claykos

That is pretty impressive - 1:20 is certainly getting around! However, the spikes you are seeing are due to the bumps on the track because of the way the accelerometer chips work. You can tell because of the very high frequency of the oscillations between ~1G and ~2G. One of the tracks we run on alot is a Roval with a big bump as you exit the banking and go into the infield. I always see spikes of 1.8-2G over this bump, but in "real" cornering it is more like 1.5 or so. I would say you are seeing more like ~1.7-1.8 as far as "real" sustained g's go. On the G2X I think the "GPS G's" are probably more accurate. It uses your speed and radius of curvature from the GPS data to calculate lateral acceleration. It is not effected by bumps/etc the same way as the accelerometers.

Johninrsf

Man, you just gave me some info I never knew about the Racepak. Thanks. I didn't know there were two ways they calculated Gs (prolly cause I've been fixated on other data) But, it's very interesting. What's funny is when I overlay GPS lat Gs over accelerometer lat Gs, I actually get similar numbers with the GPS --up to 2.30 in turn #5 and 2.13 at the exit of #9. To further confound, the largest spike in either case is with the GPS at the entry to #4 (2.69). Go figure!
Here's the graph.
http://picasaweb.google.com/Johninrs...27902932624834
P.S. I don't want to infer in any way that the 1:20.5 was my time --it wasn't. It was my car, but Kevin Roush was driving on a test day during which we were trying larger tires (320s and 280s). My best time of 1:22.4 is still on smaller tires (280s and 230s). Maybe some day............

claykos

I'm sure the GPS is also influenced by the bumps etc, but it's not to the degree of the accelerometers. You can see that in general the GPS data is smoother. When I'm looking at G-force data I always just kind of take the "average" over those oscillations. There is definitely some noise in both signals.

1:22 is still damn quick!

PS: I sent you a PM regarding your car...

mark kibort

Im wondering if I should blow off the SCCA stuff this year and run with you guys! sound like fun! Hey, are you making it down to the double regional SCCA race at Laguna again this season?

mk

mark kibort

1:22 is darn quick! But remember, Kevin also weighs about 115lbs dripping wet too!

Mk

sjfehr

Many of the accelerometers used today are MEMS-based. (MEMS is short for microelectrical mechanical systems.) It's essentially microscopic mechanical components machined into microchips using the same tools and similar techniques as used to make microprocessors. Here's a photo the MEMS accelerometer used in the iPhone:


So, anyhow, if you cracked the cover, you'd see 3 accelerometers on this chip, X, Y and Z. Most of these accelerometers are roughly about the size of the head of a pin, but the details are too small to see with the naked eye. As they're so small, they have practically no inertia, and conventional measuring techniques don't work.

To give you an idea of the scale, here are a few micro gears with a spider mite:


There are some piezoelectric materials that can be used to directly measure the deflection, but resolution is low. Most accelerometers today work by "doping" the moving bits to turn them from plain silicon into semiconductors just like they would for microchips, which turns them into capacitors. To increased the capacitance, instead of making them into nice little squares, they make what's called comb drives, with interspersed fingers.


Note: the double-indented rectangles are the only thing attached to the silicon base. (It's all solid silicon, so it's immensely strong and can measure hundreds of gs if that's the application.)

So what they do is set these comb drives vibrating at thousands or tens of thousands of times per second, with a feedback loop that maintains the vibration frequency regardless of what you do to them. Semiconductor circuits integrated into the chip monitor this feedback loop, which changes as the chip accelerates through space. It's this signal that's sent out of the chip as data. At least for X and Y, but not for the Z axis. For the Z axis, they used a cantilevered weight that instead of having interspersed fingers is pretty much just a flat plate that vibrates up and down. The feedback circuit's pretty much the same.

DarinB

Mark,
Yes, it is very fun. I don't know how your car would class, though. You would probably have to do GT with some really fast cars. The other positive note is that NASA has added the German Touring Series classes (GTS) on the west coast. So far, adding this class has only been an extra $50 for the weekend. They are trying to spark some interest in CA. Jerry Kunzman said it will probably eventually be $150 to add. For us, it sure is money well spent since we have such a long way to travel. It makes for 4 practices, 4 qualifying sessions, and 4 races in 2 days.
We do plan on doing the Laguna double regional at the end of the year again. That is one of the highlights of the season for us.
Darin

kurt M

tire shape and camber angles.

sjfer, Thanks for posting, cool stuff.

924RACR

Not exactly what I meant, but that's another discussion...

prg

sjfer, Thanks for posting, cool stuff. +1

2BWise

It has a lot to do with tire construction, and further with suspension kinematics.

sjfehr

I tried using that free g-force software on my cellphone today at the local autocross, but it was a bust. Apparently, the application gets hosed up whenever the phone turns off, and needs to be manually exited and restarted, which I didn't do. So, it recorded 2913 data points, and each and every one of them is a zero

Oh well, lesson learned. Maybe next week! At least my new video setup worked, and I finally got some in-car video

kurt M

Yes, my post was a bit over simplified.

brendo

tx for the information on the chips. i have a feeling i could get a phd in it.

sjfehr

Funny you should mention that.... I was in college studying MEMS about 12 years ago when the first commercial MEMS accelerometers hit the market, and there have been quite a few phds in the subject since then, too! The early ones were only accurate to about +/- 50gs, (yes, 50 as in fifty- they sucked!) and found their first commercial application as airbag sensors. Early MEMS gyroscopes (round comb drives, same principle as above) measured drift not in seconds per hour, but in degrees per second. They've come a long way!

How they're manufactured is interesting, too. There are various different methods, but one of the more common one is selective chemical etching. A bare piece of silicon is coated with a thin layer of photo-sensitive polymer, and then exposed to ultraviolet light shone through a finely etched mask to harden the desired design onto the polymer. The soft unexposed polymer is then dissolved in isopropyl alcohol, leaving a design in hardened plastic on the surface. If this was a microchip, they might implant boron or phosphorous ions to turn it into a semiconductor, but for 3D mechanical devices, they instead chemically etch into the silicon to create the 3D shapes. To do this, the chip is dipped in acid (Nitric Acid, for example) which etches away all the exposed silicon, but is blocked by the polymer, and relatively non-reactive with other materials on the chip like silicon nitride and silicon dioxide. It might sound slow, but the distances are so small, that only a few micrometers worth of material are etched away, so the bath might only take a few minutes. Timing is crucial, as it's too small to monitor what's going on.

There are all sorts of other tricks used I won't really go into like orienting crystal planes, sputtering, chemical vapor deposition, etc, but I always found it all fascinating. Unfortunately, my career took me off in another direction, but hey, what're you going to do