Accelerometer Noise [Archive]

11-09-2006, 07:35 PM

Hi all,

I'm using lateral and longitudinal accelerometers on last year's car in testing and running into noise issues - the engine increases the noise magnitude by a facter of around 10.

In addition to bandlimiting the signal, I'd like to try and reduce it mechanically. Has anyone here had success in mounting an accelerometer such that engine noise is reduced, and how was it done?

I have been experimenting with different mounting places/materials/methods, but some guidance would be really helpful.

Cheers,

Dan

11-09-2006, 07:35 PM

Jonathan S

11-09-2006, 08:03 PM

How do you have your accelerometer mounted?

Straight from MoTeC:
The best method is to locate the center of gravity of the vehicle and mount the sensor near or on top of it if possible. The best method of attachment is usually 2 sided tape or Velcro.

I have always used velcro, and we have no issues with noise. Depending on your datalogger's capabilities, you could try applying a filter to the channel to try and clean up the signal.

One last thing I can think of (not sure if it could really cause this or not) is a poor 0v (ground). Motorcycle charging systems are pretty "noisy".

js10coastr

11-10-2006, 01:06 AM

velcro has always worked for me.

What is the logging rate you're using? And where did you get the sensor from?

Boston

11-10-2006, 02:33 AM

Velcro mounting is a must.

formula_wally

11-10-2006, 04:55 AM

Our data acq with the 2-axis on board accelerometer is rubber grommet mounted for isolation.

If you happen to be running externals, or something from freescale, try choosing a 0~2g range with the lowest roll-off freq.

11-10-2006, 10:07 AM

I'm using two sensors - an ADXL203 (2-axis, +/-5g, 0-5V) and a MicroStrain 3DM-GX1 (Relatively slow but compensated accelerometer/gyrometer/magnetometer with wicked orientation algorithm). I will have to try mounting on velcro - thanks for the suggestion.

I checked my grounds many times and I'm running a seperate battery for the logging equipment, but you can never check grounds too many times.

Thanks for the replies guys!

Dan

Buckingham

11-10-2006, 11:48 AM

Run it through an FFT and then apply an appropriate digital filter.

Tony K

11-10-2006, 01:30 PM

The battery for your data acquisition system may be causing some of your problems as well. We ran into an issue with our engine dyno data acquisition system because we were running the NI PXI chassis on an external power supply and the engine on its own battery. One of the sensors was grounding out through its own casing (which is normally not a problem with a common ground system) and interfacing with the engine power supply and created a nice little pathway for a huge amount of noise to be transfered from the ignition system.

Jonathan S

11-10-2006, 02:27 PM

Why are you running your DAQ through its own battery?

11-12-2006, 01:32 PM

The charging system on our car isn't working because the magnets came off the flywheel, so we are running the car total loss and stopping to charge frequently. Unfortunately we had to modify the flywheel significantly to achieve sufficient crank position sensor resolution and it's going to take us a little while to get a new flywheel and modify it for the CPS.

The MicroStrain sensor is powered through a 110V adapter, so the second battery is running an inverter for it and keeps the ground loops seperate and reduces power supply noise (which is really minimal with our setup), and allows us to run a little longer before we kill the cars battery. I'll have to see if I can get a 12V DC cable for the sensor, it's really kind of a pain. It's not pressing though, as the snow is preventing us from testing at a track.

DH

formula_wally

11-12-2006, 04:58 PM

Hey DH, I dont want to flame you, just help out a fellow Canadian team.

If your car isn't charging, and a 110 volt invertor is even just close to touching your car, you might have bigger issues to deal with first.

Secondly using a 110 invertor from an already overdrawn battery, is very dirty and shouldn't produce clean dependable power.

Try and source a 0-5 volts +/-2 G acceleromter from mouser, digikey, or freescale.

What motor/management are you running that requires a homebrew CPS

11-13-2006, 01:59 PM

Hey formula wally,

There is a second battery used for the data acquisition system that is easily capable of providing for the inverter. The inverter noise is extremely minimal, as it has been set up with noise in mind. In fact, the inverter and second battery are on an isolated circuit (and yes, I have dealt with the ground loop issues inherent in this setup).

There is a very good reason for our choice in transducers, as we want to measure more than just acceleration so that drift can be estimated and rejected.

The reason we needed to modify the CPS is because the EX500 is an odd-firing engine, and requires better CPS resolution than the stock configuration to achieve fully sequential operation (the stock sensor gives 4 pulses for 720 crank degrees, just enough to run the stock CDI). Autronic is the ECU, and I don't have many bad things to say about it that don't relate to the absolutely terrible manual that comes with it.

Like I said, we are working on improving the system as it is being built. I have tested the system in many ways and the largest source of noise comes from engine vibration (and once it is dealt with, there will be very little noise).

Dan