- Apr 18, 2006
- 3,676
- 170
- 1,663
[Warning, this is long, you can just skip to the end part to get the point of all this.]
Note: G = acceleration.. 1G = 9.8m/sec. but in automotive specific use, 1G also means that the car is holding it's own weight sideways through the turn. So 1G means the car is pushing it's own weight sideways. I might be using g and acceleration interchangably, because they're actually the same thing, but different units of measure so hope that clears up any misunderstandings. lateral g means forces acting from the sides.
One thing arguing with Laptimer was making me check my facts. But the way he kept insisting he was right was epically funny, and I enjoyed myself for a while. To give credit to him, he did know some facts, or rather he googled for everything that suited his arguments.. and there came a point where he fabricated some to suit his arguments, so I took that as a cue to exit.
And as I went on reading about these GPS datalogging systems, I find that my initial argument was right, that GPS can't measure lateral G, but he was going on about how 'accurate' his GPS was, and how all measurements are derived and even threw speed as an example, and he was all right, and I was all wrong.
His comment about testing on skidpad confirmed my suspicions that he was a keyboard warrior, and likely doesn't even own a car. Which was why I gave up at that point. For those not in the know, a "skidpad" is a NOT specific place. A big enough parking lot/area will do, just have to make sure the surface is tar/asphalt and not something smooth like cement flooring. This affects grip and you know how much grip you have in those cement floorings like in some parking lots. It has to be level to minimize as much errors possible due to uneven surface. This is how those lateral g is measured by magazines to make it as close as possible for a general comparison. They don't run it at lab standards because it's difficult to replicate every single factor that affects grip like tyre compound, temps and changing surface friction.
The funniest part was him saying that g-sensors/accelerometers are have worse accuracy compared to GPS distance/time measurements, and he can actually replicate the measurements on a skidpad. ROFL!! Baited and caught. Lateral Gs on a skidpad tends to come up low, and he says he can make constant high g's.. Bro laptimer, you must drive a supercar! Yet you didn't win the TTA.. too bad...
Skidpad method
This is the most used method by car magazines you find that lists lateral gs in their car tests, and also considered the most accurate way to measure lateral G.
they use the centripetal acceleration formula which is
centripetal acceleration = velocity^2/radius
Lateral G = (velocity^2/radius)/ gravity constant (gravity is 9.8m/s for those who didn't remember secondary school physics class, though as far as I can remember, they always asked us to use 10m/s as a g constant back then. Amazing that I can remember it after 15 years leaving school lol.
What usually backyard enthusiasts do (like me) do is to draw the circle down with chalk/paint/cones/whatever, and drive around it while tracking the line, going faster and faster each round (each round must maintain constant speed, so if would be 20km/h, then 30km/h and slowly going faster each round) until the tire slips and cannot maintain driving on the line/near the cones anymore. The fastest speed going around the circle before the tires totally slips and cannot track the circle anymore, should be your fastest time, if not, it means you didn't maintain a constant speed going around the circle, or you didn't track the line properly.
My lame car can do 45kmh in a small 20m radius circle before the I can't track the circle anymore due to tires slipping.. velocity is m/s, so 45000m/3600=12.5m/s
Sorry I won't post videos of this, because someone might recognise the idiot who drove around in circles in an empty parking lot near his/her apartment/home one quiet afternoon long time ago. :P
(12.5*12.5/20)/9.8= 0.7971938775510204 Lateral g. (remember that this is a small circle.)
You guys should try this.. Your car might feel like a handling machine, but when you try this, most of you had better prepare to be disappointed.. I don't think you can get very high readings using this method unless you drive a supercar.. I think one magazine posted most sports compact cars with semi slicks having lateral G of less than 1G.... :P.
Laptimer pooh-poohed the 3m or more CEP (circular error probablility) of the GPS, but let's see what happens when I increase the radius to 23m
(12.5x12.5/23)/9.8= 0.6932120674356699 lateral g... and that's just with the minimum error probablility.
The Accelerometer method.
