Accler

DMKMM
Posts: 103
Joined: Thu Apr 17, 2008 12:12 pm
Location: Lincoln, NE

Accler

Postby DMKMM » Thu May 29, 2008 4:53 pm

This maybe totally off the wall, so...

I have a DL2 with the accelerometers, i understand they only have 2 and are limited in measuring only in the x,y axis's (for sake of arguement and misunderstanding; x= left, right, and y= back and forth)

right now i have the DL2's face pointing towards the back of the vehicle, and the botom facing down obviously.

if i give it a quarter of a turn so the bottom is facing the left can i measure g, forces in the Z axis (up and down)?

i assume that when you power up the DL2 on flat, level ground with it at the angle it would re-zero its self and potentially measure up and down.

im asking this question, this will prolly only be a one time thing so i havent been able to convince my self in buyin the IMU06.

is there a way you can figure it out mathmatically, how would one go about doing that?

faraday
Posts: 267
Joined: Sun Feb 25, 2007 1:18 am

Postby faraday » Thu May 29, 2008 10:31 pm

The logger is dumb; it doesn't care what orientation it's in. It doesn't zero itself.
Analysis software makes assumptions based on what you tell it in "Vehicle Options". If requested (under the accelerometers tab) it will do a rough re-alignment (Auto-Zero) in the software. This does not change the output of the logger, just how the software interprets it. It doesn't actually do it before the run, but uses the start of the log and dangerously assumes that the car is stationary and sitting level. This is probably never the case, or only instantaneously so if the car is in motion.
If you went to the trouble of checking this, as well as the accuracy of the logger's mounting, you wouldn't need this feature anyway. The software is not dumb, however, and will report an error if the numbers that it sees initially are way out.
There is another option to use the GPS information to correct the longitudinal error, but the two do not work together to further improve the lateral g.
This information is covered in Help, where similar warnings about accuracy are stated in bold type.

Under "Analysis Options" mapping tab of course you can choose not to use the accelerometers to help track the vehicle.
:idea: If you mount your logger as you describe, you will have to do this. The software will report acceleration correctly (or as correctly as you align the logger), but will not re-label the axes. There will be no lateral acceleration reported.
Recent DL1s have a 3-axis accelerometer that will do what you want, up to a point. I assume you want to measure vertical forces due to bumps, and try to see what the springs and shocks are doing. Even in a road car, on large bumps there will be insufficient range in the standard 2g units. This is certainly the case with typical race car ride frequencies, other than on very smooth tracks. A 6g or 10g unit would be better, but would loose resolution for its main purpose.

I'm not sure what you want to do with mathematics.
The Analysis software does the maths for you, if you use one of the mounting options described under the mounting tab of "Vehicle Options" and select "Auto-Zero". There will always be errors in "G-Force" measurements in most data loggers, because of body roll and pitch and the camber or banking on the track. The drift angle of the car while cornering also means that even in an infinitely stiff car, the accelerometers will not be orthogonal to the direction of travel. This is one of the advantages of a GPS based logger, but you probably knew that before you bought it!

We could use matrices to correct for all this if we had an IMU. Proper IMUs do this internally, but as far as I can tell, the IMU06 does not. Similarly, the logger could do a better job of Auto-Zeroing (as could the software using GPS data, but it does not have the computational power to do so. The PC does, but it would add to the processing time. The overiding concern here though, is that of the accuracy and update rate of the GPS data. If satellites were good and the car was moving slowly, etc, etc, some value would be added, but realistically, the typical results from the recreational grade module that I believe the DL2 uses, similar to the DL1, would make this "expensive" computation a waste.
The mathematics is not particularly complex, but something you would learn at university, not high school.
Engineering software (in the old days actual engineers) uses the same maths for many three dimensional problems, such as stress analysis.
Apart from simply using the peak reading as a measure of how well the suspension isolates the chassis from road irregularities, the mathematical analysis of ride, comfort and traction over bumps is very complex.
Engineers in OEM and elite motorsport use test laboratory shaker rigs to do this, partially because the noise in real world measurements masks the parameters of interest too much.

This does not mean that vertical g measurements are useless, however, because it is interesting to study the car's behavior over crests and through dips.
Similarly, an IMU06 would greatly expand the analysis possible. To take full advantage, you should equip yourself with the applied mathematical knowledge to do the matrix transformations and understand error correction. When all six degrees of freedom are measured (along and rotating about a chosen orthogonal 3-D axis system) the errors due to noise and drift in particular directions of interest can be significantly reduced.

You can't use a two-axis accelerometer to mathematically invent an orthogonal third axis, however.
If the logger (with 2-axis acelerometers) is at a known angle (other than 90 deg) to axes of interest, the third axis can be computed, but the practical benefits of this are limited by the reduction of accuracy due to the off-axis mounting wrt the other two axes.
You can't have your cake and eat it too (but somehow this saying does not bear logical scrutiny). :? Confused :?:

The benefit you enjoy with the DL2 is its more rugged construction, connectors and internal component mounting accuracy and vibration isolation. The SPEEDBOX and BRAKEBOX units are required for the GPS accuracy that makes the higher levels of mathematical analysis and error correction really worthwhile. In terms of the principal parameters of interest with BRAKEBOX and SPEEDBOX, position (distance) and velocity (speed), a good proportion of R-T's cost is the mathematical programming to increase accuracy, not just the cost of superior quality hardware.

Hope this helps.

Support

Postby Support » Fri May 30, 2008 9:56 am

The Analysis software does have an option to adjust the longitudinal measurements of the accelerometer using GPS data, this is set in the accelerometer section of the Vehicle Options. By using this option, the effects of constant gradient are largely removed from the accelerometer readings. This is achieved by looking at the change in speed over a certain period of time from the GPS data, and the same from the accelerometer data. If the road is completely flat, they should be the same. If the vehicle is going uphill the value from the accelerometers will be higher, so a correction is applied to the accelerometer data to bring it down.
Obviously the implementation is more involved than that, but that is basically how it works.

Although the DL1 and DL2 don't care about what angle they are mounted at, the AX22 does. It measures the mounting angle and then applies corrections to account for the offset in the measurements caused by the tilt, as well as applying corrections for the fact that the true readings are higher than those measured due to the measurement axis not being in line with the direction of travel.

Martin

faraday
Posts: 267
Joined: Sun Feb 25, 2007 1:18 am

Postby faraday » Sat May 31, 2008 1:02 pm

:shock: What would Isaac Newton think about this garbage?
:cry:

Namlak
Posts: 3
Joined: Sat Apr 26, 2008 11:47 pm

Postby Namlak » Sun Jun 01, 2008 2:03 am

faraday wrote::shock: What would Isaac Newton think about this garbage?
:cry:


Adjust seating position under apple trees on a slope? :wink:


Return to “Sensor-related questions”

Who is online

Users browsing this forum: No registered users and 26 guests