Poor Man's Ride Height Sensors: some (more) data

[Install425]

I love this stuff.

[Jack Olsen]

I did some calibration. It involved getting a 'before' measurement and sensor reading at resting ride height, then adding 200 pounds of sand to lower the end of the car down, then getting sensor readings for every 1/4 inch step as I raised the car back up with a jack. That gave me a chance to compare the initial ride height sensor reading (at rest) with the ride height sensor reading when the ride height was brought back to that level with the jack and the weight. I also measured a couple of additional inches up above that point in 1/4 inch increments to see if the readings I was getting from the sensors were linear or not.

Does that make sense?

It ended up looking like the sensor's output was pretty close to linear -- meaning the voltage readings proceeded in equal increments as the car raised up. The front and rear increments weren't the same, though. I'm sure that has to do with the way the sensors are mounted. But in the front, 1/4 inch of suspension travel changes the readout .07. In the back, 1/4 inch of suspension travel changes the readout .108. This lets me connect real suspension travel distances to the sensor data numbers.

I also went through the data itself to make sure the numbers I was having the computer average were from portions of the circuit where the car was always going 100 mph. In some of the data, I found the speed dropping earlier than in others. I narrowed the segment lengths so my data would have less of the data that was from not-100-mph travel.

And I also found one anomaly that I won't be able to explain until I do more tests. The at-rest ride height readings from the calibration check don't sync up with the at-rest readings from the day I did the test. Worse, the at-rest readings from that day don't really jibe with the other sets of readings I got. Something funny was happening that day when the car came to a stop. I don't know if the sensors were shifting position, or what, but everything looked much more credible and consistent when I used the no-aero-devices attached run as a baseline for comparison instead of the at-rest readings.

Still with me?

With all that said, it looks like my initial findings were not 100% right, and they weren't completely wrong, either. In other words, the front end isn't actually rising up when the rear wing is working. But the rear wing is reducing the amount that the front dips. The front drops more with no wing in the back than it does when there's a wing back there.

Also, the blue 'starting ride height point' lines in the graphs I posted were not correct. They made it look like the front end was riding higher in every test (except when the rear wing was removed) than it was when the car was standing still. This wasn't the case.

Here's the more accurate information.

If you use the run with no wing and no splitter as a baseline, the following happens:

When I put on just the front splitter, the front end runs .41" lower, and the rear runs .04" higher.

With both the rear wing and the front splitter, and the wing set at 12 degrees, front end runs .24" lower, and the rear runs .01" lower.

With both the rear wing and the front splitter, and the wing set at 16 degrees, front end runs .09" lower, and the rear runs .17" lower.

With both the rear wing and the front splitter, and the wing set at 20 degrees, front end runs .08" lower, and the rear runs .26" lower.

Now, because of my sand tests, I can ballpark actual pounds to go with these figures. But I don't know if the data is meaningful or useful, since a ride height measurement can't distinguish a force pushing the front down from a force that might be pushing the back down so hard that it works to lift the front up, in a teeter-totter effect, giving a number that's just the sum total of all the forces involved.

But my math indicates that the splitter is pushing the front down to the tune of 103 pounds when there's no wing in the back. Once you add the wing, the change up front appears to decrease (in spite of the fact that the splitter is doing the same work in every run). With the wing at 12 degrees, the front registers 60 pounds pushing down. With the wing at 16 degrees, it's 22 pounds in front. With the wing at 20 degrees, it's 20 pounds in front.

With no wing in back, there's actually 13 pounds of lift showing in back, which is probably the teeter-totter effect working in reverse, with the front 103 pounds of force lifting the back up a little. Once the wing goes on, the rear starts coming down, to the tune of 31 pounds at 12 degrees, 55 pounds at 16 degrees, and 82 pounds at 20 the degree setting.

I don't think the rear wing is only generating 82 pounds of downforce, though. Partly because the data I posted earlier says the car itself with the ducktail is generating 75 pounds of rear lift at 100 mph. So maybe that means the wing is doing 157 pounds of work to get the ride height to that point. I suspect it's even more, since the wing stretched its base much more after a lap at Willow (going up to 125-30 mph) than it did with 250# of sand loaded on it for an earlier static test.

