Sorry for how long this reply got. I post it to the list
and remind all that the <Delete> key is usually right along
side your <Insert> and <Home> keys. ;-)
Ron Katona wrote:
> Thanks for the note. Actually, .15 was on the conservative side if
> anything. Many of my runs this year showed .20 or greater differences
> and always left greater than right.
I'm a bit confused by this though. I thought you had
indicated before that adjusting the map always required you
to put in "more right" on the adjustments screen. But the
maximum adjustment available there is 0.10. It seems this
should never be enough then. Does that match your
experience?
This part isn't that crucial, btw, I'm just trying to
understand the best way that I can.
> I had to replace a damaged Palm
> Pilot this year and the problem remained with the new one. Same for
> installing new batteries in the power supply and recalibrating. The
> mounting is probably within 1 degree of parallel with the axis of the
> car, so we can rule that out too.
>
> Making a fake run on my desktop with the G-Cube just tilted up on each
> side for a few seconds yeilds almost perfect 1.0 g readings on each
> axis! However, I also went out to a parking lot and did skidpad circles
> in each direction with the G-Cube in the correct direction for one run
> and turned 180 degrees for another and got less ideal results.
When you say turned 180 degrees, do you mean that the G-Cube
was upside down, or pointing backwards?
With the
> cube mounted in the normal direction, the readings were within a few
> 10ths of a G left to right for the first time this year!
Help me to understand here...when you typed 10ths of a G you
meant 100ths, right?
Were these readings at/near your normal maximums? Were you
in racing conditions as much as possible...passengers,
tires, swaybar/shock settings?
This is curious that the problem disappeared on this simple
test done in the "normal" way. I'm searching for
differences to justify why the problem went away...but the
answer could be as simple as that the test failed to catch
the problem because the left and right runs were not
consistent enough to each other to show the problem. The
tough part here is that consistency. More on that below...
> However, the
> data displayed on the GEEZ software for that run were very strange
> looking - kind of a sawtooth effect like the sampling rate was very low.
> It also looks like the cube got confused as I tuned the car around very
> slowly in a tight 180 to reverse direction. Turning the cube around
> backwards yielded smooth looking data, but the sustained readings showed
> .25 G higher in one direction... to the LEFT!
I'm concerned about the repeatability of the test. Your
normal test didn't show the problem, and clearly it should
have, regardless of where the problem is. In other words,
whether it's the car, the cube, or anything else, the
problem should have re-occurred IF the test was done in a
repeatable way.
Driving in a completely repeatable way is always difficult
of course. But since you said that you've had the problem
on pretty much 100% of your runs I didn't expect the problem
to disappear in testing.
Here's what I would propose.
This test method is designed to try to catch the G-Cube
problem, or to rule it out. Ideally the test would be done
on a marked skidpad so that you can follow the exact same
radius, and pay attention to the speedometer as you do it.
This would give us a very repeatable drive.
Test 1: Normal mode, Calibrate the G-Cube normally. Pick a
maximum speed that is up there at the limit, but not
difficult to keep the car sustained at. In other words, one
where you have as much as possible constant throttle and
hand position. Generally to run a skidpad at perfectly
sustained lateral g's you should steer with your feet,
instead of your hands. As the car gets closer to the cones
of the skidpad give it just slightly more gas to move it
out. As the car begins to move out release just a tiny bit
of gas. This allows you to keep your hands as still as
possible and greatly minimizes changes in the lateral
g-level from driver inputs. Before you begin recording you
might do a lap or two to get your speed and method down.
Begin your recording from 0mph and pick up to your
pre-determined speed. Hold it there. Repeat going the
other direction. Make sure you hit the same pre-determined
speed. Assuming the G-Cube is in the center of the vehicle,
and that you were the same distance to the cones and the
same speed, the G levels should be extremely close. Any
difference would have to be either the G-Cube, or vehicle
lean differences between left and right. So we've been able
to isolate our potential problem down to only two choices
here. And given that neither of us think it's the leaning
of the vehicle, more than likely we are isolated to one.
Now if they are the substantially the same readings left and
right I am inclined to say that the G-Cube is fine, and that
something different happens when you are not in such a
controlled environment. If the G levels do differ
significantly, and maybe even if they don't, I would proceed
on to Test 2. BTW, I definitely think 0.10g difference is
significant).
Test 2: Upside down mode. This requires a new calibration
to work properly. When you do the new calibration the ONLY
side that we'll calibrate differently will be the TOP side.
In this one setting instead of calibrating with the Top side
up, we'll calibrate with the G-Cube upside down. That's the
only difference. When you Calibrate Left, we'll still have
the normal Left side up. Same for all of the other sides.
Now install the G-Cube just as you did on Test1, but Upside
down. The arrow should still be facing forward, and the
cord should still be facing backwards. Now do your two
directional tests exactly like you did in Test 1...same
speed, same steady method of steering with the gas pedal
rather than the hands. Again, the key is to be sure that
the speed you are hitting is exactly the same for all of the
test runs, and that you are exactly the same distance to the
cones.
