May 2000
crossing
of the
Atlantic by balloon of the Double Eagle II
in
1978, an
entirely
new era of long
dis-
tance balloon
flight
began. Soon the Pa-
cific had been
crossed, then
both oceans
were crossed by hot air balloons. Roziere’s
entered the scene with an
Atlantic cross-
ing
by the first European
team
from Bel-
gium
in
1986
and
the
famous
Chrysler
ChallengeRaceAcross the Atlanticin 1991.
These flights in many ways served as
test flights
for the final challenge, a bal-
loon
flight around
the world. As with
all
test flights, they
also served
as platforms
forrapid advancements in ballooning tech-
nology. One of the advancements born of
this era could be described as ballooning’s
black
box—the Comstock Autopilot.
Unlike the much heralded black boxes
of commercial aviation, those flight data
recorders
that
often
reveal
why
a
plane
has
crashed, ballooning’s
black box
was
designed to keep the aircraft flying and to
do so
reliably. After 20 years of develop-
ment the Comstock Autopilot has proven
it
can do
just
that.
“It
really
began
as
a
purely
recre-
ational project,” says engineer. designer,
inventorand pilot, BruceComstock. “Back
in
the late 1970’s someone else was pro-
ducing what they called a balloon autopi-
lot
that
was
really
nothing
more
than
a
thermostat
device that
didn’t
work
very
well. That got me thinking about building
a real
autopilot
and
I spent
quite a lot of
time
working
on
it
and
a balloon
flight
simulator.
Of
course
I
never
felt
there
would
be
much
market
for
an
autopilot
because
I knew that
people
bought
bal-
loons
for
the
fun
of flying
them,
so they
weren’t likely to spend
a lot of money on
something
designed
to
do
the
flying
for
them.”
After
designing
his
first
unit,
Com-
stock did his own test flights, flying with
the autopilot “until my friends got tired of
it. We’d go out and fly with this thing and
just
sit
there in the basket
with
our arms
wanted
the visceral experience of
flying
the balloon, not just riding in it, so the unit
was
placed on
the shelf.
His first so-called
production
model
was
then
installed
on the
hot air
balloon
that Don Cameron and Jim Howard hoped
to cross the Atlantic in. Unfortunately that
unit
never flew as Cameron and Howard
spent a summer in Newfoundland waiting
for a weather window that never arrived.
Enter a
character named
Steve Fos-
sett.
“Steve
first
called
me
when
he and
Tim
Cole
were
planning
their
Atlantic
flight,” says Comstock. His first reaction
to
this
call
is
now ballooning
history.
“I
didn’t know who this guy was and here he
says
he
plans
to
cross
the
Atlantic
in
a
balloon, but
no
one’s
ever heard
of him
and
I
surely
would have heard
of him or
anyone else anywhere close to having the
experience to cross the Atlantic I thought,
so
I politely
put him off.”
Ti m
Co le
wou ld
i nt erv ene
on
Fossett’s behalf and a Comstock Autopi-
lot
Model
1
was
installed
in
their
R-77
balloon. In
August of 1994
the Model 1,
Fossett and Cole made their first success-
ful ocean crossing. (This was
the first of
several
Comstock
Autopilots
used
by
Fossett.)
Next
came
Fossett’s
call
to
say
he
intended to fly the Pacificsolo and planned
to
use
the
same
autopilot.
“I found
this
news very distressing,” recalls Comstock,
“because all of these units I had built up to
then
I considered
to
be toys.
They
were
analog devices in which all of the values,
like
rates
of
climb,
altitude,
etc.
were
determined by voltages. This proved to be
a very
cumbersome way
to
build
things
and the unit was not terribly reliable. The
Model I could sometimes drift as much as
600
feet
up
or
down
from
its
assigned
altitude.
In
long
distance
balloon
flying
the
first
purpose of an
autopilot
is
to
fly
the
balloon level—very level. If you’re mak-
a long distance balloon
flight what’s
really important is flying level, for a long,
long
time.
The
ratio
of
level
flight
to
ascending and descending is very great on
flights that last several days or weeks. The
autopilot
has
to
do
this
and
must
do
it
reliably,
especially
on
a
solo
flight
be-
cause this provides the pilot his only op-
portunity
for sleep.
If the
autopilot fails
yourflight is probably going to end within
hours because you losetheability to sleep.”
This
fact
was
demonstrated
in
dra-
matic
fashion
on
Kevin
Uliassi’s
most
recent
round-the-world
attempt
when
a
capsule electrical system failure disabled
his
autopilot
while
at
29,000
feet
over
India.
“It
was
an
absolute
panic,”
says
Comstock, “because Kevin had to imme-
diately start flying
the balloon while also
trying to poke around with an ohms meter
to locate and correct the electrical failure.
It
brought
home
to
both
of us
that
if an
autopilot
were to fail a solo flight
would
definitely
end within
a matter of hours.”
If
maintaining
level
flight
and
pro-
viding sleep periods are the primary func-
tions of an autopilot, they are not its only
purposes.
“There’s
not
a
lot
of
thrill,
I
think,” says Comstock, “in
flying one of
these
round-the-world
balloons
when
you’re awake because
you’re sitting
in a
closet.
