APOLLO must get the pointing right in two respects. First, we need to correctly position the laser beam on the moon. Since we want the greatest possible laser beam flux, we want the tightest, smallest laser spot we can make on the lunar surface. The atmosphere limits the size of this spot to one arcsecond (i.e., in one-arcsecond "seeing"), or two kilometers. So we must point the beam with a precision of less than 1 arcsecond. Second, we must have our 1.4 arcsecond receiver pointing to the correct spot to see the return—also within about one arcsecond.
One might think this is simply a matter of locking the outgoing beam to point precisely along the direction in which the receiver looks. But we cannot do this because the moon is moving to the side at 1000 m/s, and the earth is rotating at about 400 m/s, so that we must carefully "shoot ahead" while looking behind. Another way to say this is: light takes 2.5 seconds to get to the moon and back. We have to shoot where the moon will be in 1.25 seconds, while looking at where the moon was 1.25 seconds ago. The offset angle is typically around one arcsecond, so we cannot ignore it. Moreover, the offset changes in our instrument as a function of azimuth and elevation, so we must dynamically track it.
We can verify the direction of our outgoing beam by looking at the return from a corner cube in the telescope exit path using our alignment/pointing CCD camera. The laser return from the corner cube looks like an artificial star: we can clearly identify its center. We compare the position of the corner cube return to the known position (pixel) of the receiver on the CCD camera. If we align the corner cube return to fall dead-center on the receiver position, we have zero offset, and call this the origin. We can then apply our deliberate velocity correction offset to this origin. We verified on several occasions that we were doing this correctly.
In December, we saw that we got strong returns when we moved the beam offset about two arcseconds (to the "left") from its nominal position. This was mysterious, but repeated itself in January, and appeared to be present again in April (though at lower signal, thus lower confidence). We struggled through the spring to acquire enough data to further characterize this anomalous pointing bias, but had difficulty doing so.
Practically, this meant that we did not understand our pointing issues, and were forced to execute searches (via time-consuming raster patterns) in both the beam offset dimensions and the overall telescope pointing dimensions—spending most of our time rastering the poorly-understood beam offset. A true search then required four-dimensional rastering, or 81 points for a 3×3×3×3 search grid. We seldom performed this in a truly systematic way, being relatively confident in telescope pointing. As a result, finding the return was difficult.
In June, we recognized the source of the problem. Our laser is reflected off of a spinning optic called the transmit/receive switch, or T/R switch. This optic can be seen as the semi-transparent thick disk at a 45° angle at upper right in the top picture on the optics-bench description page. In particular, the green beam emerges from the left, and gets a reflection downward at the T/R optic. The T/R optic is almost entirely clear, but has a reflective patch on its surface that rotates into position just before a laser pulse is sent out. The rest of the time, the clear optic passes light from the telescope to our CCD camera and photon timing detector.
Because the mirrored patch blocks the CCD camera completely, we must perform our beam alignment using the corner cube in the telescope with the optic in the clear position, meaning that only a small fraction of the laser light reflects off of the clear surface of the glass. But it is enough light to perform the measurement.
The problem all along was a tilt in the optic, such that its surface was not precisely perpendicular to the rotation axis—resulting in a wobble as it spun. In designing the optical mount, it was understood that there would be some wobble—amounting to not much more than one arcsecond on the sky. But this does not affect our laser ranging practice, as we always strike the same part of the optic (the reflective patch) with the laser every time it spins around. So any small wobble should be of no consequence. But no!
Because we perform our alignment in the clear rotation, it is in a much different position than the firing position. So a tilt in the T/R optic will translate to an offset, or bias, in our outgoing beam alignment. What we called the origin was not the relevant origin for the pulses sent to the moon.
It was not hard to verify that this was our problem, and we quickly had a quantitatively consistent explanation for our anomalous two-arcsecond offset. We were able to see this in a variety of ways—once we knew to look for it.
The fix? Simple: we can now deliberately offset the "origin" to be consistent with the tilt of the T/R optic in the firing position. It is also easy to monitor this offset.
Now that we understand the outgoing beam alignment, we can lock this down to its calculated offset (from the correct origin), and we don't have to worry about rastering this parameter. This turns our 81-point rasters into 9-point rasters in telescope pointing alone. APOLLO is only practical in this limited-search paradigm, so it is very encouraging to have landed here.
The night after this discovery, we had time on the telescope, and acquired Apollo 15 without much struggle. We did this even though the terrain immediately surrounding the reflector was in shadow—marking the first acquisition without visual verification. Once well established on the Apollo 15 reflector, we slewed the telescope directly to Apollo 11 and had a signal within one minute of the slew. After a solid Apollo 11 performance, we whisked over to Apollo 14 and promptly saw this signal (Apollo 14 was completely in the dark). This performance is very encouraging, supporting the notion that APOLLO has turned a corner and can practice efficient (even routine?) acquisition of all lunar reflectors so that we may begin our millimeter campaign in earnest.
