Date: Sun, 29 Nov 1998 12:21:22 -0600
Reply-To: Blue Eyes <lvlearn@MCI2000.COM>
Sender: Vanagon Mailing List <vanagon@vanagon.com>
From: Blue Eyes <lvlearn@MCI2000.COM>
Organization: Vexation Computer
Subject: Re: Vanagon total drag and mechanical losses
Content-type: text/plain; charset=us-ascii
Martin suggested using a video camera to record a vehicle's speedometer while
allowing it to coast down from a fast highway speed. What an excellent idea.
Love it. But I see limits caused by possible data collection errors: Vehicle
weight, speedometer lag, and accurate reading of the speedometer in the frame
images. This last objection becomes trivial if enough data points are tracked
because the pattern would imply incremental values. Camera position should
approximate the position from which the calibration readings were taken or
that would bias results. Put your head against the driver's window, hold the
speedometer on your right shoulder with the macro lens position selected, and
start the coast down test. Applying theory to practice is often tedious.
On any long Interstate highway trip, it's a vary straight forward procedure to
create a speedometer conversion table for decoding those unreliable indicated
speeds into actual speeds. I have found the toughest part of speedometer
calibration to be holding indicated speed constant. So I find cruise control
equipped vehicles best for this task. Don't even bother trying to do this
where the road isn't a constant inclination, which generally means level.
Going across Nebraska, there are long stretches where you could trust the
continuous incline or decline, depending on your direction, but that's
atypical of inclines. Incline variations yield speed variations, so don't
trust them.
Even with a cruise control equipped vehicle, I have recorded too much mile to
mile timing variation to be willing to trust distance increments as short as a
mile. I prefer 5 mile increments, but if I'm on a day long drive, I'll go to
10 mile legs to record the minutes and seconds. I observe mile marker posts
as their image passes the windshield edge or the wing vent, depending on the
vehicle's design. A steep angle compared to the vehicle's direction gives
more accurate comparisons between observations. Just hold your driving
position and turn your head and click your stop watch.
To be really confident that the number of video tape frames per second relates
to seconds as we expect, I would attach one of those cheap, pulse movement,
analog display digital clocks to the dash next to the speedometer so it would
share the video taped frame image with the speedometer. This would give a 2nd
indicator of one second time increments.
For vehicle road weight, I would fill the fuel tank and, if the test vehicle
is a Westy, note how much water is in the potable water tank, since 13 gallons
is about 105 pounds. Then stop at highway weigh stations. I've asked them to
weigh me many times and found them to be very cooperative. After all, if I
want to repeat future testing, knowing the vehicle weight depends on knowing
these tank levels or having it tested again and again.
Now I've got to introduce a couple real world problems I see with Martin's
excellent proposal. The first is the accuracy of the scale measurements.
These DOT highway scales are used to catch big trucks which are running over
their license weight limit. I don't know how much dirt and mud and junk
accounts for variations I've encountered, but they exist, and they are
significant for a comparatively light vehicle like our Vanagons and Westys.
On one trip from Iowa to Connecticut a couple years ago, just for the fun of
it (so I'm weird), I stopped at every open DOT road scale and asked them to
weigh my Mazda diesel truck. Even after compensating for fuel weight between
each scale weighing (more fun), I recorded over 100 pounds variation between
different DOT road scale estimates! I used to call them measures, but now I
call them estimates because they vary by so much. So my first concern with
Martin's procedure is getting a true weight value for plugging into our
calculation.
My second concern is known as speedometer lag. One of my long time pals has
worked for the automotive press for decades and was the technical editor for
one of the most popular US auto magazines. I asked about the use of their
expensive "fifth wheel" calibrated output device that they clamped to the back
of test vehicles. It was a lot like a motorcycle tire with a hall effect
transistor counting pulses caused by moving magnets which showed a certain
angle of rotation by the tire. By marking a spot on the tire and starting
spot on the ground, you can count off 100 rotations and measure that distance,
so you know how far one tire rotation is with good accuracy. Now all you need
is a digital timer and you can start recording accelerations, decelerations
and absolute speeds with a high degree of confidence. You can smoke a test
vehicles tires and show 40 mph on it's speedometer, but that 5th wheel just
records pulses and time increments, so it's unaffected by those inaccuracies.
I was told they had learned that speedometers typically lag toward their
previous value on both acceleration and deceleration readings. So you can't
trust them until speed has stabilized. But we want to measure speed in flux,
so we can't trust the speedometer readings. Grrrr.
An 8 to 19 year old Vanagon's speedometer is to be trusted to accurately track
even it's recalibrated to show constant speed displays for our variable speed
testing? Suppose it's showing a 2 mph lag? Of course lag would be a function
of speed. Oh more fun, . . . NOT! That would displace our entire chart by
that implied horsepower requirement difference, and at 75 mph, where the coast
down deceleration would be high, a 2 mph speed difference will be several
horsepower to a tall Westy.
I think using the video tape technique will be fun and instructive, but I
still want at least one data point which I KNOW is accurate. Since I don't
trust the single road scale weight estimates, and I expect speedometer lag
would have some biasing effect, I want one data point I can trust. Just one
very accurate 60 second per mile long cable pull tension would give us one
reliable data point by which we could adjust the values derived from timed
coast down testing. If you know the speed is 60 seconds per mile and you know
the long tow cable tension, you can know the horsepower without any other
assumptions, period. That I would trust.
I didn't know that towing vehicles with ropes was still legal in any state as
Martin seemed to imply. I hope that he's right I was wrong. Does anyone know
of a state where it's legal? In my youth, I once destroyed a motor on the
highway (don't ask, it has something to do with exceeding design limits with
my modifications and testing) and called a friend who came with a long chain
and a long pipe. He fed the chain through the pipe and attached the vehicles
together. I steered the towed vehicle for about 200 miles over the twisty
roads back home that way and it worked wonderfully, both in pulling and
braking. Just one idea for your bag of tricks.
Also, could someone comment on how long the tow cable should be? I know 10
feet is way too short. I know 1000 feet is unnecessarily long. But I don't
have a good feel about how far out you have to go to approximate the long
value in decaying induced draft effects. I'll bet there's a body of
literature about that somewhere from high speed racers testing.
John
| 
