Nice video, now I'm glad I sold my bar and put in a brake cylinder brace by LHT performance!

 

Considering you're only getting appreciable signal from rough pavement, you're not measuring chassis flex at all. You're measuring how badly the end of a very long cantilever is vibrating.

And no, I'm not saying that a strut tower is necessary at all. But give the end of that bar a tap and I bet you'll see similar measurements that you saw on track day. But if you saw long-duration DC excursions in the sweepers, that would be a good indicator of flex.

Edit: Oh, wait. You don't have to tap. At 2:50, you see how easy it was for you to manually move the end of that bar with just your hand.

 

Do you know how many lateral g's you were pulling? Because, no offense, it really didn't look like you were going very fast. I'm not saying tower bars are beneficial on the S2000 but you may need more load going through the chassis before the chassis flexes, after all Honda did put a rear tower brace on our cars.

 

Thanks for watching.
The sensor measures in two independent directions. The Up-Down motion of the bar is detected in the "Y" data feed from the sensor and is not displayed during the video. The Left-Right motion of the bar is detected in the "X" data feed from the sensor. In my testing the X motion was not affected by the arc created by the end of the bar when moving up and down from vibration.

The next time I have the sensor mounted on the car I will see if I can capture an example of how the directional data is independent.

 

Thanks for watching. I agree, it is not a blistering fast pace. The partial lap at speed shown in the video ended with a time of 1:14.2. I would say I was giving it around 80% effort of my ability and is 2.5 seconds slower than my best lap at the track.

For the left hand sweeper at around 4:50 in the video I am going around 68 mph at the apex reaching 0.67g.

In my closing voiceover I state that I will revisit the data once the car is turning faster lap times. I hope to do that later this year after some planned changes to the car.

 

Thank you for your input.

Personally, I see this as validation for a strut tower bar. I’m in the group that prefers to see my tires, bushings, and suspension which are designed to absorb the changes in dynamics rather than the chassis. My autocross surface is just a series if cracks and bumps. It would keep my chassis stiff and my suspension doing its optimal job and allow for more precise tuning.

 

Since you already have the setup figured out, it would be interesting to see how much flex is seen when you’re jacking the car up from one point.

 

Wouldn't the vibration cause the joystic to vibrate and register movement even if there is no flex? I'd imagine driving over rumble strips would vibrate it and incorrectly register deflection.

 

Great idea, and really great vid! However, I'd think the total "chassis flex" (displacement) Z would be the RSS combination of the X and Y data**, Z = sqrt(X^2 + Y^2). In fact my sense has always been that the Y flex/displacement was the primary concern of strut bars: that's the main direction of the wheel's motion over bumps and in turns, and hence the transmitted force to the chassis through the suspension.

I'd be curious to see the Y-axis data. Even allowing for a significant "noise" contribution from the cantilever effect the strut bar, there might be something interesting to glean. For example, on a given lap you could note the max Y values on straights, and then adjust the Y data in corners by ignoring anything less than that. That's crude, but I think it would be enough to indicate whether the Y-axis contribution is worth further study.

**Technically we should also allow for Fore-Aft displacement W, so that Z = sqrt(X^2 + Y^2 +Z^2). But we can agree W is negligible; and even if not, a strut tower bar wouldn't counteract it!