Accelerometers are the like a ruler. A ruler measures length, while an accelerometer measures acceleration. (the term for what accelerometers measure is 'proper acceleration' and this is measured in g-force units. 1G = 9.8m/s.)
It's only considered 'accurate' lateral g if the sensor is perfectly level when measuring, or if the sensor corrects for tilt.
In cars, one thing that affects measurement of acceleration is tilt, this is because two axis accelerometers are affected by tilt. You can see how this works as I tilt my two axis accelerometer to get the readings. Then again, I never said it was 100% accurate, and my initial comment of "can't trust the g-sensor when there's body roll" holds. This is true for anyone's RSM unit with g-sensor as much as it holds true for mine, though more accurately, I should have said, "can't trust it to measure lateral g when there's body roll" Does that mean it's useless? Not really, because we are using the g-sensor as a gauge of how much we pushed our cars through the turns, and we can use it to gauge the amount of body roll.
Does that mean that it can't measure lateral g's? Not fully accurately, but for automotive purposes where comparison is done by per run basis on the same car, it's good enough. The body roll can be considered negligible, depending on where it's mounted. That's why most g sensors tend to end up in the exact center of the car, where roll is most minimal. And one benefit of accelerometers vs GPS is that it knows where the front of the car is at all times.
Then what does it measure? It measures g-forces and acceleration. and direcly, taking into account of road camber, elevation changes, and direction of car. So a car can pull higher g's into a turn if the car rolls enough even with lower grip. I mention the reading's off because I think that the body roll(maybe due to road camber/suspension setting is making the g-sensor/accelerometer reading high even on the slow turns. Laptimer came in and exclaimed there is no error and how wrong I was because the car was using GPS to calculate the lateral g.. He loves saying that everyone is wrong/inferior to his superior intellect. LOL.
That's why I also said only the driver and pit crew knows how useful the data is. As they can measure the previous g-force and current one based on road conditions and car body roll to decide what changes they need to do on the car. The rest is off topic as it became an argument on how accurate each method of measurement is.
Some articles I read even accept the reading of the accelerometer with the tilt/vehicle roll (on level ground) as valid lateral g readings, the reason being that the suspension is already absorbing some of the lateral g's that would have acted on the car. Some articles say no, because a high angle of bodyroll (soft suspension) affects the measurement, so no 100% conclusion. But do remember that tilt plays havoc with g-sensor readings, so even though accelerometer can measure lateral g's, HOW you measure it is also important. This page from the Auxbox manual (add later) obviously shows that if you mount it wrong, the reading will be wrong.. if it's wrong, during calibration it will show E for error, and although you can still log with it in that condition, forget saying it's 100% accurate.
The GPS method.
(I don't know much about GPS 'measuring' lateral g's so I had to read a bit, and my initial assumptions are right, due to one simple fact that I didn't notice till I wrote this.. the position of the front end of the car, sometimes I miss the simplest things)
Laptimer was ranting so much about speed on GPS more accurate than tyre measurements..but we aren't measuring speeds, we're measuring acceleration. :)
The GPS method uses the skidpad formula, but slightly modified since acceleration is not constant unlike in skidpad test. It uses position vs time. Where got need speed bro laptimer...
It measures the distance/time to get the acceleration, then calculates the radius of the turn.
but is only as accurate as the GPS's accuracy, and the data on hand.
So how GPS calculates lateral G data? I tracked like usual way track cars take a turn. Out-in-out. It's called an estimation because mainly because it estimates a lot of things. A GPS doesn't know if you drifted sideways through the turn or you spun out while taking the turn. For gps users out there, you know what I mean.. reversing out of your house also it shows your car driving forward... LOL.
They call it lateral g because it takes two or more points and extrapolates it to be a part of a large circle, and uses the centripital acceleration equation to calculate the lateral G. (this is the cheaper GPS way of calculating lateral g's. Marketing speak on some of the brochures these GPS systems: Accurately measure lateral g's.... ROFL!
first X to third X = distance, 28m time from first X to third X(substitute whatever X position you want), but limit it to 1sec elapsed time, which makes the acceleration to be 28m/s. Position of first and third X has a radius of 60m. One thing I find strange is because Laptimer keep referring to speed, yes, skidpad formula can use speed & time, but that's because speed is constantly maintained. For GPS systems, why bother? can calculate distance based on two or more geolocation points, time elapsed and immediately get acceleration data.
