Does accelerometer position really matter?
#1
Does accelerometer position really matter?
I've been using a SoloDL for a few years for DE use and have never relied on the hardware accelerometer, using only GPS derived lateral acceleration to observe cornering forces.
As I look to instrument another car, I have more options and was curious (1) should I really care about hardware accelerometer use over GPS and why; and (2) if I do care, how critical is mounting position?
Does it really have to be in the cars center of (gravity/rotation)?
DB
As I look to instrument another car, I have more options and was curious (1) should I really care about hardware accelerometer use over GPS and why; and (2) if I do care, how critical is mounting position?
Does it really have to be in the cars center of (gravity/rotation)?
DB
#2
Rennlist
Basic Site Sponsor
Basic Site Sponsor
Join Date: Jun 2008
Location: Durham, NC and Virginia International Raceway
Posts: 19,180
Received 3,346 Likes
on
1,900 Posts
Depends on what you’re looking for and how much math you’re using to calculate different measures.
#3
Rennlist Hoonigan
which cost no drachmas
Lifetime Rennlist
Member
Rennlist
Site Sponsor
which cost no drachmas
Lifetime Rennlist
Member
Rennlist
Site Sponsor
Like Peter said, it really depends. For most people, it doesn't matter. If you really want to engineer or calculate things, you want it as close as possible. Many times how the device is mounted is more important than where.
#4
I just looked at GPS derived lat_a v. accelerometer lat_a for a recent event.
The accelerometer data definitely seems to have a higher sample rate and less variability; I suspect it's better resolution data which makes sense as it is directly measured rather than being inferred from GPS.
However, I was looking at an uncalibrated stream and had to make a few assumption to have it make sense (-3.2 offset and 1.1 multiplier).
So that answers GPS v. measured lat_a.
I really want to know how much different in measured lat_a there will be depending on position in vehicle relative to center of rotation (say half way between axles).
DB
The accelerometer data definitely seems to have a higher sample rate and less variability; I suspect it's better resolution data which makes sense as it is directly measured rather than being inferred from GPS.
However, I was looking at an uncalibrated stream and had to make a few assumption to have it make sense (-3.2 offset and 1.1 multiplier).
So that answers GPS v. measured lat_a.
I really want to know how much different in measured lat_a there will be depending on position in vehicle relative to center of rotation (say half way between axles).
DB
#5
Rennlist
Basic Site Sponsor
Basic Site Sponsor
Join Date: Jun 2008
Location: Durham, NC and Virginia International Raceway
Posts: 19,180
Received 3,346 Likes
on
1,900 Posts
I can count on one hand how many club and HPDE drivers carry any significant yaw, much less a level that would even remotely affect lat_a.
You'd need a gyro to test. Here's a good paper. https://arxiv.org/pdf/1704.06053.pdf
The idea that noise is greater than the accelerometer resolution and measure change is telling.
Instant can lead you astray. A curb strike can be far more of a variance than the idea of some kind of lat_a null during rotation.
Or maybe I don't understand what you're trying to ask?
__________________
-Peter Krause
www.peterkrause.net
www.gofasternow.com
"Combining the Art and Science of Driving Fast!"
Specializing in Professional, Private Driver Performance Evaluation and Optimization
Consultation Available Remotely and at VIRginia International Raceway
-Peter Krause
www.peterkrause.net
www.gofasternow.com
"Combining the Art and Science of Driving Fast!"
Specializing in Professional, Private Driver Performance Evaluation and Optimization
Consultation Available Remotely and at VIRginia International Raceway
#6
As an automotive engineer who deals with vehicle dynamics I will always prefer a physical accelerometer over a GPS calculation based on doppler shift. The sensor will provide more accurate data and it'll provide more data points. The higher data resolution allows post processing and filtering to remove some noise without neutering all of your peaks. Locating the physical sensor is easier than you think. Ideally you want it at the CG of the vehicle, but that's seldom possible. Just get it mounted along the centerline of the car where the center console would be. If the car is front-heavy then move the sensor in front of the gear selector. If the car is rear-heavy move it inline with the driver and passenger seat backs. You'll end up close enough to the CG that lateral and longitudinal acceleration aren't massively influenced by distance from CG. Your yaw rate, should you choose to log it, will be much more accurate than having a logger attached to the windshield or dash. It won't be perfect without knowing an exact CG and then calculating for the offset, but it'll be close enough not to lose sleep over it. The majority of OEM inertial mass units are located in the center console somewhere between the dash and back of the front seats.
