NOTE: Still under review, by no means gospel or in any way truthfull...
Overview:
RASCAL is a scanning laser altimeter. The instrument itself simply sends a
beam of light, at various angles, a couple thousand times a second towards a target (usually the earth) and measures
the time it takes for the light to return. While measuring this time of
flight, other components of the system must measure the position of the aircraft
(GPS)* and the alignment of the aircraft with respect to gravity (INS)*.
Gateing is performed to make sure the signal measured was the proper one, and
a total return energy for the laser is sampled. This will help in post
processing calibration of the system.
All measurements are tied to one time frame, so they can be merged later in
post processing. This is all that rascal does as an instrument.
* INS = Inertial Guidance Unit
* GPS = Global Positioning System
After Processing and a Summary of the Processing Steps and Methodology:
While this is all the RASCAL does as an instrument, what it produces is
in post processing is a swath of 1 meter diameter footprints, spaced
approximately every 1.5 meters in both directions (assuming a non accelerating
aircraft) 100 meters wide. Each measurement is accurate to less than 20 cm, but
typically on the plus or minus 5 cm range.
To get to this point, a few keys steps are necessary. The laser range (or
round trip time divided by 2) must be transformed into an elevation of the
surface of the earth in a useable reference frame. The laser range is
tagged with a time. Independantly, we measure the position of the aircraft
with GPS, which produces a traectory for the aircraft. This is a position
of the airplane for a given time. So, we have the position of the aircraft
when the laser was fired.
Another piece of the puzzle, is where is the laser. The position from the
GPS trajectory is the position of the antenna, not the laser. So, we come
up with a physical offset or translation from the antenna to the laser. This
has been done through brute force measuring, and through static GPS surveys
of the aircraft. (we mark positions and then move the aircraft and set up
static GPS surveys of the marked spots). Ok, now we have the position of
the actual laser when it fired. We then must rotate the beam to the proper
angle, because odds are, the plane wasn't flying with zero roll, pitch and yaw
angles.
The INS is where we go next for information. This device tells us what
roll, pitch and yaw the aircraft is at with an update rate of 64 Hz. We
interpolate to the time the laser was fired and solve for what the angles
the aircraft was positioned at during the measurement.
One more device comes into play, and that is the digital wheel encoder.
There is a spinning mirror at the heart of RASCAL. It is a 45 sided polygon,
which essentially looks like a wheel. The laser beam is directed into the
polygon while it is turning, which creates 45 scans per revolution of the
wheel. The beam travels 16 degrees in one direction, and then resets to the
other side and begins to travel again. The position of the wheel is recorded
and stamped with a time. These positions are interpolated for the time of
the measurement, and you have the position of the scan beam for every shot.
With all the pieces at the ready, many matrix transformations and rotations
are performed to bring the range to the surface, to an elevation measurement
at some latitude and longitude of the earth.
Technical Information:
The laser is a spectra physics CW pump NYd:YLF which is frequency doubled to produce
a 6 to 10 nanosecond pulse at 523nm wavelength. We run the system at 5 kHz,
and the laser is optimized for 2 kHz, so we are not allowing the laser to
pump to full power before each firing. Our telescope is a fixed field of view
with 12 degree across and a couple milliradians forward and back (direction
of aircraft travel). It has a 12.5 cm diameter and a very tight filter in front to only allow 523nm light which is funneled into a photo
multiplier tube. We typically run the tube at 700 to 1200 volts.
The beam is steered by the 45 sided polygon spinning mirror. The position
of the polygon is sampled at 100 Hz with a 16 bit resolution position with
respect to a single revolution. The time interval unit we are using is
a LeCroy 4204, which has a time resolution of 156.25 pico seconds. Our
return energy is measured by integrating the return pulse with a LeCroy
2249 analog to digital converter.
The data system is a standard IBM PC AT compatible currently using a
90 MHz pentium processor. The laser control and data are routed from the
PC through the CAMAC crate. All gating and laser related wiring occur
in the crate. The PC also ingests a very accurate 1 Hz time synchronization that is
produced by the GPS receivers and tied to GPS time (and UTC). In addition, the reciever provides the
data system with a crude position in the form of a GGA serial message passed
on the COM port of the PC. The encoder is read through a custom wiring layout
into a generic digital I/O port. The INS is read through an off the shelf
ARINC-429 interface card.
All data is recorded to generic SCSI hard drives, and then dumped to tapes
post flight. Drives are written to directly, instead of writing files, so
that data cannot be lost with improperly closed files (possibley caused
with power outages during data acquisition). All information is stored in
a custom formatted C structure, and all programs are written in C by us.
We then write a C program on the SUN (currently a Sparc Ultra 2 with 2x200MHz
processors) to convert this C structure to an IDL binary structure on the
SUN. We then move to IDL for all processing.
Short Comings and Plans to Correct Them:
The overall accuracy on a shot to shot basis is not as good as it should be.
This can be attributed to a couple things that we have learned while processing
the first missions worth of data.
Most of our range error is due to our method of sampling the return energy,
and in future iterations of the system, I plan on changing our method of
return energy measurement and add a start energy measurement. The wide pulse
width also contributes to overall accuracy. A new laser is in development
which will have a 1 nanosecond pulse at up to 10 kHz, giving us much less
range walk for different amplitudes of pulses. This should reduce the overall
error a good deal.
Return to the RASCAL home page...