So using the centripetal equation divide by g constant, (Velocity^2/radius)/9.8
(28^2/60m/9.8)=1.3g
Another method measures it versus the turn radius of the track (dotted lines in red). This is the more expensive GPS datalogging systems which has track mapping data. It tracks the acceleration of the car vs. the track's turn radius, not radius between two or more points. Accuracy is higher than the earlier method, but ultimately depending on how accurate the positioning data is. Without ground stations, our geolocation data is not fine enough for more accurate positioning data of the car, so that's why I said it's not accurate.
Summary
So, to make Laptimer satisfied, because he said my comments were making people think that GPS cannot read/measure lateral G at all. Was I wrong? You be the judge.
Sure it can measure Gs, because G = acceleration, but it can't figure out the direction, it just assumes the direction based on positioning data.
So, 3 sentences for those who want to know whether who is right or wrong....
Statement 1) GPS can be used to estimate lateral G an object through a turn, not lateral G of the car. It cannot be used to measure the lateral G of the car due to statement no 2. I bet google didn't tell bro laptimer that. :P
Statement 2) GPS assumes that the car is 'facing' forward at all times, or the side of the car always faces the side of the corner.
(no drifting or going sideways :P) < simplest explanation that I missed. LOL.
Statement 3) GPS gives positioning data, not directional data. Directional data is assumed by comparing current location from previous location via tracklogs. These readings are not real time, they are processed only when the tracklogs are downloaded to a computer.
That's why the article he quoted said it was an estimation. Somehow an estimation became an object of high accuracy, because the some of the product brochures he mentioned used the word 'measure' lateral Gs, plus the word 'accurately'.
Counter argument is that there can be two GPS receivers recording the location of the front vs the rear of the car. But is our Malaysia GPS geolocation data that accurate without ground stations? And who in the right mind in Malaysia runs two GPS receivers on one car? Raise their hands please... :P That particular about position of your car ah?
Oh, if it so happens that someone thinks it is accurate, here's a simple test.. lock coordinates of the front end and rear end of your car.. then measure distance between two points.. If you can get it accurate, you must have some super GPS receiver, and can sell that to Malaysian military for use.. hehehe.. You can see how this error makes it easy for gps to overestimate or underestimate the lateral G.
Heck, I experienced this a lot tagging POIs, and although I'm standing right in front of the shop when tagging POIs, checking back on shows the shop as being across the street. Seems like sometimes GPS signal bounces off buildings and gives slight geolocation errors, and the 'across the street' is definitely more than 3 metres.
I think any reader here can judge how accurate this estimation is compared to accelerometers. even the two axis accelerometers knows where the front of the car is.. :P
Tilt might be an issue, but you'd know if the road is level, and you're having very high g's when cornering at slow speed = excessive body roll.. or if your car doesn't feel like it's rolling much, but you still register high g's at high speeds.. you have a very grippy car :P.. probably the only car that doesn't roll much is those F1 or SuperGT cars, where their G sensor readings are probably the closest measurement to lateral g that's not done on a skidpad.
Nothing so clear cut on right or wrong. :). More you read on lateral g measurement methods, can argue till next year.. :P
That's how laptimer argues his point, and why it sounds convincing. Google up 'keyboard warrior' if you don't know what that is. :P
Is he TOTALLY wrong? NOT Really. GPS with ground stations do get it down to +/-1m accuracy in OTHER COUNTRIES, but in Malaysia got such high accuracy that can use as track datalogging tool for lateral g's? I guess Laptimer really a 'boleh' person.
Does that means it's useless? People do use it, in fact Laptimer's links to a PDF shows that it can be used to estimate, but that's just what it is, an estimation.