As for your Solo DL and offsets, does the older hardware not allow you to zero / offset the accelerometer in Race Studio? I picked up a Solo 2 DL a few months ago and really like that feature. I can mount the unit in the car, connect to the laptop, and zero the values so there is no offset. That being said, I still log the values from the vehicle ECU as they are coming from that inertial mass unit located near the CG and already compensated for distance from CG. I wish AiM had a channel for the yaw rate from the inertial unit, but it seems they haven't reverse engineered that yet.
As for your Solo DL and offsets, does the older hardware not allow you to zero / offset the accelerometer in Race Studio? I picked up a Solo 2 DL a few months ago and really like that feature. I can mount the unit in the car, connect to the laptop, and zero the values so there is no offset. That being said, I still log the values from the vehicle ECU as they are coming from that inertial mass unit located near the CG and already compensated for distance from CG. I wish AiM had a channel for the yaw rate from the inertial unit, but it seems they haven't reverse engineered that yet.
#7
As an automotive engineer who deals with vehicle dynamics I will always prefer a physical accelerometer over a GPS calculation based on doppler shift. The sensor will provide more accurate data and it'll provide more data points. The higher data resolution allows post processing and filtering to remove some noise without neutering all of your peaks. Locating the physical sensor is easier than you think. Ideally you want it at the CG of the vehicle, but that's seldom possible. Just get it mounted along the centerline of the car where the center console would be. If the car is front-heavy then move the sensor in front of the gear selector. If the car is rear-heavy move it inline with the driver and passenger seat backs. You'll end up close enough to the CG that lateral and longitudinal acceleration aren't massively influenced by distance from CG. Your yaw rate, should you choose to log it, will be much more accurate than having a logger attached to the windshield or dash. It won't be perfect without knowing an exact CG and then calculating for the offset, but it'll be close enough not to lose sleep over it. The majority of OEM inertial mass units are located in the center console somewhere between the dash and back of the front seats.
As for your Solo DL and offsets, does the older hardware not allow you to zero / offset the accelerometer in Race Studio? I picked up a Solo 2 DL a few months ago and really like that feature. I can mount the unit in the car, connect to the laptop, and zero the values so there is no offset. That being said, I still log the values from the vehicle ECU as they are coming from that inertial mass unit located near the CG and already compensated for distance from CG. I wish AiM had a channel for the yaw rate from the inertial unit, but it seems they haven't reverse engineered that yet.
As for your Solo DL and offsets, does the older hardware not allow you to zero / offset the accelerometer in Race Studio? I picked up a Solo 2 DL a few months ago and really like that feature. I can mount the unit in the car, connect to the laptop, and zero the values so there is no offset. That being said, I still log the values from the vehicle ECU as they are coming from that inertial mass unit located near the CG and already compensated for distance from CG. I wish AiM had a channel for the yaw rate from the inertial unit, but it seems they haven't reverse engineered that yet.
As to mounting. I am considering an EVO5 and my optimal mounting position would be in the passenger dash airbag compartment. I can also square the mount to the car in 3-axis in this location. Unfortunately, this is still offset from the centerline, but not as much as using a windshield mounted logger. The car is a Boxster (mid engine), so I assume the center of gravity is probably right against the engine bay in the cockpit. I might be able to mount the logger there, but that ex-airbag space is really convenient for many reasons.
Trending Topics
#8
Rennlist
Basic Site Sponsor
Basic Site Sponsor
Join Date: Jun 2008
Location: Durham, NC and Virginia International Raceway
Posts: 19,180
Received 3,346 Likes
on
1,900 Posts
As an automotive engineer who deals with vehicle dynamics I will always prefer a physical accelerometer over a GPS calculation based on doppler shift. The sensor will provide more accurate data and it'll provide more data points. The higher data resolution allows post processing and filtering to remove some noise without neutering all of your peaks. Locating the physical sensor is easier than you think. Ideally you want it at the CG of the vehicle, but that's seldom possible. Just get it mounted along the centerline of the car where the center console would be. If the car is front-heavy then move the sensor in front of the gear selector. If the car is rear-heavy move it inline with the driver and passenger seat backs. You'll end up close enough to the CG that lateral and longitudinal acceleration aren't massively influenced by distance from CG. Your yaw rate, should you choose to log it, will be much more accurate than having a logger attached to the windshield or dash. It won't be perfect without knowing an exact CG and then calculating for the offset, but it'll be close enough not to lose sleep over it. The majority of OEM inertial mass units are located in the center console somewhere between the dash and back of the front seats.
