
How to "measure" periodic error?
#1
Posted 16 March 2009 - 10:27 AM
I used my Atlas EQ-G with the Orion 102ED f/7 scope this weekend to shoot my first set of astro pictures (will post some results here at a later time). However, since I don't have an Auto-guider setup and I heard a lot of good things about the Atlas I figured I'll see how long the mount can track accurately and was a little surprised to only get relatively short exposures. At 60s I had to throw out almost half of the exposures due to some star trailing (in RA direction), 30s exposures consistently looked good, except for a few. I also took some 120s exposures and also had to throw out at least half. Not quite what I had in mind. Did I expect too much here?
Anyhow, I drift aligned the mount to the best of my abilities actually using the DSLR since I also don't have a cross hair eye piece, yet. I used the technique where you expose for 5s to mark the star and then move the mount forward in RA for about 60s at twice the siderial rate and then essentially stop the tracking for another 60 seconds, all while the shutter is open. The result is a V shaped line in the image if there is any misalignment. Worked like a charm and I might actually perform the alignment this way in the future instead of using the eye piece. I adjusted the mount as needed and got no more drift in the image for up to 3 minutes.
So, to make a long story short, the only reason for the star trails that I can think of now is RA tracking errors in the mount. I'd like to actually "see" the periodic error, etc. somehow in an image but can't quite figure out how I would go about doing that. Do you guys have any suggestions?
Thx in advance,
/ThJ
#2
Posted 16 March 2009 - 11:14 AM
Take a series of short exposure images (may need a brightish star) that totals longer than the period of the worm (typ 10min).
Use a stacking program that measures and records (to a file) the x,y coordinates of the star (the program should find the star's centroid). AIP4WIN does this.
Import the recorded coordinates into Excel (or another spreadsheet program) and plot the x and y values vs exposure number. The PE will easily be seen in the plot.
Some calculation using the scopes focal length and the pixel sizes will give you PE in Arcsec.
If you align the camera so that RA is along the pixel rows (x-coordinate) then there should be no movement in the y direction if your polar alignment is perfect. Any change in the y is polar misalignment.
I have a spreedsheet at home from my Super Polaris mount. Let me know if you need more help on this part.
#3
Posted 16 March 2009 - 11:24 AM
Many folks who use autoguiders with the Atlas/EQ6 say they don't bother with PEC training.
P
#4
Posted 16 March 2009 - 12:14 PM
I'd like to actually "see" the periodic error, etc.
Thomas, If you would like to see just what your PE is, intentionally setup your mount about 5° (angle is not critical) off polar aligned. By setting up "OFF Polar", what you are doing is introducing drift into your image. Now take a image that is at least one worm cycle. The image that is displayed will look like a series of saw teeth. It is the step of the saw tooth that is your PE. This is a "Real" time graph. Now how to measure that error,

#5
Posted 16 March 2009 - 12:20 PM
Another thing to look for is a smooth 'sinusoidal' shape to the periodic error. This is easier to guide out and also indicates that the mount is operating smoothly. If the PE shape is jumpy or erratic then the mount could use some tuning up.
#6
Posted 16 March 2009 - 01:35 PM
Thomas, If you would like to see just what your PE is, intentionally setup your mount about 5° (angle is not critical) off polar aligned. By setting up "OFF Polar", what you are doing is introducing drift into your image. Now take a image that is at least one worm cycle. The image that is displayed will look like a series of saw teeth. It is the step of the saw tooth that is your PE. This is a "Real" time graph. Now how to measure that error,
I see what you're saying. I'll have to try this! It would be great to get a squiggly line of some sort to see the PE. I guess if I know the arcsec/pixel (should be fairly straight forward to figure out) I could actually "measure" it too.
Where should the star that I image be? Near Equator? Meridian? East Horizon? Does it matter?
#7
Posted 16 March 2009 - 01:43 PM
Makes sense, I'll try to do that.That will work Carl. To measure the PE just measure the amplitude of a 'saw tooth' and convert to arcsec.
Good point. I'll definitely look for that. I do have some play in DEC but RA seems fairly solid (no backlash or play that I noticed visually). To avoid the DEC play I simply balanced the mount slightly heavy on one side and it hasn't really given me issues but I'm not opposed to the idea of taking the mount apart to tune it if needed.Another thing to look for is a smooth 'sinusoidal' shape to the periodic error. This is easier to guide out and also indicates that the mount is operating smoothly. If the PE shape is jumpy or erratic then the mount could use some tuning up.
#8
Posted 16 March 2009 - 05:02 PM
Where should the star that I image be? Near Equator? Meridian? East Horizon? Does it matter?
It probably doesn't matter. But if it was me, I would be somewhere near the equator & just on the West side of the meridian. Each worm cycle will crate a saw tooth, so the longer your image is (worm cycles), the more teeth will show up. Post your image so we can see what you have.
