001: Enlightenment  [HSD cutaway]

As I continued reading about the engineering that went into Toyota's hybrid
effort, many of my preconceived notions were turned completely on their ear,
especially where battery technology and engine management are concerned.
Every little feature I read about the Prius seemed to have undergone some
really brilliant thinking in its development.  It was clear that there were
a lot of good minds in the customer base, too.  Several sites and forums
pointed to obscure, tech-heavy but very excellent papers that explained many
of the principles.  Forum archives and various owner sites offered plenty
of reverse-engineered hypotheses about Toyota's hybrid designs, and I began
finding myself quite in awe of the things that modern automotive engineers
need to think about.  It was very informative to look back ten or fifteen
years in various patents and see how some of the design thinking had evolved,
and how many of those ideas had obviously been adopted in the Prius too.
It's often hard to stay awake through something as dry as patent text, but
in this case it was worth it.

The "Hybrid Synergy" drivetrain is actually beautifully simple from a
mechanical standpoint.  Nothing ever physically shifts -- the two motors and
gas engine are linked through a planetary system that always remains meshed.
Everything else about power allocation is done electronically, by simply
playing a torque-balance game between engine demand and motor loads.  Even
"Neutral" is just an electrical state -- no mechanical disengagement of
anything.  And the motors are fairly simple 3-phase machines with permanent
magnet rotors -- no moving contacts or screwing around with induction and
phase slip, and exactly the right thing for producing plenty of regeneration
power.  Almost like big stepper motors with tight position feedback.

Another major change in my thinking came concerning batteries.  From the
old school viewpoint of ni-cads that often had to be "zapped" back into life
to remove that infamous memory effect, Toyota's approach to battery management
seemed completely strange.  "Deep-cycle" capacity isn't even mentioned --
instead, it turns out that modern nickel-metal-hydride chemistry will provide
a much longer useful service lifetime if it is *never* deep-cycled or
overcharged.  Thus, the battery management deliberately ignores a certain
amount of capacity and keeps the state of charge between 40 and 80 percent,
centering around 60.  This keeps the battery chemistry in a very stable
region, but still allows room for a useful amount of either driving energy
or regeneration overhead at almost all times.  The electronics surrounding
the battery are designed to keep *very* close tabs on charge state and
communicate it accurately to the rest of the system, unlike a lot of cell
phone and laptop packs that seem to rapidly drift away from reality and
kill their own cell chemistry in a couple of years despite periodic
"recalibration" attempts.

What this means is that the stock battery pack is sized so that that's only
good for 2 or 3 very gentle miles on a "full", i.e. 80%, charge.  This isn't
going to golf-cart you around through a full day's shopping, even if it did
allow itself to run all the way down -- the Prius is not an EV, although
many people like to think of it as one.  The battery is for temporary energy
storage, not primary propulsion.  But for moving between parking spaces or
something, running solely under electric power is absolutely the right thing
to do -- avoid the rich-mixture hit of cold engine startup if you don't need
it.  The car's brain can't predict where you're *going*, and sometimes needs a
little help making the decision.  Very early in the study process I read about
the "EV button" hack that several people added to their cars, for which it
turns out that the functionality is already there and all that's needed is a
switch to momentarily ground one of the pins in the controller and enable it.
That immediately went on the list of pending mods, of course, assuming that
the functionality would also be present in other hybrid vehicles.

If a cold engine startup does happen, the effects are mitigated in a really
entertaining fashion -- at the prior shutdown, the hot coolant is pumped off
to a thermos-style holding tank that can stay hot for something like three
days, and then all pumped back in around the cylinders just before starting
up.  Like a built-in block heater.  Certainly not enough to bring a cold
block completely back up to running temperature, but it still gives a head
start on heating, to reduce startup friction and fuel consumption and make
emissions reach a low level that much sooner.  Cold start engine management
emphasizes getting the catalytics warm as soon as possible so they can
begin working right.

And even more thermal efficiency comes from the Atkinson/Miller cycle
engine, which uses a variable intake-cam offset hack to allow different
compression and expansion ratios.  It's very similar to a standard Otto
cycle engine but winds up extracting more energy from each charge of fuel,
at the expense of a little bit of torque.  Toyota's hybrid-adapted engine
can set the valve timing on the fly under computer control with the VVT
system, and thus best tailor the compression/expansion to demanded output.
And they even decided to slightly offset piston axis from the crank axis
to allow a more "straight-down" shot on the power stroke, reducing sliding
friction and piston-skirt wear.