It should also be noted that these numbers all rely on my ability to do basic math. I assure you that the odds of me making a key error at some point in all of this are very high. I haven't taken a math class since high school, and that was back in the eighties.

But it's more data. I want to try repeating it after I make sure that there's no slop in the way the ride height sensors are mounted. I want to figure out why I was getting the goofy at-rest readings during the test. I'd also like to test the car with a flat rear decklid, and the 3.8RS lid I used to use, and also with the big wing I used to run. Now that my 'Poor Man's Wind Tunnel' is on the verge of being reliable, I want to tie up all the loose ends and see if it will generate consistent and repeatable data.

Of course, I've got to fit it into my spare time. So this thread will be popping up again in a few weeks.

[cstreit]

Subscribed!


Jack,

I made a set of "Poor mans" DTM style front winglets a few weeks ago in hopes of reducing some of the understeer my car was experiencing...

BOY did it! If you have the time I'd recommend investigating those pictured above. Took me about two hours to make and mount. First attempt had too much angle. Second attempt worked nearly as well and gained me about 100 RPM on the main straight.

Mine pictured below:

[Jack Olsen]

It doesn't take all that much of an excuse to get me to drive 300 miles on a Saturday, does it? I took the 911 out again this morning to test some ideas for bringing the front down. The drive was fun, but I still haven't found a great way to get the nose lower.

My first idea was to extend a curtain below the 5-inch spoiler to reduce the amount of air that gets under the car. I built a pair of bolt-on pieces that would be easy to replace after an off-track exctursion. It wasn't pretty, but it kept the clearance to a minimum.



I also threw together a mount for the front wing I'd built. I had read the concerns that the low pressure zone under the wing might cancel out the wing's downforce as it pulled 'up' on the car below it. I also had my own concerns about how ugly a front wing would be on a 911. Those concerns became worse when I came up with a scheme to extend the wing a little bit further forward.

It was might ugly.

I'm serious about hat. Cover your eyes if you're sensitive. This is going to sting.









Well, the nice thing about a wing up front is that you can see when it's working. At 100 mph, it was pushing down on the support rods enough to bend them. At the same time, though, the lightweight hood was actually fluttering in the turbulence (or the low pressure area, I guess). It actually pulled part of the rubber seal out from under the trunk.

The good news? The front wing didn't work. The wing is pushing down, but I think it's also pulling up on the car in an equal amount. This means I don't actually have to put it on at an event where other drivers might see it.

The not-so-good news is that with or without the wing, the vertical front curtain had almost no effect at all. According to my ride height numbers, the front is riding less than 1/16 of an inch lower with the curtain under the splittler. That's maybe in the neighborhood of 4 pounds of downforce -- less than a gallon of gas in the tank.

So, no solution for the front end, yet.

I'll keep thinking about other possible solutions.

[Elombard]

Jack and Chris have you seen the "winglets" that Mitch Rossi put on the front spoiler of his car? They are on shown on the picture on the front of his new book about modifying a 911 for street/track. He claims they are effective and they dont look too bad.... Jack you have probably seen them live in So cal.

[Jack Olsen]

Everything I've read about the front corner winglets is that their effect is pretty minor. But I'm going to look at the Rossi book and see what he says about them.

Here's a graph that shows how acceleration effects front and rear ride heights. Part of what I'm seeing with front ride height is from the fact that the air is pushing the car backwards as the car pushes forward. In the graph, the car goes up to about 133 mph and then deccelerates. The front ride height line trades places with the rear ride height line. (Remember, with these sensors, a low point on the graph means that end of the car is high, and vice versa.) So under throttle, the front is high and the rear is low. Then the front dips down and the rear comes up as the car slows down.

[Quicksilver]

I'm wondering if a portion of your front end lift is coming from the decrease in rake that pushing the rear down causes. The reduction in rear 'under tray height' would reduce the scavenging of the air under the car but the front would be allowing the same quantity of air under the car. (or increase the quantity of air from the cantilever effect...) This would cause the air flow under the front to slow down which means increased air pressure.