As for your Solo DL and offsets, does the older hardware not allow you to zero / offset the accelerometer in Race Studio? I picked up a Solo 2 DL a few months ago and really like that feature. I can mount the unit in the car, connect to the laptop, and zero the values so there is no offset. That being said, I still log the values from the vehicle ECU as they are coming from that inertial mass unit located near the CG and already compensated for distance from CG. I wish AiM had a channel for the yaw rate from the inertial unit, but it seems they haven't reverse engineered that yet.
As for your Solo DL and offsets, does the older hardware not allow you to zero / offset the accelerometer in Race Studio? I picked up a Solo 2 DL a few months ago and really like that feature. I can mount the unit in the car, connect to the laptop, and zero the values so there is no offset. That being said, I still log the values from the vehicle ECU as they are coming from that inertial mass unit located near the CG and already compensated for distance from CG. I wish AiM had a channel for the yaw rate from the inertial unit, but it seems they haven't reverse engineered that yet.
The inertial unit is much more sophisticated in the MXx line of display loggers and the EVO4S/EVO5.
As you say, attaching a separate accelerometer sensor (IMU), for those systems that offer that option, usually located on top of the center tunnel (between the seats on a GT3 Cup and R platform is common) is optimal.
Thank you for the detailed explanation. 1. I can remove an offset and provide a first order calibration in RS - there is an ability to apply a multiplier and offset (y=mx+b) to the measured acceleration. So, an offset is not enough if the sensor (SOLO) is not physically squared to the vehicle which it isn't. With a little algebra, I can get the calibration pretty close.
As to mounting. I am considering an EVO5 and my optimal mounting position would be in the passenger dash airbag compartment. I can also square the mount to the car in 3-axis in this location. Unfortunately, this is still offset from the centerline, but not as much as using a windshield mounted logger. The car is a Boxster (mid engine), so I assume the center of gravity is probably right against the engine bay in the cockpit. I might be able to mount the logger there, but that ex-airbag space is really convenient for many reasons.
As to mounting. I am considering an EVO5 and my optimal mounting position would be in the passenger dash airbag compartment. I can also square the mount to the car in 3-axis in this location. Unfortunately, this is still offset from the centerline, but not as much as using a windshield mounted logger. The car is a Boxster (mid engine), so I assume the center of gravity is probably right against the engine bay in the cockpit. I might be able to mount the logger there, but that ex-airbag space is really convenient for many reasons.
#9
Rennlist
Basic Site Sponsor
Basic Site Sponsor
Join Date: Jun 2008
Location: Durham, NC and Virginia International Raceway
Posts: 19,180
Received 3,346 Likes
on
1,900 Posts
Given that you are now considering the EVO5, the wireless feature works well, using the SD card as a backup. So mounting options are more open with this device than they would be on an EVO4S. On historic cars, I’ve fashioned a USB bulkhead extension cable to bury the EVO4S and still access for easy download.
#10
Given that you are now considering the EVO5, the wireless feature works well, using the SD card as a backup. So mounting options are more open with this device than they would be on an EVO4S. On historic cars, I’ve fashioned a USB bulkhead extension cable to bury the EVO4S and still access for easy download.
ping as well - perhaps useful to minorly clean up acceleration measurement)
#11
Rennlist
Basic Site Sponsor
Basic Site Sponsor
Join Date: Jun 2008
Location: Durham, NC and Virginia International Raceway
Posts: 19,180
Received 3,346 Likes
on
1,900 Posts
Yes, perfect.
The AMB should be relocated, so as not to foul the mounting of the EVO5 or the connecting cables, but that’s just the spot.
Would not recommend screw mounting. Would recommend degreasing the painted surfaces on the tunnel, and the underside of the EVO5, and using the High Strength, hard plastic 3M Dual-Lock on both the car and the EVO 5 surfaces for mounting.
This will achieve the correct damping and get rid of the highest frequency noise, while still being very, very secure.
The AMB should be relocated, so as not to foul the mounting of the EVO5 or the connecting cables, but that’s just the spot.
Would not recommend screw mounting. Would recommend degreasing the painted surfaces on the tunnel, and the underside of the EVO5, and using the High Strength, hard plastic 3M Dual-Lock on both the car and the EVO 5 surfaces for mounting.
This will achieve the correct damping and get rid of the highest frequency noise, while still being very, very secure.