Tracking issues

[admin]

Why do I get different configuration results with different pointing files

one might assume that if you make 2 pointing files you will get the same result regarding the calculation of Polar Alignment (PA) error or encoder resolution and other factors. This is not the case. Why ?

Mathematical background:

There is a set of formula that describe the pointing error of the mount
1. Polar alignment error in Azimuth
2. Polar alignment error in Altitude
3. Collimation error
4. Mount error (axes not perpendicular)
5. Tube flexure
6. Encoder resolution error
7. etc.

 

What Autoslew does when it gets a pointing file from you is to make a best fit of the variables in these formulas that give a minimum RMS pointing error in this pointing file.

If we assume we have a perfect mount and a stiff telescope we can use the configuration and make the same pointing file again and we will get the exactly the calculated RMS error you received as result after optimization. This is definitely true but why is it that your configuration shows different values after every new pointing file ?

Reason 1: Different number of stars in pointing file and thus different number of fitted
formula

When you add more stars in your pointing file Autoslew will automatically increase the number of variables that can be fit. If you take only few stars, it would not make sense to optimize more than PA and encoder resolution. When you add more stars, thus adding another formula to fit (like tube flexure) Autoslew will probably find out that the RMS error is smaller if it adds a certain amount of tube flexure instead of fitting all the error into an Altitude polar alignment error. Thus, you can have a different result now in PA since the extra Formula (in this case Tube Flexure) has Autoslew allowed to make a better fit. The more stars you use the more formula are used and the better Autoslew can find out which error really belongs to the PA (or encoder resolution). Certain errors can cause very much the same pointing error behaviour. As an example, when you slew from SouthWest 60 degree alt to SouthWest 20 degree alt and Autoslew measures 20.1 degree altitude there instead of 20 degree you can exactly explain this result by assuming that

a.) You have some tube flexure
b.) You have a Azimuth polar alignment error
c.) Encoder Resolution is wrong

Autoslew has only the chance to find out which of these errors might be the “real” one when you measure MORE stars between these 2 points because each of these errors will cause different slopes between the 2 above measured positions or use stars at different sky positions. For example if you continue the measurement exactly in South direction and measure the same error again the Azimuth alignment error CAN’t be the source because an Azimuth PA error would not be able to cause a Declination error at the Meridian. Of course Autoslew is not so smart to think like that. It does the same thing that all other programs like Pinpoint and programs that do pointing optimization for the very large telescope do: It varies the polar alignment factor, the tube flexure and the encoder resolution until it gets the minimum RMS pointing error of all the measured stars. As you might understand, when you measure only the 2
stars mentioned in above example the result will be random and chaos because it can fit this behaviour with any of the above formula. This changes when you add the 2 stars in South direction because the optimization will now give less RMS error when the Azimuth error is zero but everything is put in the Tube flexure (or encoder resolution...). When you add more stars at the right places, the optimization will also be able to distinguish between encoder resolution error and flexure and so on...

 

Reason 2: The telescope behaviour cannot be perfectly fitted with the used formula
This will nearly always the case. The ideal mechanical world will cause the telescope to flex like the assumed Alt-Flex=A*Sin(Z) where Z is the zenith distance. So if the telescope has a flex that is getting stronger towards horizon than the calculated flex constant A will be different depending on the altitude of the stars you choose in your pointing file. It might even try to fit the residual mistake into a PA error or another formula and thus adding 3 arc minutes PA error which in reality is not existent. The problem that you can never find formula that exactly describe YOUR telescope is also the reason why the optimized formula (=configuration) will change depending on which stars you choose to make your pointing file.

 

The next typical question that now comes up is:

 

WHY DON’T YOU ADD MORE FORMULA IN AUTOSLEW?