I did install it before, but took it out because wanted to repaint the car, including the interior, and never actually installed it back, since not planning to mod the suspension yet at the moment, since already running spoon dampers.
Note: G = acceleration.. 1G = 9.8m/sec. but in automotive specific use, 1G also means that the car is holding it's own weight sideways through the turn. So 1G means the car is pushing it's own weight sideways. I might be using g and acceleration interchangably, because they're actually the same thing, but different units of measure so hope that clears up any misunderstandings. lateral g means forces acting from the sides.
One thing arguing with Laptimer was making me check my facts. But the way he kept insisting he was right was epically funny, and I enjoyed myself for a while. To give credit to him, he did know some facts, or rather he googled for everything that suited his arguments.. and there came a point where he fabricated some to suit his arguments, so I took that as a cue to exit.
And as I went on reading about these GPS datalogging systems, I find that my initial argument was right, that GPS can't measure lateral G, but he was going on about how 'accurate' his GPS was, and how all measurements are derived and even threw speed as an example, and he was all right, and I was all wrong.
His comment about testing on skidpad confirmed my suspicions that he was a keyboard warrior, and likely doesn't even own a car. Which was why I gave up at that point. For those not in the know, a "skidpad" is a NOT specific place. A big enough parking lot/area will do, just have to make sure the surface is tar/asphalt and not something smooth like cement flooring. This affects grip and you know how much grip you have in those cement floorings like in some parking lots. It has to be level to minimize as much errors possible due to uneven surface. This is how those lateral g is measured by magazines to make it as close as possible for a general comparison. They don't run it at lab standards because it's difficult to replicate every single factor that affects grip like tyre compound, temps and changing surface friction.
The funniest part was him saying that g-sensors/accelerometers are have worse accuracy compared to GPS distance/time measurements, and he can actually replicate the measurements on a skidpad. ROFL!! Baited and caught. Lateral Gs on a skidpad tends to come up low, and he says he can make constant high g's.. Bro laptimer, you must drive a supercar! Yet you didn't win the TTA.. too bad...
Skidpad method
This is the most used method by car magazines you find that lists lateral gs in their car tests, and also considered the most accurate way to measure lateral G.
they use the centripetal acceleration formula which is
centripetal acceleration = velocity^2/radius
Lateral G = (velocity^2/radius)/ gravity constant (gravity is 9.8m/s for those who didn't remember secondary school physics class, though as far as I can remember, they always asked us to use 10m/s as a g constant back then. Amazing that I can remember it after 15 years leaving school lol.
What usually backyard enthusiasts do (like me) do is to draw the circle down with chalk/paint/cones/whatever, and drive around it while tracking the line, going faster and faster each round (each round must maintain constant speed, so if would be 20km/h, then 30km/h and slowly going faster each round) until the tire slips and cannot maintain driving on the line/near the cones anymore. The fastest speed going around the circle before the tires totally slips and cannot track the circle anymore, should be your fastest time, if not, it means you didn't maintain a constant speed going around the circle, or you didn't track the line properly.
My lame car can do 45kmh in a small 20m radius circle before the I can't track the circle anymore due to tires slipping.. velocity is m/s, so 45000m/3600=12.5m/s
Sorry I won't post videos of this, because someone might recognise the idiot who drove around in circles in an empty parking lot near his/her apartment/home one quiet afternoon long time ago. :P
(12.5*12.5/20)/9.8= 0.7971938775510204 Lateral g. (remember that this is a small circle.)
You guys should try this.. Your car might feel like a handling machine, but when you try this, most of you had better prepare to be disappointed.. I don't think you can get very high readings using this method unless you drive a supercar.. I think one magazine posted most sports compact cars with semi slicks having lateral G of less than 1G.... :P.
Laptimer pooh-poohed the 3m or more CEP (circular error probablility) of the GPS, but let's see what happens when I increase the radius to 23m
(12.5x12.5/23)/9.8= 0.6932120674356699 lateral g... and that's just with the minimum error probablility.
The Accelerometer method.