Spark generation is done directly at the plug, using an integrated "igniter"
containing the coil and switching electronics, mounted directly over the plug.
A low-voltage pulse triggers it.  Thus, no spark plug wires, external coils,
distributor, etc.  The igniters sit down in deep wells in the head cover,
probably reducing EMI as well.  Coil-per-plug has been around for a while
but seems to be appearing in many more engines these days, because really,
it just makes sense.  Toyota's approach also utilizes normally wasted space
down in the plug holes to contain the electronics.

I had never realized before digging into all this that an engine's efficiency
*increases* under load.  But here I was, facing the possibility that my usual
method of babying a car might actually *reduce* gas mileage, and that better
gain might have come from pushing it a little harder!  Another old notion
scrapped, and with that came the realization that traditional cars all have
engines sized to handle *peak* demands which actually don't occur that often
in normal driving and certainly not in steady-state cruising.  Thus, they
all have an oversized engine loafing along at partial loads most of the
time, but still needing to crank a significant volume of air and fuel through
themselves to stay running.  Even if you drive like a total butthead, you
can't be placing a high demand on an engine *all* the time.  That's one of
the main reasons for crappy mileage in most vehicles -- overpowered, yet
having to accomodate widely varying power demands.  Fewer worries with a
hybrid system, though, since the engine can be much smaller and engine
management and electronic throttle control should try to keep it nominally
loaded when it's running and much closer to its efficiency sweet-spot, or
eliminate the entire problem by shutting down and consuming no fuel at all.

Of course having a hybrid system eliminates the alternator and starter motor,
two frequent sources of automotive headache, and moves all that functionality
into a protected, lubricated, and cooled environment.  No more G-r-r-r-r-r
and klunky solenoids trying to chip teeth off the flywheel -- instead, the
engine is spun up to 1000 RPM or so and confirmed to be turning *before* any
gas and spark are applied.  This eliminates the traditional method of using
combustion [and associated low-RPM lugging, knocking, and connecting-rod
stress] to bring the engine speed up the rest of the way.  A brief burst of
battery current, and suddenly it's magically running.  This happens whether
the vehicle is moving or sitting still.  And on shutdown, the motors are
actually used to reposition the engine in a way that optimizes forces and
timing during the next restart.

Three relays are used in the main battery pack -- one on each terminal,
and a third to connect the positive side through a resistor long enough to
charge the filter capacitors in the inverter more slowly and avoid high
inrush currents before bringing in the main contactor.  Standard technique
for electronic devices with large capacitors such as power amps, but amusing
to now find in a car.  A fused safety plug connects both halves of the
battery pack together in the middle, disabling the whole deal if it's pulled
out.  Toyota refers to this often in the emergency-response info hung out on
their website, making it very clear how to completely disable the high
voltage side of the hybrid system when needed.

Toyota evidently had to take a bunch of lessons from the Japanese bullet-train
control systems and design their own switching transistors from scratch, to
achieve sufficient current-handling capacity and breakdown resistance.
They're mounted in the inverter module on a cold-plate that gets its own
coolant loop along with the motors themselves.  This makes things a little
harder to work on since that coolant loop must be drained first, but it
gives plenty of heat-removal capacity to protect the inverter electronics.

Braking is done as regeneratively as possible, resorting to physical brakes
only on high stopping demands or at very low speeds when the motors aren't
turning fast enough to produce useful zoobs.  [Yes, that's a technical
term..] There are lots of electronics involved in modern braking already,
when systems like ABS and skid-control and traction control are taken into
account.  So we already have hydraulics passing through many little solenoid
valves under computer control.  With the newer hybrid system, Toyota simply
decided to merge *all* of that functionality into one place and went with
a fully electronic braking system in which each wheel is independently
controlled and front/rear proportioning, ABS, etc just happen as a natural
consequence of how the braking computer responds.  Including to apply little
or no physical braking at all if gentle stopping is requested and things are
stable and within comfortable margins -- at that point it can *all* be
regenerative using the main electric motor, and the braking controller keeps
itself out of the picture until really needed.  If the electronic braking fails
somehow, there is still a physical hydraulic path between the foot-controlled
master cylinder and the front wheels, giving a reasonable safety backup and
allowing the brakes to function when the car is not powered.