 

As an example, we could try to fit the Flex by a more complex formula like : Alt-Flex=A*Sin(Z) + B*Z^2 + C*Z^4 etc. If we only add enough factors in this series we can be sure that we can fit ANY flexure behaviour that is purely Altitude dependent. But wait – does that really make sense ? Yes, in case you want to make a pointing file with infinite number of stars it would indeed make sense to fit 50 different variables (assumed that we find a mathematical optimization algorithm that is able to find the exact minimum of a function with 50 variables). If you have only a limited number of stars and you want to make the best of it trying to fit too many variables is a good recipe for a complete desaster. The result will be random and chaos at positions where you don’t have measurements. The rule is: If you have few stars, fit few variables and only the most important once. However you might have noticed that if you add even more stars (like 30) Autoslew will add another fit on top of the formula fit – the Fourier Fit. That is a very powerfull fit to model the residual errors that are left after the Formel Fit. But it makes only sense to use this if you have enough stars. If you do Automatic Configuration Calculation you don’t have to care about all this because Autoslew will calculated how many variables to fit.

Reason 3: Hysteresis, non reproducible and random errors
This is probably very easy to understand. If you make 2 measurements with different random results (like depending on where you have pointed before), the configuration will also have different results. Sounds logic ?

Conclusion and Remarks:
You can expect to get different results with every new pointing file. This is NORMAL. From my experience the major error when making pointing files are hysteresis errors followed by flexure problems that cannot be modeled by Autoslew. When you make a 20 star pointing file and you get a residual RMS pointing error of larger 1 arc minute it simply does not make any sense to correct a 5 Minute PA error that Autoslew will show you because the chance is high that even without any change the next pointing file will show a complete different result. In this environment, you would simply not NOTICE any improvement of 5 arc minutes polar alignment because your other errors (Hysteresis, Formula residuals etc.) are much more prominent than the residual PA error. As a rule of thumb the determination accuracy of PA is about 5x the residual RMS error. Another option to find out when to stop PA is when you make a 5 star pointing file in East and then do the same in West. If you get -2’ on one side and +5’ on the other side you should stop the PA process because your measurements are not accurate enough. So people who have such bad results like 1arcmin RMS should (and I am aware that a lot of you will not like to hear this again and again) improve the reproducibility of the telescope and / or selectmore stars to allow also the Fourier Fit to take place.

Final remark about encoder resolution of DDM 60:

The Encoder resolution of the DDM60 is only another formula (like PA or flexure) and the accuracy of the factor depends on the quality of your measurement. If you get different results you will usually not be hurt by that because the errors during your pointing and tracking will be the same errors that caused your measurements to be non reproducible and they will be larger than the error caused by the error in your resolution error.

 

Tip:
When you did a reasonable PA (+-7 arc minutes) make a big pointing file one time, save the resolution and never touch it again. You should only redo this procedure if you have the feeling that you enhanced your telescope flexure and hysteresis problems and thus allowing you to make more accurate measurements.

 

For a manual how to meassure the Hysteresis and Time flex please click the icon below:

 

SequenceHyst-Timeflex.pdf

[admin]

The following is an information letter written by Philipp Keller regarding some tracking issues:

 

Dear ASA customers

This is going to be a long post…

I had some discussion with an ASA customer about tracking problems with his small Takahashi 500mm and I think that this might also be helpful for other astronomers going through the same problems.

I have put the email exchange in a PDF which can be downloaded here:

http://www.astrooptik.com/asa/Manual%20Optimizing.pdf

I had to hear recently all kind of harsh critics about the tracking, pointing files etc. Let me allow some comments:

The models and used formula work since 10 years on 100’ eds of telescopes now. Some telescopes perform better, some worse. If there are problems with tracking accuracy it makes more sense to take care about the quality of the input (how accurate and reproducible are my measurements during making of the pointing file) when to think that the software does it all wrong.

The funny thing is that some years ago people tried to guide with guide telescopes. At some point they accepted that you have flexure between the 2 scopes and thus the exposure time is limited. Or they tried to make this connection as stiff as possible by all means. Then maybe if they wanted long exposure times they turned to off-axis guiding to get rid of this problems. Today, people buy a mount with very good tracking abilities and expect, that it also takes care of all the same flexure problems they had before with their guide scopes. But trying to blind guide with a precise tracking mount means that you are going all over these flexure problems again. It’s true that to some degree especially if they are reproducible, they can be corrected by software. Software can very well correct polar alignment errors, collimation problems and also flexure that happens regular (in form of sinus waves or multiple of them). But large flexure nearly always means large hysteresis. Hysteresis (pointing errors depending from WHERE you come) can NEVER be corrected by any software.