Accelerometers are the like a ruler. A ruler measures length, while an accelerometer measures acceleration. (the term for what accelerometers measure is 'proper acceleration' and this is measured in g-force units. 1G = 9.8m/s.)
It's only considered 'accurate' lateral g if the sensor is perfectly level when measuring, or if the sensor corrects for tilt.
In cars, one thing that affects measurement of acceleration is tilt, this is because two axis accelerometers are affected by tilt. You can see how this works as I tilt my two axis accelerometer to get the readings. Then again, I never said it was 100% accurate, and my initial comment of "can't trust the g-sensor when there's body roll" holds. This is true for anyone's RSM unit with g-sensor as much as it holds true for mine, though more accurately, I should have said, "can't trust it to measure lateral g when there's body roll" Does that mean it's useless? Not really, because we are using the g-sensor as a gauge of how much we pushed our cars through the turns, and we can use it to gauge the amount of body roll.
Does that mean that it can't measure lateral g's? Not fully accurately, but for automotive purposes where comparison is done by per run basis on the same car, it's good enough. The body roll can be considered negligible, depending on where it's mounted. That's why most g sensors tend to end up in the exact center of the car, where roll is most minimal. And one benefit of accelerometers vs GPS is that it knows where the front of the car is at all times.
Then what does it measure? It measures g-forces and acceleration. and direcly, taking into account of road camber, elevation changes, and direction of car. So a car can pull higher g's into a turn if the car rolls enough even with lower grip. I mention the reading's off because I think that the body roll(maybe due to road camber/suspension setting is making the g-sensor/accelerometer reading high even on the slow turns. Laptimer came in and exclaimed there is no error and how wrong I was because the car was using GPS to calculate the lateral g.. He loves saying that everyone is wrong/inferior to his superior intellect. LOL.
That's why I also said only the driver and pit crew knows how useful the data is. As they can measure the previous g-force and current one based on road conditions and car body roll to decide what changes they need to do on the car. The rest is off topic as it became an argument on how accurate each method of measurement is.
Some articles I read even accept the reading of the accelerometer with the tilt/vehicle roll (on level ground) as valid lateral g readings, the reason being that the suspension is already absorbing some of the lateral g's that would have acted on the car. Some articles say no, because a high angle of bodyroll (soft suspension) affects the measurement, so no 100% conclusion. But do remember that tilt plays havoc with g-sensor readings, so even though accelerometer can measure lateral g's, HOW you measure it is also important. This page from the Auxbox manual (add later) obviously shows that if you mount it wrong, the reading will be wrong.. if it's wrong, during calibration it will show E for error, and although you can still log with it in that condition, forget saying it's 100% accurate.
The GPS method.
(I don't know much about GPS 'measuring' lateral g's so I had to read a bit, and my initial assumptions are right, due to one simple fact that I didn't notice till I wrote this.. the position of the front end of the car, sometimes I miss the simplest things)
Laptimer was ranting so much about speed on GPS more accurate than tyre measurements..but we aren't measuring speeds, we're measuring acceleration. :)
The GPS method uses the skidpad formula, but slightly modified since acceleration is not constant unlike in skidpad test. It uses position vs time. Where got need speed bro laptimer...
It measures the distance/time to get the acceleration, then calculates the radius of the turn.
but is only as accurate as the GPS's accuracy, and the data on hand.
So how GPS calculates lateral G data? I tracked like usual way track cars take a turn. Out-in-out. It's called an estimation because mainly because it estimates a lot of things. A GPS doesn't know if you drifted sideways through the turn or you spun out while taking the turn. For gps users out there, you know what I mean.. reversing out of your house also it shows your car driving forward... LOL.
They call it lateral g because it takes two or more points and extrapolates it to be a part of a large circle, and uses the centripital acceleration equation to calculate the lateral G. (this is the cheaper GPS way of calculating lateral g's. Marketing speak on some of the brochures these GPS systems: Accurately measure lateral g's.... ROFL!
first X to third X = distance, 28m time from first X to third X(substitute whatever X position you want), but limit it to 1sec elapsed time, which makes the acceleration to be 28m/s. Position of first and third X has a radius of 60m. One thing I find strange is because Laptimer keep referring to speed, yes, skidpad formula can use speed & time, but that's because speed is constantly maintained. For GPS systems, why bother? can calculate distance based on two or more geolocation points, time elapsed and immediately get acceleration data.