The various elements of control software appears to take *rate of change* of
inputs into account.  Some of this is described in the patents dealing with
the brake-by-wire system.  Sudden hard braking, e.g. quick pedal travel,
causes different response than slow braking, even if the same final braking
amount requested is the same.  Similarly, a sudden step input to the
accelerator yields a temporarily higher surge of engine output power, if
it's running, assuming that the driver thought it was needed.  Or it
encourages the system to start the engine in the first place.  This allows
a *little* more explicit control over run/stop via quick blips on the
accelerator.  I have always been a very "predictive" driver and thought
down to the level of d(input)/dt with regard to throttle and brake, based
on terrain and surrounding conditions.  So I'm all in favor of a control
system that is also aware of such things and responds in a reasonable manner.

In general, the more I read the more it was evident that Toyota had really
made a strong bid for reliability, reduced stress, and carefully staged
startup in many design areas.  Which also happens to totally match my own
approach to handling gear and using technology, because I understand many
of the real-life analog effects and how to minimize the deleterious side of
them.  Anyone who's heard my lamp-preheat or hard-drive-warmup rants can
attest to this.  The more I kept discovering the little stuff like this
that they'd thought of, the more it made me want to throw money at them.

Some people are uncomfortable with the idea that a car engine can randomly
start and stop itself when it sees fit.  But do you specifically turn the
furnace in your house on and off?  No, you set the thermostat and forget
about it, and it fires up and shuts down automatically when needed.  Often
that's under computer control too, especially in larger complex HVAC systems.
Heck, any other car has a high number of automated systems too -- radiator
fans, for example.  Did I begrudge the old wagon's automatic decision to spin
them when the coolant was too hot?  No, but I *would* let them finish cycling
before shutting down the engine -- but that's just me and my usual hyper-
awareness of what's going on under the hood.  Numerous factors affect a
hybrid system's decision to run the engine -- not just propulsion demand,
but batteries needing to be charged, a call for cabin heat or cooling, and
sometimes just to keep itself and the emissions system properly warm.  In
general it's making the right choices, although there are still times when
the human can help beneficially sway the decision.

Now, as some people are quick to ask, what about computer failure?  Think
about how many lines of code you're already running to read this, or to
make a call on your cellphone, use a camera, or in fact to drive most cars
nowadays.  Concern over the reliability of dedicated processing devices goes
down as their ubiquity rises.  Let's face it, almost *any* car on the road
today has at least one reasonably complex electronic control unit [ECU] of
some sort, without which the engine doesn't run, period.  The industry has
gotten way beyond those days of fire-breathing carburetors and bimetallic
choke actuators and centrifugal timing advance and brute-force points and
spark coils, however brilliant some of the work in that arena may have been
at the time.  Embedded control is everywhere, and increasingly is the *least*
likely component of a system to break.  Many of those early growing pains
of semiconductor-based automotive control are over -- they've gotten much
better about dealing with device and power stability, input and thermal
protection, and even software design.  Now that it's a well-known and fairly
proven approach for engines themselves, designers are going after the larger
concepts of how the entire powertrain works together.  Better efficiency
tweaks can be done entirely in software, as well as early detection of
potential problems ...  and in general it *is* working out for the better.
Yeah, I thought *I'd* never be saying that.  The downside, of course, is that
when components do fail, they're wicked expensive to replace and for the most
part infeasible to D-I-Y fix.

Exceptions to robustness may arise if those means of control are allowed
to converse arbitrarily with the outside world.  Here is a rambly but
important digression about computer security and reliability in actively
hostile environments, which is not only a major concern on the desktop
but also a concern in the embedded-control world -- whether anyone's
really aware of it or not.

Part of my research had also more or less become a crash course to update my
knowlege of modern engine-control systems, and while computers are rapidly
numbering the days of the traditional weekend car mechanic, one only needs an
OBD-II scan tool and the newer control modules could pretty much tell you the
cause of any given problem right up front.  People are way too scared of that
"check engine" light, but all it takes is the equivalent of plugging a serial
terminal into the device in question and issuing a few commands to figure out
what's going on.  Having maintained routers and firewalls for several years, I
can certainly relate to that -- as long as the *protocol* is well-documented.
The cars with parts simple enough to get apart and scrape the gunk out of with
a Swiss Army knife or even [in the words of an old buddy] a plastic fork from
McDonalds to get running right again were already fading into fond memory.
The new plastic fork is something almost as ubiquitous nowadays -- a laptop
or a PDA.  The old carburetor-tweaker in me was slowly, grudgingly coming to
grips with this.