When you look at the Test1 data it will appear normal, while
the Test2 data will be mirror-image to what you drove...your
left circles will go right and your right circles will go
left. If by following this repeatable testing procedure you
were able to produce high readings on the Left for the runs
as displayed and low readings on the right, on both tests,
then clearly the problem is in the G-Cube. If you ran the
runs the exact same way and had consistent readings between
all four (Test 1 Left, Test 1 Right, Test 2 Left, and Test 2
Right) then there no problems with the Cube. If you see a
pattern like Test 1 Left and Test 1 Right are both high
while Test 1 Right and Test 2 Left are both low, we are back
to a lean problem (not likely). And if you get three the
same and one different, then we have good evidence that the
test was not consistently executed.
I know this is a lot of trouble, and I don't mean to put you
through a lot of hassle. However the controlled test you
did produced a result of even side to side. The second test
showed a higher level problem than normal. Both are
different from your normal problem by quite a bit, one high
and one low. So currently we have your symptoms from the
year not reproduced in either of your tests, but I suspect
even that the tests weren't done in the best of test
conditions.
> You'll have to trust me when I say my car corners flat as a pancake (500
> lb springs on an M3) and doesn't turn one direction better than the
> other. Sam Strano just drove my car and certainly he would have picked
> up on that. ;-)
I definitely doubt that lean is to blame. Regarding
noticing differences in left and right, I think it's a rare
person who can do that. I liken it to the musical abilities
of "perfect pitch" and "relative pitch". Pretty much all
musicians have excellent "relative pitch". A person with
relative pitch can tell you what an "A" should sound like if
you play for him a "C". However, if you just walk up to him
on the street and ask him for an "A", he'll be lost because
he has no point of reference. I've known only one musician
in my life who had perfect pitch. Her daughter has it too!
It's a weird and exceedingly rare gift. "Perfect G's", if I
can coin that term is even more rare. In fact, I think that
there are lots of great drivers who exhibit only mediocre
"relative g's". I believe that driving isn't a matter of
knowing what 1.2g's feels like, and putting the car there,
but rather a matter of knowing what behaviors the car
exhibits as it approaches it's limit, and putting the car
there. If that limit is lower on one side than the other,
but the behavior is the same, we just automatically adjust.
But hey, that's a topic for another thread. (And one that I
find fascinating. Some day I want to test some theories
down this line to see if we can in fact reproduce a g-level
without having feedback from the vehicle to influence us.)
> My ending speeds also always way high. In full stop
> events, I often see 30-40 mph speeds after the braking in the stop
> garage where I know I came to a full stop.
This is usually indicative of lots of oversteer, especially
at corner exits. For me it's the normal adjustment to be
too high on the end speed and have to tone it down, though
usually not this far. I know that some drivers have
reported that their adjustment has changed as their
smoothness gets better. In short, not necessarily a
problem.
> I've also had a difficult
> time with the autostop feature ending runs prematurely even using a 3
> second .2 G threshold. I'm pretty baffled as well, but I have the
> feeling something's not quite kosher with my setup.
Again, I don't know if you typed that the way you meant it,
but if you did, then yes, you'll have premature stops with
that setting. The Palm only checks your longitudinal G's
for AutoStop (the laptop checks both lateral and
longitudinal, so it won't be so easily fooled). On the Palm
then if you run a steady state turn with very little
throttle or braking changes, such as our test runs above,
you'll have the machine shut off on you. To make your
AutoStop less sensitive, try going down in settings to
0.05g.
> If it's not the car
> or the Palm, it's got to be the G-Cube or power supply unless I'm
> somehow calibrating this thing all wrong... but the directions make it
> pretty clear how to properly do that. Oh well, let me know if you want
> to look at some of this data. Thanks,
Well, we found plenty to discuss above. I don't know where
we'll find the problem. Oh, and more about the adjustments
for "more right". Usually when you need this setting you'll
find that your straights are not going straight, but rather
bending left all the time. This indicates that the
Calibration that occurs at the start of your run (or the
start of the AutoStart session) occurred on tilted ground,
throwing off your "center point". The adjustment setting
works to move the center point back to middle. However,
that adjustment doesn't cause the readings to be altered to
include the "corrected" values.
Hey, call me, Ron, and let's talk about whether all of this
testing is a good idea or not. I need to weigh the
possibility of just replacing the cube for you with creating
a precedent that wouldn't be reasonable. I don't mind
replacing the Cube for free if it's bad. Yeah, I know you
bought it over a year ago, but that doesn't really matter.
But I don't want to start a trend of replacing Cubes that
aren't bad. You see my concern I'm sure. But most
importantly, I want you to be happy.
--Byron
800-775-9511
|