So using the centripetal equation divide by g constant, (Velocity^2/radius)/9.8
(28^2/60m/9.8)=1.3g
Another method measures it versus the turn radius of the track (dotted lines in red). This is the more expensive GPS datalogging systems which has track mapping data. It tracks the acceleration of the car vs. the track's turn radius, not radius between two or more points. Accuracy is higher than the earlier method, but ultimately depending on how accurate the positioning data is. Without ground stations, our geolocation data is not fine enough for more accurate positioning data of the car, so that's why I said it's not accurate.
Summary
So, to make Laptimer satisfied, because he said my comments were making people think that GPS cannot read/measure lateral G at all. Was I wrong? You be the judge.
Sure it can measure Gs, because G = acceleration, but it can't figure out the direction, it just assumes the direction based on positioning data.
So, 3 sentences for those who want to know whether who is right or wrong....
Statement 1) GPS can be used to estimate lateral G an object through a turn, not lateral G of the car. It cannot be used to measure the lateral G of the car due to statement no 2. I bet google didn't tell bro laptimer that. :P
Statement 2) GPS assumes that the car is 'facing' forward at all times, or the side of the car always faces the side of the corner.
(no drifting or going sideways :P) < simplest explanation that I missed. LOL.
Statement 3) GPS gives positioning data, not directional data. Directional data is assumed by comparing current location from previous location via tracklogs. These readings are not real time, they are processed only when the tracklogs are downloaded to a computer.
That's why the article he quoted said it was an estimation. Somehow an estimation became an object of high accuracy, because the some of the product brochures he mentioned used the word 'measure' lateral Gs, plus the word 'accurately'.
Counter argument is that there can be two GPS receivers recording the location of the front vs the rear of the car. But is our Malaysia GPS geolocation data that accurate without ground stations? And who in the right mind in Malaysia runs two GPS receivers on one car? Raise their hands please... :P That particular about position of your car ah?
Oh, if it so happens that someone thinks it is accurate, here's a simple test.. lock coordinates of the front end and rear end of your car.. then measure distance between two points.. If you can get it accurate, you must have some super GPS receiver, and can sell that to Malaysian military for use.. hehehe.. You can see how this error makes it easy for gps to overestimate or underestimate the lateral G.
Heck, I experienced this a lot tagging POIs, and although I'm standing right in front of the shop when tagging POIs, checking back on shows the shop as being across the street. Seems like sometimes GPS signal bounces off buildings and gives slight geolocation errors, and the 'across the street' is definitely more than 3 metres.
I think any reader here can judge how accurate this estimation is compared to accelerometers. even the two axis accelerometers knows where the front of the car is.. :P
Tilt might be an issue, but you'd know if the road is level, and you're having very high g's when cornering at slow speed = excessive body roll.. or if your car doesn't feel like it's rolling much, but you still register high g's at high speeds.. you have a very grippy car :P.. probably the only car that doesn't roll much is those F1 or SuperGT cars, where their G sensor readings are probably the closest measurement to lateral g that's not done on a skidpad.
Nothing so clear cut on right or wrong. :). More you read on lateral g measurement methods, can argue till next year.. :P
That's how laptimer argues his point, and why it sounds convincing. Google up 'keyboard warrior' if you don't know what that is. :P
Is he TOTALLY wrong? NOT Really. GPS with ground stations do get it down to +/-1m accuracy in OTHER COUNTRIES, but in Malaysia got such high accuracy that can use as track datalogging tool for lateral g's? I guess Laptimer really a 'boleh' person.
Does that means it's useless? People do use it, in fact Laptimer's links to a PDF shows that it can be used to estimate, but that's just what it is, an estimation.
Last edited:
