Re: [APML]: Autoguiders comparision

[Steve Bell <sb635@delphi.com>]

Pedro Breda wrote:
> 
> 
> I am about to buy an autoguider but havent been able to decide yet between
> the 201XT, 208/216XT or the ST4.
> First I have to say that althought everybody says that the bigger pixels in
> the ST4 are better, I think quite the opposite....
> 
> BUT,
> althought I think the 201XT chip is better, the chip is not all. The most
> important factor is the guiding algorithm....
> 
> 
> I have been reading the "Shortage of Guide Stars" thread with very
> attention and Wil Milan refered to this problem:
> "
> I had other problems with the 201XT as well. It seemed to chase
> scintillation,
> so it was continually making little twitchy adjustments back and forth. The
> ST-4 avoids that by being able to guide on faint stars which don't twinkle
> so
> much, but it also averages several pixels around the star to determine its
> exact position, allowing it to pinpoint the star location on the CCD array
> within 1/5 of a pixel each time. It doesn't seem the 201XT does anything
> like
> that, and thus it not only chases twinkles but has lower guiding resolution
> as
> well.
> "
> So what I would need to know is what is the algorithm used by the
> 201/208/216XT and ST4 guiders.... 

I don't know about the Meade autoguiders, but this is how the SBIG
autoguiders operate. Say a guide star has been imaged within the guide
box. Each pixel in the guide box has a certain intensity count. Given
the guide star has been centered in the box, the pixels around the edges
will be at the background count, while the central pixels falling under
the star will be registering high counts because of the star's light.
Think of each pixel's count as the weight of a "bar" located at the
pixel's position. Picture each of these "bars" to have a height above
the pixel, where each bar's height is reflective of its weight, ie,
count. Over the 2-D area of the guide box, the 3-D "histogram" of the
bars will produce some mean, ie central, location that is the center of
mass of the histogram. Hence the term "centroid."  The words "to within
1/5 of a pixel" refer to the fact that the centroid is computed to
within 0.2 pixel in each of the x and y directions on the chip. In order
to perform an actual autoguiding, the autoguider must first be
calibrated. During the calibration process, the scope is purposefully
moved in +/- RA and +/- DEC, and the guide star's centroids are computed
after each movement. A "velocity" in terms of pixels per seconds of
movement are computed from the observed shifts in the guide stars
position. There are four velocities: one each for +x and -x (usually set
up to be in the RA direction), and one each for +y and -y (usually set
up to be in the DEC direction). Onec these calibration "velocties" are
computed, they are then used during an autoguiding session to compute
how much time each x and/or y rellay needs to be activated in order to
null out the observed centroid location error (from the beginning (0,0)
point in the guide box) during that particular guidance update.

Now, just because the centroid is computed to the nearest 0.2 pixel,
that certainly doesn't mean your scope and mount will be able to track
with pixel errors no larger that 0.2 pixels. Sure, the centroid is
accurately computed, but if your mount can't track well, it could easily
be several pixels away from the desired (0,0) point. How well your
scope/mount keeps the computed centroid close to the (0,0) point over
time is the real quantification of how well it is tracking.

Just how big you want the pixels of your guiding CCD to be is related to
several dependent factors. First, there is the focal length that's doing
the actual imaging. Generally speaking, the longer the imaging focal
length, the more accurate your guiding needs to be. A good rule of thumb
is to use a guiding focal length as long as the imaging focal length,
which is easily accomplished using an OAG. If you use a guide scope, put
on a Barlow to bring the guiding focal length to at least 1/2 the
imaging focal length, and preferrably up to at least 3/4 the imaging
focal length. That is, up to about a guiding focal length of about 2000
- 3000 mm. Some folks experience with an ST-4 seems to show that
regardless of the imaging focal length, it's not best to autoguide with
a focal length any longer than about 2000 - 3000 mm. I have a feeling
that this is related to seeing effects. With long guiding focal lengths,
the guide star nearly fills the guide box and the star's image jumps
around more (due to seeing) than with a smaller focal length, and the
scope will begin to chase the seeing, which is not good. So you might
say, use a smaller guiding focal length so that seeing does not affect
the guide star image so much. Ok, that's fine, but you can't go so short
that you don't get enough magnification on the CCD's autoguider's chip
to be able to adequately see the drive's periodic error. And what size
pixel is correct for all this? You don't want to use a pixel that's too
large for the guiding focal length such that too much tracking error has
to build up to be seen by the autoguider before a correction is made.
And you don't want to use a too small pixel size so that the star image
jumps around in the guide box due to seeing effects. So you see, there
are several inter-related factors that come into play.

I would advise, regardless of what make of autoguider you use, to use a
guiding focal length equal to or at least 3/4 as long as the imaging
focal length (I'm talking about deep sky here - you can usually always
guide a piggy-backed camera through the main scope). Then, experiment,
experiment, experiment. Start off with short guidance exposures of about
1 sec. Take some test 15 min shots. If you get trailing, then it's
likely that either you have a fairly bad drive, or your scope is chasing
seeing. Keep increasing the guidance expousre in 1 or 2 sec steps and
see what happens. If increasing the guidance exposure doesn't help, then
your drive is probably too inaccurate to track well. You might try
shorter than 1 sec guidance exposures, but then available guide stars
get scarce, and you really need good seeing so your scope doesn't start
to chase it. If you have only so-so seeing, and you've got a good mount
and drive, then increasing the guidance exposure should help. The longer
guidance exposures actually allow the guide star to average out it's
scintillations, and the scope won't chase the seeing as much. On my
16.25" f/4.7 Newtonian that's mounted on a mount with a 7.5" Byers drive
that delivers a system periodic error of +/- 9 arcsecs, at my site with
only so-so seeing, I can use a 5 sec guidance exposure with an ST-5 in
an OAG configuration and usually get good tracking results.

Oh, and with regards to what brand autoguider to buy? I'm biased, but
I'd get an ST-4.


-- 
Clear skies,
Steve Bell
email: sb635@delphi.com
Astrophoto page: http://www.mindspring.com/~sb635