You are having so much fun with this. Why don't you make a multiport manometer rig and collect air pressures from different location on the car. (such as under the front)

On a different subject... A suggestion on your rear diffuser arrangement. You are pushing the cooling air down into the diffuser's airflow. I would suggest a NACA duct or a stepped outlet kind of like the cowl flaps on an airplane. I would think it would smooth the airflow and possible help the cooling airflow.

BTW - I think the to Lamborghini’s front wing was actually for US bumper height laws. (someone know for sure?)

[gestalt1]

i'm sorry i posted that photo of the countach - i was joking. i did not realize you would actually build one to test. i'm glad it didn't work. i guess you could sell it on ebay so some kid can put it on there corolla.

the front skirt idea is interesting, i guess the volume of air splitting under the car is important for speeding up the velocity of the air under the car. i would think that the right amount is nessesary for a rear diffuser to work.

[Jack Olsen]

No apologies necessary; the craziness was all mine. I'd built the front wing weeks ago. I originally came up with the idea back in July, but wasn't able to find an example of anyone actually trying it on a production car.

Here was my original drawing:



Based on the airflow over a 911, I also thought one could be put right above the windshield (C), although it might suffer the same 'cancelling out' problem. In front (A) would probably work the best, but it would be hard to maintain its height relative to the ground because of suspension height variations. It would also be a nuisance to other drivers (and pedestrians).

[Quicksilver]

There is no reason that you can't mount a wing in front like example "A" but make it higher above the ground. For stability's sake I would think you would want it about 1-1/2' to 2' above the ground. If you put it close to the ground the downforce will oscillate in an unstable fashion as the height changes.

In the case of a rear wing, when the wing pushes harder the rear of the car lowers and the wing's angle of attach slightly decreases. This is a stable reaction because it tends to regulate the force. In the front this will be unstable because when it pushes down the wing gets closer to the ground and increases the venturi effect, which pushes it even harder down. When the wing's height goes up it loses downforce, which deceases the venturi effect and makes it go even higher. This creates a situation where it wants to create an oscillation instead of stabilizing.

To combat this winged cars have insanely high spring rates, which you don't want to get into on a 911. (unless you want to build a whole new chassis that can deal with the pounding) The car will have more of a tendency to porpoise over any ripple in the road anyway.

I think you have two conventional solutions you can use.
- One is extend the splitter and change the spoiler so it isn't so vertical but more angled like a snow shovel. This would help keep the high pressure in front of the car from pushing more air underneath the car.
- The other idea is to manage the air around the front tires to get the air scavenged out through the sides. The tire is going to squeegee all of the air off of the road and you want as much of it as possible to go around the outside of the tire as opposed to the inside.

Anything you can do to let the low pressure area in the rear of the car extend as efficiently as possible all the way to the front spoiler will get the job done.

[Jack Olsen]

Some quick and dirty origami-style canards:







I have no idea if these will do me any good, but I'm going to try them out tomorrow.

[cstreit]

Slick!

I'm really curious to know what their impact is... Thanks for doing this research Jack!

[BertBeagle]

Originally posted by Jack Olsen
Some quick and dirty origami-style canards:







I have no idea if these will do me any good, but I'm going to try them out tomorrow.

These are /were used on ground effects cars (ie the IMSA GTP cars of the 80's). What they accomplish is creating a vortex down the side of the car, filling the low pressure area that exists along the sides of the car, and that kept the air entering the underside of the car from the nose under the car and effectively increased the down force with a low drag penalty. Aero guys are always looking for something for nothing (ie downforce for no drag penalty) and this was a good find.

The primary function of the devices (dive plains, as they were sometimes called) was not to create down force by forcing air up over them and pushing down on the nose.

They were very misunderstood in their function.

For some good reading on aerodynamics (and race cars in general) explore these two great sites.

http://www.mulsannescorner.com/

http://www.suzukaracing.com/ (The owner of this site is a well respected aerodynamicists who has designed many Le Mans cars. He brought the dive plane back in the 80's. It had not been used in a number of years.)

