Training of the LX200 PEC with autoguiders seems to fall into to two categories, that done with the ST-4 and that with the 216XT. The report here is from
Philip Perkins on the use of the ST-4 (note: mixture of English and US
spelling intentional)
The reports on the use of Pictors for this purpose are very scarce at
this time (January 1999) See short comments below for
the 216XT data.
The very best information I have seen on the issue of Guiding Techniques for Astrophotography (film) have been compiled by Philip Perkins and can be found at his fine web site <http://www.astrocruise.com> Mr. Perkins is, by the way, one of the finest film photographers ever to use an LX200, in my opinion. I have freely extracted from some of his MAPUG-Astronomy posts on this topic. (slight editing imposed, Ed.)
Reports from Philip Perkins, who writes about his experience with the ST-4.
For what it's worth.. I've been training the PEC on my LX200 for about 4 years now. I have found that on my LX200 accurate PEC training is absolutely essential to achieve accurate guiding over the 3 hour exposures that I sometimes make. If I do not train my PEC then sometimes the errors have sufficient magnitude to drive the star right out of the ST-4 guiding box. I achieved the most accurate training last June. It is so good that I have hung on to it ever since. The big surprise is that this training was done with the ST-4. This directly contradicts a post I made a year ago in which I said that I had not had any success in training the PEC using my ST-4.
I think that three factors made the difference: - The seeing was exceptionally good - I chose an integration time of .3 seconds on the ST-4 - I selected Faint Mode on the ST-4 Following the training, I did one update.
I believe that all three factors were important, but Faint Mode especially so. Faint Mode has a special algorithm that smoothes the star image, as well as boosting its brightness. This has the tremendously useful effect of canceling out rapid star movement due to scintillation. When we train the PEC we want to record only star movements associated with worm error, and we must be careful to avoid programming in movements due to scintillation.
Faint Mode meant that only genuine drift corrections got programmed in, and the .3 second integration meant that errors were corrected very swiftly. When using Faint Mode, star brightness readings should be in the range of about 8 - 18. The combination of .3 seconds and Faint Mode means that some time may be required to select a star of suitable brightness. But it could be worth it. I continue to be very happy with the training I achieved 6 months ago. In retrospect, the effort seems very worthwhile.
During my four years of experience in PEC training I had tried using an autoguider many times, using many different settings, with both my original 201XT and my current ST-4. None of these attempts worked very well, and until last June I also believed that manual training was the only viable method. But there was one very significant change that I think was primarily responsible for the breakthrough I made last June. As stated earlier, I used Faint Mode. As well as boosting star brightness, this mode also smoothes the star image which almost completely cancels rapid star movement due to scintillation. The effect is dramatic, and it overcomes about the only major limitation in using an autoguider for PEC training.
I agree that in theory it should be possible to achieve excellent PEC training every time using the manual method. The problem with the manual method is that it needs to be done by humans, and humans are all different. There are some very real and practical limitations to the manual method from my own experience. I suspect that many LX200 owners will nod in agreement at these points. Despite best attempts, it is very difficult to maintain split second concentration over the whole 8 minute period, twice over. How many times does one's attention wander, or is one distracted by something? When one's concentration is restored the guide star is often found well outside the limits.
How many times does one over-or-under shoot when making corrections? Despite years of experience in discriminating between seeing induced motion and worm induced motion, it remains difficult to do this accurately. All too often one is left madly stabbing the keypad while trying to bring an over-corrected star back to the centreline. An autoguider completely overcomes these foibles. Corrections are very precise, occur very rapidly at the moment of the error, and are completely consistent over the whole 8 minute period. However there are two problems with using an autoguider, which may be overcome as follows:
Normally an autoguider is completely unable to discriminate between seeing induced motion and worm induced motion, whereas the human eye can, to a limited degree. This is a major problem, but is very effectively overcome by using the Faint Mode of the ST-4. As mentioned above, the effect is dramatic. Note that this solution is unavailable with other imagers. Note that binning is not the same as Faint Mode.
The results with the ST-4 seem to indicate that it is more than rapid enough for accurate PEC training. Yet higher frequency can be achieved by using 0.1 second exposure. Moreover, each correction is very precise compared to what an average human can achieve. There are others out there with ST-4 experience. In fact the vast majority of the most advanced astrophotographers use the ST-4, and many of them use sub-second integration times with their AP refractors.
My best training (using the ST-4) is somewhere around 6 -8 arc seconds peak to peak. The variance is due to recording slightly different results at different times. It's difficult to measure accurately because of star scintillation. Under ideal seeing and expert manual training error residuals in the 4 to 8 arc seconds area have been reported. This is exceptionally good with the known limitations of the Meade PEC system which averages over 2.4 second intervals..
The reason why the autoguider method is so interesting is exactly because of repeatability. It takes out the human factor. I suspect it may do
so for several other LX200 users. It sounds as though you are able
to achieve superb training absolutely repeatably, but I wonder how many
other people can? Not only do I wait until seeing is optimum but
I program the PEC during the time just before dawn when seeing is steadiest. Opportunities for optimum PEC training are pretty rare which is why I
need to hang on to an optimum training for as long as possible.
This topic has had a good airing now and I feel comfortable that the pros and
cons on both sides (manual training and CCD guider training Ed.) have come to
light. But at the risk of wearying the patience of those not interested
in PEC training I would like to make a couple of observations.
