Jesper has created an automotive performance meter using a handful of ICs. The key chip is the ADXL202 from Analog Devices. It measures acceleration while the AVR 8515 keeps track of time and does all of the calculations. All you need to do is enter the weight of the vehicle and the meter will calculate your 0-60 mph time, 60-0 braking distance, and maximum horsepower. There is a complete schematic on Jesper’s site, but the code still needs to have a few bugs worked out. This project is essentially a reproduction of the first generation G-TECH/Pro meter (bunnie has a picture of the original board). The new generation G-TECH meters are fun to play with and do interesting things like determining the engine RPM by measuring the noise in the electrical system.
That’s pretty nifty, It might be interesting to see an expanded application, perhaps with lateral Gs and a graphical readout.
It’s a cool thing, but unfortunately more on the useless side in the real world. The cummulated error over time of the acceleration measurements will make the readings have large errors after a short while. This isn’t something one can correct completely in software.
Aircraft inertial guidance systems are considerably accurate (and much more complex). To give you an idea of how accurate these things can be: “(2) For flights over 10 hours’ duration, a tolerance of +/-20 miles cross-track and +/-25 miles along-track on 95 percent of system flights completed is permitted.” (Federal Aviation Regulations, Appendix G to part 121). So, fly to Paris from the U.S. using only your INS, and you should be able to see the Eiffel tower at your destination if visibility is good. Not bad if you ask me.
bogdanm, could you elaborate? Are you referring to errors in the short term readings, such as what was my peak acceleration when I stopped at that stop sign after 1 hr of driving, or are you referring to the fact that it can’t be used as a reliable inertial guidance system for any useful distance?
Well, I guess the best answer is given by Analog Devices themselves (the manufacturers of the acceleromter part):
http://www.analog.com/library/analogDialogue/archives/33-08/accel/index.html
Look especially at point G. It’s the integration error that’s killing your application. We’re talking about cummulative error here, and this is the worst of all. The error will just get bigger and bigger in time. One way to overcome is again specified by them: in conjunction with a GPS system. As for the aircraft guidane systems, I know they’re amazing, but I’m also quite convinced that they involve more that just accelerometers.
I have a G-Tech Pro meter and it’s always been pretty accurate. When measuring quarter mile times, it’s been within around +- .3 seconds of the actual times when tested at the drag strip. That’s a pretty reasonable margin of error if you ask me.
yes no 5. INS does have more components that just accelerometers. they also included gyros. old ones used real gyros, new stuff uses laser ring gyros.
actually to be honest each airliner uses three ins setups and uses triple mix (not averaging) to get the position. they are updated by gps, older ones flew over known groundstations and updated the error manually.
i have a g-tech pro comp meter that i use after tuning my corvette. i have found that the g-tech device is within 1% of what i get at the dragstrip & at the dyno. it’s pretty awesome for $250… plus it does a lot of other things too!
bush 2 is here http://www.dbg2950.blogspot.com
This is fine for a car performance meter. The integrated bias error is not going to build up significantly over a quarter mile.
If you are trying to navigate with it, it will not work due to accumulated bias errors. I’ve been working with putting an INU in a r/c airplane. Ours has static and dynamic pressures, 3 axis magnetometers, 3 axis rate gyros, 3 axis accelerometers, and GPS. This combination will allow you to guide an airplane pretty well without crashing.
Sparkfun has a 3 axis accelerometer that would work pretty well for this. Third axis will tell you if you flipped over.
Yes, I suppose a quarter mile would be fine. I was talking about ‘slightly’ longer distances though :) In the range of tens of kilometers and similar to this. It all depends on the application in the end. If you want it to be accurate just for a quarter mile, I’m quite convinced it can do the job.