#9
Posted 16 March 2009 - 06:21 PM
Calculate how long it'll take for the star to move from top to bottom of the image, at this tracking rate. Start the mount to tracking, open the shutter, then push the 'down' button on the hc. Hold the down button till the star should be at/near the bottom. Release the button, close the shutter, now open the shutter for the next exposure, and push the up button for the same amount of time. Repeat until you have gone a full worm cycle.
The resulting frames should allow for lots of measuring.
My 8 inch sct has a fov of 25x38 arcminutes with the canon 350 in there. Allowing for a little slop around the edges, it's 2300 pixels high, so I'll go about 2000 pixels per shot, and it's 0.66 arcseconds per pixel. 2000x0.66=1320 arcseconds useable in the frame, with a tracking rate of 1.875 means I can run about 700 seconds (just over 11 minutes) at this rate.
Low and behold, even with my long focal length and small fov, I can pretty much measure a full rotation in a single frame, without messing anything up in terms of alignment.
I'm gonna have to dig out an hc, and double check the slowest vertical tracking rate I can set with the hc, but this may actually be a very simple way to do this measurement.
#10
Posted 16 March 2009 - 06:33 PM
This may require a little program that talks to the synscan and sets a 'properly low' dec tracking rate, but, it's not really difficult to do.
#11
Posted 16 March 2009 - 11:15 PM
However, since I don't have an Auto-guider setup and I heard a lot of good things about the Atlas I figured I'll see how long the mount can track accurately and was a little surprised to only get relatively short exposures. At 60s I had to throw out almost half of the exposures due to some star trailing (in RA direction), 30s exposures consistently looked good, except for a few. I also took some 120s exposures and also had to throw out at least half. Not quite what I had in mind. Did I expect too much here?
As noted above, the focal length you're using will directly affect how long an exposure you can take without star trails. On my CG5-GT mount, I can take a 1 minute exposure at 320mm f/l and get nearly 100% good images. My average PE is a little better than +/- 20 arc seconds.
I'd like to actually "see" the periodic error, etc. somehow in an image but can't quite figure out how I would go about doing that. Do you guys have any suggestions?
The best way I've found to measure PE is with a webcam and two programs called K3CCDtools and PEAS. First, K3CCDtools is used to collect the raw PE data into a file. Then the file is imported into the PEAS program which generates the PE curves. Once you figure out how to use the programs, it's relatively straightforward to develop the PE curves.
K3CCDtools is not free, but it is a good webcam tool and is also useful for planetary photography. The PEAS program is completely free. The documentation that comes with K3CCDtools explains how to create the data file...you can read up on it in the on-line manual (found under 'Downloads') before you buy the software.
Here is an example of my CG-5GT mount's Periodic Error captured as described above.
Patrick
#12
Posted 17 March 2009 - 02:34 AM
OK, here's what I did tonight: I roughly set up the mount and then extended one side of the tripod to make it point a few degrees above the pole. I then pointed at a star roughly at the equator and towards my Eastern horizon.
Here's what I got...Post your image so we can see what you have.
I took a few exposures and the PE is really obvious now. The first image (top) is two 30 minute exposures and the second (bottom) is a single 64 minute exposure. The rate at which the star drifted slowed due to the fact that the mount was misaligned and slowly drifted North of the Equator. Either way, the PE shows up nicely as ~8 minute long "waves".
As for the amount of RA error. I am (very roughly) guesstimating around +/- 25 arcsec from peak to peak based on my chip size, number of pixels and FOV I get with the camera.
#14
Posted 17 March 2009 - 05:53 AM
#15
Posted 17 March 2009 - 05:59 AM
#16
Posted 17 March 2009 - 06:16 AM
http://eq-mod.source...et/ppindex.html
#17
Posted 17 March 2009 - 09:30 AM
I don't know anything about your mount, but does it have PEC training? If not your PE looks to be easy to autoguide away.
Yes, the Atlas does have PEC training. I'll definitely give that a try when I get a chance. We'll see what difference that makes.
#18
Posted 17 March 2009 - 01:09 PM
Or a quick and easy method is to use EQAlign (free) and a webcam.
jc
#19
Posted 17 March 2009 - 03:37 PM
#20
Posted 05 May 2009 - 12:42 AM
Actually PEMPro does handle uploading directly to EQMOD as well as external playback to a hand controller while PEC recording is active. I am the author of PEMPro and I went through great pains to make it very easy for users to acquire PEC data, create a useful PEC curve, and upload it to EQMOD. In fact I even dove into the source code of EQMOD and fixed a few bugs in the external PEC interface and the user interface of EQMOD. There are detailed instructions in PEMPro's help and online at the link below.Atlas mound can't handle Pempro training directly but you can do it with Eqmod and Pecprep:
http://www.siriusima...ProV2/eqmod.htm
BTW, PEMPro's 60-day free trial should be long enough time to measure and program your PEC, so PEMPro is essentially as free as PECPrep! :-)
-Ray Gralak
#21
Posted 05 May 2009 - 01:07 AM
