelope of the traction circle. There's no easy answer. To quote my favorite stock market analyst, "you must do your research" on all the data that you can collect.
12-27-2006, 11:01 AM
#10
Addict
Rennlist Member
Rennlist Member
Compare with my data trace in the other thread, look at the steering angle and how yaw rate responds to it. From what I've seen, an understeer condition is pretty obvious in that the yaw rate stops increasing and in fact usually falls off once the driver's saturated the front tires and starts plowing. Equally, you can see when the understeer stops by a small spike in the yaw rate as the tires return to their peak slip angle and then fall below (as the driver steers out of the corner).
Of course, that's when looking at a flat corner; with banking or elevation changes, you may expect and drive through some degree of understeer, and so training the driver to fix that understeer may not improve lap times.
Sorry, I haven't been able to draw such definitive conclusions from lateral g traces...
Of course, that's when looking at a flat corner; with banking or elevation changes, you may expect and drive through some degree of understeer, and so training the driver to fix that understeer may not improve lap times.
Sorry, I haven't been able to draw such definitive conclusions from lateral g traces...
