Inspired by the two threads going on about this, I hooked up
the Autoenginuity and went out for a bit of data-collection
this afternoon. First of all, here's my data -- *all* of it:
->
http://techno-fandom.org/~hobbit/cars/sweet/loafing.csv
for anyone's statistical pleasure. I collected spark advance,
RPM, engine torque [which I believe is calculated by the hybrid
ECU from motor torque, but who knows], injector time, and speed.
The file has a label line at the top. I didn't include LOD
because I know that my non-upgraded ECU reports it bizarrely
and I can more or less see it via my vacuum gauge and/or the
torque figure. Then I went back and re-read both threads
in parallel with composing this just to make sure I had the
whole picture. What else to do on a nice sunny Labor Day
afternoon? I already got my labor in, smoothing down and
painting six big cinderblocks for use as tent weights.
.
First of all, you guys have been averaging about 50 mph for
this experiment, right? As xcel points out, that's hard to
do on most of our interstates without a good excuse; today
I managed to poke along mostly at less than 55 but with all
the returning Labor Day beach traffic and big-ass pickups
and boat trailers tailgating everyone in sight and diving
on and off the ramps, that wasn't easy. But my usual "test
track", I-95 on the north-shore section, has 4 and sometimes
5 lanes, so it's a little more reasonable there. Still, at
60-ish people tend to just pass, but down closer to 50 there's
definitely a lot more stink-eye goin' on.
.
So on a relatively fresh tank I tried both techniques for a
while; about 25 miles of 14-advance and then another longish
leg back to my usual minimum-1700-RPM and WS routine. I
oscillated between maybe 48 mph and 60 mph; possibly a little
faster than others' experiments. After the 14-degree time
I was sitting at 66.5 mpg average. After the pulse-n-WS
routine I was seeing 66.4. This wasn't a totally controlled
terrain -- I made a big loop on the local highways, rather
than trying a rigorous A/B two-round-trips test. I figured
if there was going to be any significant difference, I'd
see it simply by eyeballing the 5-minute bars and watching
the average drift. What I really wanted to determine was
the conditions that y'all have been talking about WRT this
s'posedly new technique. Besides the .CSV data, I did note
a couple of other things.
.
The 14-deg mode appears to be able to hold about 47 mph
sustained on the flat. Not useful for interstates by itself,
but obviously allows for a longer glide-down. That's
probably just the balance of output power vs. air resistance.
Spark timing seems to advance farther while engine RPM is
increasing, and retard a little on spool-down. Idle is 10
deg as Dan noted. What's rather puzzling is a> why there's
a dip in ignition timing like that around that certain
RPM/torque range -- backing off even farther sends it more
*advanced* again until you actually hit idle, and b> why
retarding the ignition at all would have any hope of being
efficient. Perhaps at very light loads, there's so little
mixture in there that it's not worth lighting it off earlier
since it's done burning before the power stroke is finished?
.
There is a little operational plateau in there with respect
to accel position. In that state I usually saw 14.5 degrees,
the iFCD hovering just under 100 MPG, 8 in-Hg of vacuum, and
4.4 or thereabouts ms injector time. So without the laptop,
my indicators of that state are the vac at 8 and the iFCD just
shy of 100 [sans scangauge, I can back off slowly and guess
when it just went off the top]. [Note that injector time is
usually 6 - 7 ms under most pulse conditions, and about 1.2
ms at idle.] The plateau shows up when wiggling my foot a
little through the 14-deg range -- the vacuum and RPM don't
increase or decrease. What does change just a little is
battery current. At lower demand I see about 5 amps of
charge, and at higher demand [just before vac starts to drop
and spark-timing changes] it goes back to 0. If I'm at 60%
SoC already the time that that 5 amps of "background charge"
happens is fairly short, and then the plateau more or less
disappears. That's when battery current [and the mimic] start
bobbling back and forth. If I've held the state long enough
for this to happen, chances are I'm down under 50 mph by then
and it's high time to pulse again. For what it's worth I
rarely saw 13 degrees, and if I did it was almost impossible
to maintain even if my foot was frozen. Must have showed up
briefly on transitions.
.
This is in keeping with how the Prius does a lot of its
engine management -- certain torque ranges are maintained
under varying demand, simply by shunting a little more
engine output to battery-charging instead of between the
MGs when needed. We see this pretty often at low speed,
too, when SoC is under 60%. If I backed off *too* far in
14-degree mode, the whole system just fell right into WS
by itself. So obviously this is how the Prius handles the
next demand region just above that.
.
So I can't see much MPG difference so far between this and
what I've been doing, but I've got the possible advantage
of being able to manage my WS a little more closely using
the battery-current meter and actually keep it close to
zero for long ones or use a little bit of 'lectric to sustain
speed without re-lighting until I need to. My 5-minute bars
wandered around somewhere between the 50 and 75 lines pretty
much the whole time, without any profound differences stemming
from either glide-mode. What still baffles me is how xcel is
pulling 80 MPG segments out of this where everyone else seems
to be working hard for 70. We're not worthy. No, the
aforementioned cinderblocks were *not* in the car.
.
I will note that holding the iMPG somewhere north of 75 was one
of the techniques I was trying two years ago. It was likely
similar to what's going on here. I couldn't see any profound
differences there either; of course I was doing a lot of that
in winter weather, so nothing was likely to hand me 70 mpg
highway runnin' for free. Bottom line that we've all known is
that the Prius ICE is pretty efficient in general, and the air
resistance curve is a bitch. We also know that trying to truly
steady-state at 55 - 60 mph is likely to see an efficiency drop.
.
But what I'm looking for is a higher "crossover point". If
I'm actually trying to get somewhere, it's nice to put a mile
a minute under the wheels and get the occasional thumbs-up from
a passing SUV instead of fists and fingers from all of them.
.
Side note -- my usual "S1 done" indicator is the change in
engine/exhaust note, which is pretty obvious and should
happen about a minute after startup anyways. I also see
propulsion begin to come from the engine instead of just
the battery, which has been causing me to try to take it
really easy on the go-pedal during that first minute to
avoid pulling the pack way down. 10-amp crawl out of my
road, and then pulse/EVB until it's warm enough for S4.
.
_H*
Re: the above, I have (up until now) done this by "S1-limping" up to 12MPH or so, watching the MFD report instantaneous 11, 13, 15... 17... then the sound of the engine changes, MFD flicks up to 25 or so, and S1 is complete (=EV enabled). 'Grats on finding a hard indicator of this.
Linear Mode