I can't claim to have achieved anything much better than 10 arc seconds peak to peak with manual training. With the average scintillation that we get here I find it near impossible to record this much more precisely than 1-2 arc seconds either side. It's possible that with the training on July 8, 97 I achieved 8 arc seconds, but I also recorded 10 arc seconds on a couple of occasions. I use a Meade 4000 series 2x barlow in conjunction with a 12.5 mm illuminated reticule Plossl. This gives about 400x, which is about the maximum for any meaningful work under most conditions.
Discussion of Faint Mode in the ST-4
Faint Mode is potentially very important for any LX200 owner who has to work under less than ideal seeing conditions. For me, and I suspect many others, that means most of the time. Faint Mode works by replacing each pixel value in memory with the sum of the 3 x 3 box centered on that pixel. Hence it greatly boosts the brightness value of each pixel and it also performs an averaging / smoothing function across the entire star image. This means three things:
- the overall star brightness is greatly increased
- if a star is in rapid motion owing to scintillation, the star's average position
in that range of motion will be accurately uncovered by the smoothing function.
- sub pixel accuracy can be reduced where a pinpoint star image is available
in completely stable conditions.
It is true that the resulting star image will be expanded, but if you look at the mathematics of the function you will see that in practice it has very little effect on the centroid calculation. As an ST-4 user you will be aware that it has a particularly sophisticated and effective centroid calculation algorithm that is capable of detecting the centroid to within 0.2 of a pixel. However this is only made possible by the pinpoint star images delivered by precision short focus instruments in stable conditions, where typically the star image is mostly contained within 1 pixel. This will never be true for those of us using long focus SCTs, where the star image will always cover several pixels, even under very stable seeing conditions. The real answer is that for SCT users, there is no perceptible loss in guiding resolution through use of Faint Mode because the star image always covers several pixels to start with, and the accuracy of the centroid calculation is hardly affected under these conditions.
Well, that's the theory side of it, but in practice the rationale for Faint Mode gets much better. In practice, most of us have to contend with less than perfect seeing, most of the time. This has two main effects on the star image.
- The star's position remains fairly stable, but it is bloated and distorted by the seeing. The distortion often causes false brightness spikes in the star's image that are offset from the centre. A short integration can mean that the star is captured in such a distorted shape, causing the autoguider to make a false correction. The smoothing function restores the round shape of the star and restores the brightness weighting towards the centre of the star.
- The star image is fairly crisp, but it moves ('wobbles') rapidly about its true position. This can mean that the star image is smeared across several pixels with a brightness peak usually offset from the true centre, causing the autoguider to make a false correction. The smoothing function averages out the brightness variations, allowing the centroid calculation to fix the centroid on the average position.
Of course, there can be combinations of the above, where the star image both bloats / distorts and wobbles at the same time. In such conditions the benefit of Faint Mode is even more dramatic.
The result of all of this is that in moderate or poor seeing conditions using a long focus SCT, the use of Faint Mode _significantly_improves_ guiding accuracy. This probably sounds like a lot of theory. Not a bit of it - in the early part of 1997 I exposed a large number of frames comparing Faint Mode with Normal Mode. The results were so compelling that from March 1997 onwards I completely standardised on Faint Mode. Every single image on my web site has been exposed with Faint Mode enabled.
I'll pick an interesting example: <http://www.astrocruise.com/m51.htm> because I have been asked a couple of times if it was taken using an AO system. I suppose the answer is that SBIG's AO system doesn't work too well with film ;-) - no, it was simply taken with a 'virgin' LX200 using an ST-4 with Faint Mode enabled.
I expect that most LX200 users with autoguiders have seen this problem: the
guiding errors are large and frequent large corrections are made. The
LX200 is constantly over-or-under correcting, and the result is trailed stars. What's happening is that the autoguider is 'chasing the seeing' - exactly what
it should not do. Enable Faint Mode on the ST-4 and the result is dramatic
- guiding errors are cut in half and the whole system stabilises - the system
is guiding on true star drift, as it should do.
Comments from Doc G
The above related experiences are similar to others I have seen over the past year or so both as to experience for typical self guiding function and in use for training the PEC. I thought it worthwhile to post these comments so that a complete picture of the dedicated use of the ST-4 can be had. It is clear that a CCD guider can be used for PEC training, but that everything has to be just right. Mr. Perkins' experience with the ST-4 is very encouraging.
I have found training for the full 8 minutes to be tough. Eight minutes can seem like a very long time when staring into the scope, bumping it and waiting for it to settle down, and punching the buttons like mad at the same time. In fact, at magnifications of 300X I found it takes upward of 5 seconds for the LX200 to settle down to the point where meaningful corrections could be made. This leaves a gap of 2 or 3 sections without correction. I did manage to get a ten times reduction in the worm errors with one learn and one update and was at the 6 to 8 arc second level. This is about is as good as possible I think. But, I would certainly like to train my LX200s with a CCD imager. It is clear that the ST-4 does send information often enough to effect a good training. Since the PEC averages over 2.4 second sections of time, it might only be necessary for the CCD imager to send corrections at that rate. One must apply sampling concepts, such as the Nyquist criterion, with caution when the effective bandwidth of one element in the control chain is so limited.
Comments On the Use of the 216XT for PEC Training
I was not very successful using the 216XT to train my LX200s, but that was about a year ago, before I did the bearing modifications. I used the 216XT with an exposure setting of 1/4 second and choose a star bright enough to get a brightness reading of 10 to 20 on the 216XT readout. This seemed to give corrections every second or so. I found that this was satisfactory for guiding after the PEC was manually trained but was not useful for actually training the PEC.