For
me, this project is always
something I’ve been interested in, having toyed with car
computers
many years ago, and with my interest in flying, I was keen to
volunteer an avionics package when Darren decided to build a plane.
But,
rather than bore you with my
history, I thought I should give a run down on the proposal for this
electronic instrument system for the KR.
In
my tradition of trying to put a name
to the project, I’ve come up with this,
DiKRepid (Mark 1)
Digital KR enhanced
primary instrument display
Then
follow the specs, which are based
on every neat piece of electronics that goes into a modern aircraft
cockpit, that we think we could possibly build. (along with a few
really nice to have, but maybe a tad more difficult to build,
extras).
The
guts of the system will be
Main flight instruments, (AI, HSI, Alt, Airspeed,
Turn rate and Slip etc..)
Flight data recording, and cockpit voice recorder.
(Records GPS pos, and pretty much every other piece of data the system
can measure)
GPS Navigation assistant.
Initially,
there is no intention to make
the system comply with all the TSO requirements, which would make the
initial build rather difficult to achieve.
If
we can get all of the sensor systems
working as expected and with some level of reliability, then maybe
TSOs and future production will be possible, other than a basic
knowledge of the systems operation, most of the precision gear is
first experiences.
For
example. The gyro system for
attitude indicator and horizontal situation indicator. My first
attempt a building a gyro was less than successful. It seemed to
work on the bench, but when mounted in a model helicopter –
well
lets say the heli has never seen the light of day since. Lets not
dwell on the past though.
Work
has started on the attitude
indicator system, altitude and airspeed sensor circuits, and some
experimenting with an overall system spec that allows for easy design
and testing.
So check
out the work.
udty
Attitude
Indicator.
I
always
thought that an attitude
indicator would be labor intensive, but otherwise, fairly
straightforward to build.
Based
on a 2
axis gimballed gyro, we
just need some sensors on the two axis and hey presto.
So,
my usual
design thought process gos
something like this. What do I have in my junk collection that would
suit a Gyro. After some rummaging around, I destroyed about 3 or 4
VCRs and figured that the heads were excellent candidates for both
the gyro motor and the axis. They are well balanced, plenty of meat
on them to machine and attach other components and have nice bearings
in them. They also have a rotating transformer that is used to
transfer the normal video signals from the heads, back to the video
boards in the VCR, that might be able to provide a frictionless
method for power and sensor signals transfer.
Here’s
a drawing of the first
proposal.
After
some time in the tool room, this is what we have. This is one
axis, with a VCR head as
the gyro and motor mounted in the middle.
It rotates
about two end shafts. The
one on the left has an insulated rod for power through the center and
a spring loaded touching the machined point on the outside.
The shaft on
the right has a 500 line
optical encoder for position sensing, and also a contact touching the
outer end. Both shafts are 4mm diam, mounted in low friction
bearings. The bearings are lightly oiled rather than the usual
greased variety, and don’t have side covers, which means that
after
the machining work, they are now low friction bearing with grindy
bits in em. They will be replaced with two new ones when I’m
happy
with the rest of the setup.
This
is one
axis only though and we
need two to be able to detect pitch and roll. So this arrangement
gets mounted on another VCR head. The rotating transformer on this
second head is supposed to transfer high frequency AC power across to
the inner gimbal, but there’s a few problems with this at the
moment.
The
mechanical
arrangement looks ok
though.
Here’s
the stationary side of the
head, mounted into a PC power supply case.
On
the right is the connector for the
transformer in the head, and mounted on the near end of the shaft
with a screw on the end, is a small adapter that will support another
500 line optical encoder. The three hex supports will hold a small
circuit board, and the detector part of the encoder.
On the
inside of the box it looks like this.
The
rotating part of the VCR head sits
just through the hole in the case. I drilled some holes and tapped
M3 into them to support the inner gimbal. The green and white
printed circuit board in the the middle of the head connects to the
moving side of the rotating transformer, and the heads. The heads
wont stay – they get removed to save weight and circuit
complications.
With
the inner gimbal mounted, this is
what it looks like.
Case clearance is a small problem at
the moment. The screws that hold the optical encoder module (brown
plastic thing on the left of the inner gimbal) stick out a bit too
far.