[Driver Ed]

First, let me echo the other posts which have said "thanks for doing (and sharing!) this research." Second, I have some specific questions about equipment. First, what datalogger are you using? I've been gathering notes to build my own from scratch, and I thought it would be useful to learn what others are currently using, especially if you have opinions on what features are especially useful/useless. At this point, I'm sure I'd like to gather data, but there seems to be a wide divergence of opinion as to what data channels are useful. For instance, I'm not so concerned about engine management, since I don't plan to redesign it, but I might be interested in suspension monitoring, wheel speeds, tire temps and steering angles to monitor things I might want to change. Anybody have words of wisdom there?

Ed

[Jack Olsen]

Originally posted by BertBeagle
These are /were used on ground effects cars (ie the IMSA GTP cars of the 80's). What they accomplish is creating a vortex down the side of the car, filling the low pressure area that exists along the sides of the car, and that kept the air entering the underside of the car from the nose under the car and effectively increased the down force with a low drag penalty. Aero guys are always looking for something for nothing (ie downforce for no drag penalty) and this was a good find.

From the Suzuka Racing site:



I'd heard that the canards didn't do much for downforce. Now it makes more sense to me. I'll run a test with these and see if they make any difference, even accidentally.

Maybe I should try wings off to the side of the headlights.

Originally posted by Driver Ed
For instance, I'm not so concerned about engine management, since I don't plan to redesign it, but I might be interested in suspension monitoring, wheel speeds, tire temps and steering angles to monitor things I might want to change. Anybody have words of wisdom there?

I use a DL1 data logger, made by Race Technology in Britain. I like the multiple inputs for different sensors (I use throttle position and brake as well as front and rear ride height. But I also like the integrated GPS and accelerometers. And of course the software allows a short-bus guy like me to actually make some sense of the data.

One question on this line, though. Why do data loggers generate such high lateral G numbers? I show sustained g's in the 1.4 range, maxing out at about 1.6. Common sense tells me my car isn't generating 1.6 lateral g's. I know how to recalibrate the accelerometers, but that won't change the lateral g inflation that all the different mainstream units seem to share.

[randywebb]

what is the sample rate? that might be the cause...

[jeffgrant]

Originally posted by Jack Olsen
One question on this line, though. Why do data loggers generate such high lateral G numbers? I show sustained g's in the 1.4 range, maxing out at about 1.6. Common sense tells me my car isn't generating 1.6 lateral g's. I know how to recalibrate the accelerometers, but that won't change the lateral g inflation that all the different mainstream units seem to share.

I think it's due to where the unit is mounted. I've played with the location of where I install my unit (Traqmate), and it seems to make a bit of a difference. Not sure if it's just relative to the CofG or how level it is, but it seems to have an effect of some sort.

Assuming I'm not riding on the same short bus, I would think that a unit mounted on the inside radius of the turn would tend to have less G's measured than if it were mounted on the outside radius of the turn. Differences with front-to-aft mounting could also come into effect, I'd imagine, but nowhere near as much.

With my original placement (basically right of center in the front trunk), I found that my G's at Seattle (left-hand turns) were substantially higher than my G's at Portland (right-hand turns). It seems that my left hand turns were being recorded higher.

I've been meaning to go out to a skid-pad (or parking lot) and test the measurements from both directions (left and right turns) and see if they measure "off" equally for both, or if they're biased to one direction over the other.

I've also been tempted to mount the unit 90 degrees to normal, just to see if it's a difference between the lateral and longitudinal accellerometers. My bet is that the same "errors" would be shown.

[Jack Olsen]

Well, differences left to right can be diagnosed more easily. Just slide the unit out of its harness and rotate it 180 degrees along all three axes. I did it with mine and got different readings, which led to me learning about resetting the accelerometers.

But the artificially high numbers come from all the big brands of loggers, as far as I can tell. We're all getting numbers with our production cars that are better than Le Mans GT cars that cost 300K and have full undertray aero.

[jeffgrant]

It could just come down to the cheapness of the parts... after all, the typical 2-axis g-meter found in these systems are only about $25, 3-axis are about $45, and they probably have issues with sustained, accurate readings. Maybe it over-reads the initial load, and then settles down to something reasonable, etc.

Data is only as good as the sensor.

[randywebb]

set it to avg. over some time period -- 0.1 to 0.4 sec

I think it it s sample rate issue. Your instantaneous G may well be that high.