Engineering a sway bar

[fiftytakedowns]

how much wilil you be sellin these for when your done? do you plan to sell your design to Ireland engineering?

[TwoJ's]

I won't be selling the design. If I did, I certainly wouldn't sell it to IE.

I also won't be producing these to sell either. In the future that is a possibility, but I may have one or two things to change on it if that is the case.

[george graves]

Don't drop that scope! (Or you'll be on KP duties till you graduate) - actually it's not that bad it's like $1500. I just bought a $400 fluke multimeter - so I can feel the sting.

Did you get the signal filtering to how you like it?

[TwoJ's]

The noise is acceptable as is, but that was done using the power supply in that picture. I may run in to much more interference when I attach it to the 12V from the car. I went to test it last night and the cig lighter was dead, so I am off to figure that out real quick like.

[george graves]

The sparkfun's break out board has some filtering on it's power supply. You should be ok. I'd plug in the accelerometer after you've started the car though to avoid negative voltage spikes.

Or you could build a small battery pack for it. The chip only needs 3.3v. The regulator on board needs a bit more - around 4 volts. 4 AA batteries would do nicely. I know this project is nickel and dime'ing you - but another $10 at radio shack should get you take care of. You can't get cleaner power than batteries. A battery holder, 4 AA's, and a small switch. Oh and some double sided foam tape - attach it right to the bottom maybe.

Also, I looked, and the break out board already has some output filtering to it. 0.1uF caps on each of it's 3 outputs. If you need to filter it a bit more you could add another more caps.

How does the accelerometer respond to vibration? You might be trying to measure a 1.0g lateral force in the car, but if you have a car that is vibrating it's center console 1/100th of an inch at 2 g's, back and forth, you're going to get some noise? I don't know - I don't know how that sensor responds to vibration. Usually when that chip is used it's used with some type of hardware filtering, or software filtering. In software you'd grab 5 smalls real quick, and average them together to get 1 sample. Or you might keep sampling them to an array and each time you write to the array, you re-compute it's average. Kinda a running average. In hardware you might just add some more caps to the output. It looks like sparkfun already added caps to the board for you - so you may be ok. Sorry I feel like I'm confusing you more than helping you. that was not my intention. Can you go for a practice run with the set up?

[TwoJ's]

Thanks george for the input. I'm not too confused at all by that. I'm going to go in to the lab and find someone to ride with me and be my data logger. I'll do a few test runs, and I'll host the excel files so you can take a look at them if you want to give me some input.

[McGyver]

wow, im just finishing up my sophomore year of ME at vcu in richmond. its nice to see the applications of what im learning! good lock with your job and i hope the bar comes out great!

[TwoJ's]

George, you hit the nail square on the head! I went and did three test runs and I had a ton of noise -- enough to make it absolute garbage data. I put it in excel and it appears there was a 100 mV spike for each ignition pulse. I just went and got a 9V battery, and surprise, radio shack doesn't have the proper connector for it. I am really starting to hate RS.

I'll report back later, hopefully with some real world data.

[george graves]

So you had ignition noise *and* physical/mechanical vibration noise?

[TwoJ's]

Yes, I was getting about 500 mV of ignition noise, not the 100 I originally thought (had it on a different scale than I thought). I fixed that issue with the battery, but I am still getting the mechanical noise.

http://www.twojsengineering.com/files/test1.xlsx

That is the excel file from a run I did. A sharp left followed by a long bender and then a sharp right turn. Channel A was the accelerometer and B was the gyro. Notice the Gyro is basically giving me nothing. Play around with it if you would like. I quickly made a chart with a 5 period moving average (which is much higher than I would like to use).

I have no time to play with this tonight. I'll have to deal with it more tomorrow.

I've got some work to do. There is a 4.5x scaled output on the circuit board for the gyro that I am likely going to have to use.

[TwoJ's]

Okay, so I have finally gotten some workable data. I have abandoned the gyroscope. For starters, the roll angle of the car with my sway bar is so low that the roll rate is within the noise level of the circuit. On top of that, I came to the realization that I was not placing the gyro directly on the roll axis of the vehicle. I won't go in to that too much, I'm sure you can visualize that scenario and realize the effect it would have on a gyro.

So I decided that maximum lateral cornering acceleration would be my primary test data. Through a friend, I was able to go down to Renegade Raceway in Yakima and do some "skidpad" testing. I did 5 runs going left and 5 going right with my sway bar installed and the stock sway bar installed. I was able to get some good data and will proceed with my analysis and testing report.

Here is a plot of lateral acceleration vs. time. It is negative because this was a right turn. You can clearly see the point of maximum cornering force, and then loss of traction. This particular run induced oversteer.



I ended up with a 10.6% increase in cornering acceleration with my bar vs. OEM.

[enigmaticdream]

Dude, that's an awesome senior engineering project. Congrats on the 10.6% increase. Also, Go Seahawks!

[TwoJ's]

go seahawks!

yes!

[fiftytakedowns]

Those are the kind of numbers that sell products

[ThatOtherGuy]

dude...

You are fucking awesome