For this unit to work smoothly, balance
is of the utmost importance. However, for some reasons do be
discussed, as neat at this looks (esp when its running), there are
some basic problems that I have learned about, that may not be able
to be resolved with this design.
So at the moment, this is on hold and
some other work is being done.
Solid state devices are now being
investigated as an alternative to the mechanical gyro, and work is
proceeding on the interfacing circuits.
PC Interface
I’ve
been interfacing devices to computers for over 20 years now.
At first
it was basic digital interfaces like printer ports, memory expansion
modules and the like. The designs were never complicated
because there
were Microsoft was not yet big enough to tell computer manufactures how
to build hardware.
How
the world has changed !
The
options I have now, are : Serial
port (slow and messy, but with an RS232 networking protocol a friend
and I designed about 10 years ago, quite reliable). USB
The high speed, much more versatile successor to the above. Parallel
port – Good performance, no voltage conversion issues, most
PCs have a
standard interface, and programming’s not too hard, but
Win2K/XP make
things messy here. There are drivers around to let you use
traditional
IN / OUT commands and write directly to the PC parallel port chip, but
I would prefer not to be running too many connections between the PC
and peripherals in an aircrafts environment. Dedicated
interface board. Easy with ISA cards. PCI requires
dedicated drivers
in windows I think, and then IU have to deal with plug and
play.
Although there are books on the topic, I prefer to get my help from the
net and the PCI community seems pretty closed. This rules out
PCI
express too. So
USB seems to be the best. There are heaps of microcontroller
chips
supporting USB in hardware, and there is plenty on the net to support
it (and me).
Check
out
MJOY if your interested. This
is a very neat joystick controller that implements USB in
software. It
has 6 analogue ports and 28 digital ports. It uses the
ATmega8-16PI
which is available here from Jaycar for about $20. The
programming
hardware is just a connection to the parallel port of a PC with some
resistors, and the programming software is downloadable (ponyprog).
Here
is the result – working without any major issues.
I havn’t yet tried to mod the software on this, but there are
some minor issues with the way it connects to Windows, which need to be
resolved. Of course we expected this.
The device works as a standard Joystick interface, and so no special
drivers are required. It can be programmed using standard MS
DirectX / Direct Input compatible code. But the nature of
this is that direct input make the data coming from the device work
like joystick and I don’t seems to be able to get the raw
data from the chip. I don’t know for sure yet if
this is a problem with the software on the chip, or a DirectInput
translation problem. Anyway – it doesn’t
matter because the web is at hand. There is another project
on the web (from Objective Development) using an almost identical
circuit that has much better interface code. It just requires
a lot more development software – like the Windows driver
development kit. (Didn’t want to go down this path
but I always did want to check out the kit).
So the software is being reviewed now, and some mods have been done to
the circuit to interface some sensors that will be doing airspeed,
altitude and temperature sensing.
More on this soon.
Unfortunately, I am also building a house and this takes priority for
me.
For those interested, you can check out these USB related sites.
Well,
Darren and I both work for a
Lighting company, so this one better be good.
To
start with, there are several
lighting requirements that we want to meet, not because we have to,
but because lights look good, and they need not be expensive.
Lighting
Inventory Exterior
Position lights
(red on left, green on right and white on tail tips, non flashing)
Anti collision
lights (strobes on left and right wingtip, and on top of tail)
We
intend to design and manufacture all
of these, using readily available components, but with the aim of
keeping costs low.
Anti-collision
Lights
We’ll
start with the Anti collision
strobe, partly because I’ve never worked with these before,
but
mostly because I’ve been pretty slack with the web site
updates,
and in the mean time, most of the hard work is already been done.
To
make a xenon strobe flash, you
require a few hundred volts available in a reasonable sized
capacitor, to provide the flash energy, and a short 6000 Volt burst
to trigger the flash. Of course, in an aircraft, all of this must be
derived from your available supply – 12V in our case.