ASA mounts are better than any guide scope because they can to some part model this flexure that guide scopes can’t but there is some limitation, they can’t perform miracles.

 

There really are limitations how accurate things can get. The Autopointing Grid that Sequence creates for you is maybe showing to small errors (compared to reality) because the mount always starts in East and then continues West in a repetitive and regular pattern. Thus, also your pointing errors will show maybe better results compared to real life (where you slew your objects in a random pattern coming from different sky locations).

 

The next myth people like to believe in is that things get much easier with very short focal lengths. This is simply not true because the absolute focuser shift in mm causes the same star trail length in terms of pixel. Sometimes longer focal lengths are thus easier to track because the seeing will blur our small tracking errors whereas with 500mm focal length you will even see sub-pixel elongations. What exposure time can you expect ? I already wrote something to this subject 1 year ago in the pointing manual of the DDM85 and DDM60

http://www.astrosysteme.at/images/DDM60_PointingModel_Manual_E_091124.pdf

Limitations of pointing models
You can only fit pointing errors if they are reproducible.
This means that the pointing error must be the
same at a certain location in the sky no matter if the
telescope slews to this point coming from deep East
or far West. You can test the repeatability of your
scope by synching on a star and then slew again to
the same star coming from different positions in the
sky. If you are more than one arc minute off in this
test, try to optimize your telescope (mirror cell, focuser
etc.). Your final pointing will always be as good
as your telescope. In 99% of all cases where people
complain about bad pointing the problem is in the telescope
and not the mount. We can build mounts with
5arc sec RMS pointing but with a Schmidt Cassegrain
put on this mount with 1arcmin mirror shift you will
never reach the mount performance! However, please
accept that there is NO telescope in this world that has
zero hysteresis. Especially reflectors will always have
a little mirror shift and if you try to fix your mirrors
until nothing moves you might have the mirror bent in
its cell and have good pointing but bad image quality.
Try to be realistic with your targets. If you can reach
5m unguided then it’s often easier to stack more CCD
frames than to try hard to get up to 20m unguided. I
remember one 1m telescope from an Australian company
where the mirror was GLUED to the mirror cell.
This telescope reached 2arcsec RMS pointing but at
the same time the image quality was a disaster. Since
the telescope was only used photometric, there was
no problem.

 

I still believe that 5 minutes unguided is a goal that is realistic and makes sense. Trying to reach much more will nearly always end in a fruitless and frustrated search for the error sources which will nearly always be within your TUBE and not the mount.

 

This does not mean that in some cases you cannot succeed to make 45m unguided. I did this several times but I could not rely on it that it would work on the next day with different temperature etc…

 

A very common specification for professional telescopes where a 1m costs around a million euros is 0.3 arc sec. RMS in 5 minutes. RMS is always a nice specification since it allows 1 arc sec. PtV deviation or even more and still can be within these tolerances.

 

ASA mounts can do 0.25” RMS in 5 minutes and I can guarantee you that they are much better and we could prove that with ANY of these mounts with a suitable test telescope. We are not able to specify what happens on top of this mount, how could we?

 

 

The tracking results show what can be expected from the pure mount if you have absolutely no hysteresis or temperature problem in your scope but in my view it’s totally unrealistic to expect this on a regular basis with “normal” amateur telescopes.

With today’s CCD with 6 micron pixel size and short focal length you can accept maybe 3 micron trailing. That is 1/300 of a 1 mm. Now please take a look at your focuser chain, mirror cell or whatever and tell me that with your telescopes things will not most likely move 10 x more than that. Please also note that there are seeing conditions which are not isotropic which means that you have seeing influenced image shift that is directional and causes elongated stars (jet stream conditions, slowly rising thermal bubbles). Dome seeing AND tube seeing is nearly always directional and will cause elongated stars. Try to find out with different exposure times if this might be the case.