Therefore, the other item on the shopping list was an interface for my laptop
to plug into the diagnostic port on a car and pull out the trouble codes and
other info.  In parallel with all the ongoing research I had also dug into
OBD-II somewhat, and discovered that by now there are something like five
different interface "standards" that can appear at that little plug under
the dash and a vast array of parameters and pathologies that can be reported
through it.  *Sigh*.  So much for those golden Saturday afternoons fooling
with the little screws on the side of the carburetor and listening to how
the engine changes.  If I was going to do my own diagnostics and implement
the list of hacks I had already started forming in my mind, I'd need that
modern-day plastic fork -- and not just for a hybrid, but to help me diagnose
any other late-model car including the existing wagon.  The newer Toyotas all
speak Controller Area Network [CAN], which is supposedly slated to be THE
standard for new vehicles by 2008 or so, but vehicles using other protocols
will still be around for some time so it's good to be versatile.  The tool I
settled on was Alex Peper's "CarCode" with a CAN-compatible interface, the
installation instructions for which amounted to "unzip into some directory
and run OBD2.EXE to start it up".  A very nice philosophy -- no hidden DLL
conflicts, no turds in the registry -- could run in the same very cut-down
windoze environment I use for running lights, using generic display and
serial-interface calls without a lot of broken Microsoft fru-fru.  So I
liked this guy's style, at least at first.  More on him later.

If I'd had such a kit six months earlier and some of the newly-refreshed
knowledge I was sponging up about control systems, I could have pinpointed
and possibly fixed the "check engine" light that came on in my parents' car
while I was visiting for the holidays.  But they wound up taking it to the
Toyota dealer, and getting shafted.  Mere posession of a diagnostic tool
doesn't ensure that the person wielding it actually knows what they're up
against, as I explained to them quite after the fact:

	I also happened across descriptions of typical EGR valve
	problems, such as what supposedly went south in your car.  There
	is a strong chance that it was merely due to needing the carbon
	deposits scraped out of the little pipe that connects between exhaust
	and intake -- that's a very common cause of that "insufficient flow"
	sort of problem.  But of course the dealer will seize any opportunity
	to sell you $450 worth of parts you probably didn't need yet.  I was
	rather stunned by those guys' obstinate unwillingness to actually
	*diagnose* what the real problem was, instead of blindly parts-
	swapping it away.  Even with cars the way they are now, if one digs
	in a little, genuine problem causes can still be ferreted out.
	Unfortunately that level of dedication/interest isn't cost-effective
	anymore, but I still think it was a rather telling indictment of
	their ability to think.

Thus, I was totally unafraid of the electrical maze I would inevitably face.
That's what service manuals are for, and once I found that they were available
for a nominal download fee at techinfo.toyota.com, I felt much more confident
about it all.  One thing that the CAN bus or other diagnostic interface
provides is more than one option for adding instrumentation.  Numerous sites
show how people have gone the "pure" digital route, especially on the
Japanese-only prius-hacking sites, designing embedded systems to query the
ECUs and show figures on a small LCD display.  Alternatively, one can read
analog signals straight from various sensors in a more traditional fashion,
or design something with a mix of both methods.  My own forte is more on the
analog side, so that's the way I would generally head.

And whatever form it would take, I really wanted to get started on the rest of
the learning experience.  Somehow.  Perhaps a mild sense of desperation had
set in?  I found email addresses for a few dealers in the group archives, and
started trying to ask some of them directly if they knew *anything* about the
Estima or a US-market equivalent.  Well, dealers are evidently the *last*
people to find out any real info about new or future vehicles, so that was
pretty useless.  They also have about zero influence back up the chain to
Toyota corporate.  I found myself sending *them* URLs to good info resources
because of all the things they didn't know.  One of these guys was relatively
local but never answered the email, so I decided to just try calling him one
afternoon.  That would be Gilles Labelle, at Westboro Toyota.  And it turned
out to be a very enjoyable conversation, because we seemed to be on the same
page about the whole reduced-consumption concept in the first place and he's
got genuine passion about it.  By the end of that call I had agreed to haul
down there and visit him in person.  Not that he knew anything about Estimas
either, but that was rapidly becoming immaterial.

Next
Prev
Index