Monday, March 31, 2014

RA drift might be flexure after all

Measuring Flex


There was a thread on Cloudynights recently that got my attention: http://www.cloudynights.com/ubbthreads/showflat.php/Cat/0/Number/6429454/page/0/view/collapsed/sb/5/o/all/fpart/all/vc/1

In the discussion, a user talks about a slow creep in RA (with some variance in DEC).  Sounds pretty familiar!

In the thread, a person linked to a review done by Craig Stark (of PHD guiding fame) where he looks at the Borg 50mm guidescope using the XY stage. http://www.sciencecenter.net/hutech/borg/guidescopes/attnov08.pdf  In it, he describes a test to determine differential flexure where he images a distant daytime object at one extreme of the scope position, then slews to the other side and images again.  The idea is that the differential flexure between the two imaging cameras would create a double image. 

I've considered that the issue I've been seeing has been an RA rate problem.  I went back to some older data that I've published before.  That data looks at two different plate solves pointing at two different locations of the sky. 

From Mar 22, I have some points where I can compare the plate solves from the SSAG and the K10D at different times of the session.

When the target is High in the sky
Mp22 3 (SSAG)  01:47

Solving using 54 image and 260 catalog stars
Matched 10 of 54 image and 260 catalog stars;
Average residual 1.3 arcsec; order 1
RA 13h 14m 58.9s,  Dec +18° 18' 15.8"
Pos Angle +357° 07.9', FL 418.5 mm, 2.56"/Pixel           

mar 22 mp 0012 K10D

Solving using 363 image and 1196 catalog stars
Matched 293 of 363 image and 1196 catalog stars;
Average residual 0.6 arcsec; order 4
RA 13h 15m 01.3s,  Dec +18° 17' 45.6"
Pos Angle +267° 20.4', FL N/A mm, 1.91"/Pixel


When the target is Low in the sky (note that this time is earlier than the High)
mp 22 2 22:58 SSAG

Solving using 44 image and 264 catalog stars
Matched 9 of 44 image and 264 catalog stars;
Average residual 0.6 arcsec; order 1
RA 13h 14m 59.6s,  Dec +18° 18' 24.8"
Pos Angle +357° 18.4', FL 419.5 mm, 2.56"/Pixel


mar 22 mp 005 K10D

Solving using 218 image and 1196 catalog stars
Matched 182 of 218 image and 1196 catalog stars;
Average residual 0.6 arcsec; order 4
RA 13h 14m 58.4s,  Dec +18° 18' 01.6"
Pos Angle +267° 26.4', FL N/A mm, 1.91"/Pixel

low
ssag    RA 13h 14m 59.6s,  Dec +18° 18' 24.8"
k10d    RA 13h 14m 58.4s,  Dec +18° 18' 01.6"
                    1.2s                23.2"

high
ssag    RA 13h 14m 58.9s,  Dec +18° 18' 15.8"
k10d    RA 13h 15m 01.3s,  Dec +18° 17' 45.6"
                    -2.4s                30.2"
                   
K10D
low     RA 13h 14m 58.4s,  Dec +18° 18' 01.6"
hig     RA 13h 15m 01.3s,  Dec +18° 17' 45.6"
                    -2.9s                16"

SSAG
low        RA 13h 14m 59.6s,  Dec +18° 18' 24.8"
hig        RA 13h 14m 58.9s,  Dec +18° 18' 15.8"       
                    .7s                    9"


During the above session, I had the K10D free - it was supported only by the T mount ring.  Also, I had the SSAG supported by the three point ring attached to a dovetail.  I since have examined the SV70ED mounting and have added an additional 2 inch drawtube extension (thus allowing the focuser to be retracted a bit for more support), and have added a locking screw to the dovetail on the three-point ring supporting the SSAG.  I noticed that it would move then pushed, so this is no longer a flexing point.

I went back to some old data where I supported the K10D just as I did at Calstar 2013.  I didn't have enough data to use from Calstar because of the way I was using the film camera didn't provide the chance to set many plate-solveable images.  I needed to find data sets where I had recorded an image from the SSAG before the meridian flip.  This would allow a comparison of any relative change over time for the SSAG vs K10D.

I finally found the proper sets from last August where I was working on IC1396:

ic 1396

ssag
low
RA 21h 39m 13.2s,  Dec +57° 28' 04.0"
Pos Angle +01° 58.6', FL 1013.3 mm, 1.06"/Pixel

high
RA 21h 39m 13.5s,  Dec +57° 28' 03.1"
Pos Angle +02° 17.5', FL 1013.9 mm, 1.06"/Pixel


k10d
low
RA 21h 39m 15.1s,  Dec +57° 26' 53.4"
Pos Angle +270° 09.9', FL N/A mm, 1.91"/Pixel

high
RA 21h 39m 20.2s,  Dec +57° 26' 40.3"
Pos Angle +270° 29.9', FL N/A mm, 1.91"/Pixel


ssag
low     RA 21h 39m 13.2s,  Dec +57° 28' 04.0"
high    RA 21h 39m 13.5s,  Dec +57° 28' 03.1"
                    -.3s                0.9"

k10d
low     RA 21h 39m 15.1s,  Dec +57° 26' 53.4"
high    RA 21h 39m 20.2s,  Dec +57° 26' 40.3"
                    -5.1s                7.1"


differences
ssag l    RA 21h 39m 13.2s,  Dec +57° 28' 04.0"
k10d l    RA 21h 39m 15.1s,  Dec +57° 26' 53.4"
                    -2.9            01' 10.6"

ssag h    RA 21h 39m 13.5s,  Dec +57° 28' 03.1"
k10d h    RA 21h 39m 20.2s,  Dec +57° 26' 40.3"
                    6.7s            01"    22.8"

                    


What this shows is:
Guiding with the SSAG did not allow much drift to occur - less than one arc second from the platesolve.  However, the K10D did allow some movement, nearly 5-7 arc seconds.  This is nearly 3-4 pixels and thus is enough to blur the image.

Also, notice the rotation of the field in the original plate solve results.  These rotations (IC 1396 features nearly 18 arc minutes of rotation and the March 22 set shows 10.5 arc minutes) are both from polar alignment issues.  Better alignment is the only way to solve this problem.

What to do:
As I am still waiting just another 24 hours for the replacement thermistor to arrive from Oven Industries, I will not take apart the K10D just yet.  Once it arrives, I plan to insert the sensor wire in place, replace the PWM controller for the Peltier devices, and then hot glue the floating CCD sensor into position.  I believe that this last step is critical, based on all the other efforts that have been done.  Once this is completed, I'll replace the K10D on the dovetail clamp to eliminate flexure from the Feathertouch focuser.

Two more efforts handled recently:
1. I spoke with Franck of Ovision on email and he says that the Ovision worm period is exactly 239.4 seconds.  This suggests that there might still be some residual from the fractions after that .4.  We'll see what happens with properly tuning PEC.

2. I also received replacement pads for the G11.  The old pads were scored, polished and thinner than they should have been.  I ordered two sets and put both in place to give a thicker amount, allowing better tightening.  I also noticed that the RA pad was indented by screws touching the face of the plate.  This is interesting and I may have to invest in a set of small washers to shim these screws to avoid the collision.  Once I replaced the pads, I was surprised and pleased at how smoothly the mount moves.  I'm a little concerned that it may not tighten down strongly enough, however.

Lastly, I want to check to see if ASCOM pulse guiding is reliable enough vs using the ST4 cable.  I stopped using the cable from the SSAG to the mount due to some issues with the pulse guiding causing bad calibration moves in Maxim.  There have been changes in the software and this might no longer be an issue.