You
basically
want
to
be
flying
level
but you
don’t
want
to be a slave to
flying the balloon all those hours.” So the
autopilot
serves
to
provide
free time for
even the awake and alert pilot—time that
can be used for any number of other tasks
besides
constantly
actuating
the
burners
on and off.
Fossett went on to fly the Pacific with
the Model I, or “Vegematic” as Tim Cole
had dubbed it because it was packaged in
a Tupperware® vegetable crisper. How-
ever, when Fossett then announced
plans
to
fly
around
the
world,
Comstock
laid
down
the law.
He must
not
fly
with
the
Model I, it simply was not reliable enough.
In
fact
Comstock
told
Fossett
he needed
May 2000
“The reason for two was again reli-
ability,” Comstock
explains.
“Roughly
speaking if you have a unit that has a one
in one hundred chance of failing, if you
have two then the chances of both failing
is about one in ten thousand.”
What resulted was the development
and
testing of
the
first
version of
the
Model II, an autopilot based on a digital
computer that can fly virtually any bal-
loon that operates on a “on-off” cycle.
(There have since been seven major re-
finements to the Model II computer pro-
gram.) Its actions are
based purely on
sensing atmospheric pressure, yet it can
fly a balloon level at 10,000 feet with an
accuracy of plus/minus 10 feet! “It uses
pressure altitudebecause flying at a con-
stantpressureismore likelytokeepyou in
the same
wind pattern than flying at a
constant geometric altitude,” says
Com-
stock.
The
Model
II
is
built around
a
single board,
industrial control,
digital
computer with an operating speed of less
than
10 megahertz,
chosen
not
for
its
speed but for its own reliability.
“Atthe time I selected that computer
there were 20,000 of them that had been
used in harshindustrialenvironments for
years,” says Comstock, “and all the bugs
are out of that computer.” While itsoper-
ating speed has been questioned by some
Comstockremindsthem thatnothinghap-
pensthatfastina balloon,itsimplydoesn’t
need the speed of today’s blindingly fast
500, 600 or 750 megahertz processors.
The
program
that runs
the autopilot is
writteninLanguage C. The algorithm that
makes the decisionto heat and how much
is only one
line
of
an
8 page,
single-
spaced,
program.
The
rear
of
the
unit
incudes five connections. Two for the 12
volt DC power supply, another for a 1/4-
inch tube thatcan be plumbed toanexter-
nal port in the event the unit is within a
pressurized
capsule,
and two
electrical
switch contacts that can be used to turn
somethingonand off, typicallya solenoid
valve to fire the balloon’s burners. It’s all
housed in a unit that measures 4 x 12 x 8
inches and costs?? “Around $10,000” is
as specific as Comstock will say.
Each unit is also equipped with an
ear-shattering90decibel altitudealarm. It
serves as
a
watchdog,
checking about
once per secondonthe performance ofthe
autopilotand ready to sound off if neces-
The current Model II alarms at a
variance of 500 feet of altitude, above or
below the set altitude, or if the ascent/
descent rate is more
than 300 feet per
minute. Why the need for an alarm? “Re-
member these units were designed to be
used by a sleeping solo pilot that could
end up deadif they fail towake up to deal
with a
problem,” says Comstock.
The
most likely cause of an alarm is running
out of fuel in the tank connected to the
burner.
The unique design of the Model II
program allows the autopilotto adapt on
the fly; since most long distance Roziere
balloons are flown only once, on the real
flight,
not
a
test
flight,
this was
very
important. The unit has to reliably adapt
toany balloon withoutthe benefitof a test
flight during which various parameters
could be preset. Most of his testing was
done in lightly loaded hot air balloons
because of their greater degree of diffi-
culty to control.
“IhavearoutineI taketheunitthrough
now,” says Comstock,
“including shut-
tingoff the fuelsupplyuntilthe balloonis
ina steep descent, thenturning onthe fuel
and watching as the autopilot stops the
descent, roundsoutthe balloonandfliesit
back to the desired, presetaltitude. Foran
experienced balloon pilot thisis a simple
task, but to teach a black box to do this
reliably is another matter.”
Seven Model II autopilotshave been
built;twothatflew withSteve Fossettare
two
more
were
built as
replacements.
Kevin Uliassi has two and a spare had
been on the shelf until it recently went to
a
team
staging for
an upcoming
long-
distance flight to the North Pole.
There isalso a ModelIII on the shelf.
That unitwas tofeature a numeric keypad
for easyinputof altitudes,rate of climbor
descent and variousother control factors.
However, when balancing the extremely
limited market for such a device against
the years of development and testing re-
quired to create a reliable design, Com-
stock says the Model III will stay right
there, on the shelf.
Currently he is at work on a Model
IV.
It’s primary
advantage
will be the
addition of
an altitude shift device. By
dialing in a
preferred rate of
climb or
descent the autopilot will be capable of
morethanjustlevelflight.Thiswillfacili-
tatealtitudechangesfrom one cruiselevel
to another.
Fornow,withthe successofBreitling
Orbiter III, the potential market for even
the Model IV
can be
measured by less
than
the
fingers of
one
hand.
In
fact,
Comstock sayshe hasnoactive interestin
another unit. Still he takes great pride in
the fact that the Model I and II combined
have flown more
than 57,000 air miles
and achieved three
absolute world dis-
tance records for balloons. Quite an ac-
complishment for a little black box!