WHAT
– 6000 Volts ? – Yes. Stop
for a second and do some reading on Xenon bulbs – they (and
their
driver circuits) can be deadly. Check out http://members.misty.com/don/xesafe.html
Xenon
bulbs made from quartz (as
opposed to glass) can also emit dangerous levels of UV radiation. When
building and testing this circuit, keep your wiring neat, your
bench clean and try and keep the high voltage ends away from other
bits and especially fingers. I keep voltmeters connected to the high
voltage terminals and if anything looks out of wack, I throw a
screwdriver or large resistor (physically large, but low resistance
–
10ohms or less), across the capacitor terminals to discharge them. A
resistor is better for this, as screwdrivers can weld themselves to
your circuit – just make sure the resistor is big enough to
hold in
the middle without having your fingers near the wires.
Our
initial circuit, (and associated
excellent source of information on the topic) came from a recent
article in “Kitplanes” magazine – Jul,
Aug and Sep 2007
editions, titled “Flash Yourself”, written by Jim
Weir.
As
mentioned at the end of part 3, the
circuit could do with a number of improvements, and that is where we
will be starting.
The
components specified in Jim’s
article, are not hard to find, but the specs on them are nothing
special, and I wanted to use some parts that I already had, so by
changing a few parts from the original circuit, this is what we have.
Changes
from original: The
4049 chip has been changed to a
4069, which doesn’t have quite the output drive, so the
trigger has
been modified a bit to suit. The
MOSFETs I have used have a lower
Rds (on resistance), but a similar gate threshold voltage. The
transformer T1, has been extracted
from an old PC power supply. The only inconvenient thing here, is
that the transformer secondary (which in our circuit becomes the
primary), has on terminal coming from the top of the winding, instead
of below with the rest of the terminals. For testing this is fine,
but on board a plane, this will vibrate and create problems. If you
have an old PC power supply, you can use the power transformer (which
should be the largest of two or three similar looking devices) as T1
in our circuit. We just need to make some adaptations. The
Xenon flash tube I have used, is a
$5 job from Jaycar electronics. No idea about its power output or
life, but I figure that when it burns out, we’ll replace it
with a
quality unit. Jaycar also sell a trigger transformer to suit it. In
case you have one without data, here’s how mine is setup.
Other changes include the 22nF 630V
firing cap. The cap doesn’t need to be rated at 630V, it just
one
I had. It should be rated to at least 450V though. Mine is the
light blue in the photo.
Here’s
my first version of the
circuit.
I’ve noticed too, that this
particular design (like most actually), has a tendancy to completely
blow it self to kingdom come, if the power is connected backwards.
Out
of all the components on the
circuit, the main transformer, and the heatsink came from the old PC
power supply. I spent about $15 on parts to make it work (Aussie
dollars – a couple of rupee for the rest of the world).
So
now, we have one cheap xenon flash
bulb, with a broad daylight visibility range of about 300m, which
needs some improvements.
This
is what we want to do.
Increase
brightness – considerably.
Operate three
bulbs in a double quick flash pattern that repeats each second.
Reduce the heat
produced by the 1R0 resistor on T1’s primary. – As
suggested by the original article, the switching MOSFETs need to
alternate with no overlap. The current circuit has some small overlap
due to the 4069 inverter between the gates of Q1 and Q2.
Although
our bulb flashes ok, we need
to increase the voltage across the bulb. Without trying to get a
transformer with a higher output voltage, we will substitute the
4-diode-capacitor setup on T1’s secondary, for a voltage
tripler. So after extracting the 2 large caps from the old PC power
supply,
and doing some other diode changes, we now have this. Just to test
the parallel bulb theory too – I’ve added a second
xenon tube. The voltage applied to the flash now get to about 500V, as
opposed to
the 250V that was present before the tripler went in.
This
diagram shows the change to up the
voltage.
Next step, is to replace the 4069 chip
with a small microcontroller. It adds to the cost and complexity,
but as the one circuit will do three bulbs, it also allows me to add
some features. With a small program change, I can tune the switching
times of Q1 and Q2 to maximise the circuits efficiency. I can also
generate any flash pattern I like, and it is easy to put some checks
into ensure safety – like a charge monitor on the main
capacitors. Put these suckers over-voltage and you will know about it.
If
you want to make one of these, then
I suggest you get the Kitplanes article, available from www.kitplanes.com
