Project - 10" F4 Newtonian Telescope on Losmandy G11 On-Step Mount

Since I switched to new OnStep hardware (Arduino Mega2560 R3 based) I wanted to check out the tracking. I took an image of M51 first, programmed PEC, then took an images of M13 and M5. These were shot in color using my Sky Planetarium and the modified DSI-II Pro camera with it's integral filter wheel.

Each image was a stack of 20 subs of 20 seconds ea. in LRGB.

It was an overcast moon-lit night, so I worked on PEC again. This time I recorded the PEC data using OnStep's built-in recording mode while Sky Planetarium did the autoguiding. Up to now, most serious PEC programming has been done by recording in K3CCDTOOLS, processing in PECPREP, then uploading to OnStep. So this is a different way to accomplish the same thing. I'm pleased to report that it worked very well. I got clouded out early, but before that happened I captured the lumensence data for M97. No "low quality" frames were thrown away to form this image.

M97, 10 x 40 seconds, capured in Sky Planetarium, stacked/log-stretched in Autostar IP, touchup/hotpixel removal in Photoshop:

I knew that more data would give a better PE correction so I measured 6 cycles of the worm and programmed in the resulting curve (from PECPrep). Again, the curve appearance and phase were similar to those that came before. I wrapped things up by actually imaging something to see the results of my work. These images are unguided, you can see that my polar alignment is a little off since the stars are bloated along the Declination axis. Along the RA axis things look ok, but could be a little tighter in the three minute exposure. Also, the optical collimation is off towards the lower-right corner of the image. I'll fix these things next time I'm out there and get a clear night and at the observatory, that's the downside to star gazing 250 miles from your telescope. I have another tracking accuracy problem, but it's of a very small magnitude (couple of arc-sec) and long period. Could be OnStep or could be something flexing, hard to tell. I'm shooting to get every last bit out of PEC and my OnStep's tracking then add autoguiding to clean up what little mess remains. Anyway, I'm quite pleased with these results... my LXD75 at less than half this focal length couldn't image past a minute unguided. Of course, it doesn't hurt that I was lucky enough to purchase a well cared for used G11 mount with such a wonderfully accurate RA worm gear.

M51, 12 x 60 seconds, capured in Sky Planetarium, stacked/log-stretched in Autostar IP, touchup/hotpixel removal in Photoshop:


M51, 1 x 180 seconds, capured in Sky Planetarium, converted to jpg - otherwise untouched:

I went ahead and made numerous improvements to the OnStep PEC code, again. I was hesitant to make changes since it was working, but I knew better performance was still possible, read about the firmware changes here. The ability to easily visualize and to import/export the PEC curves from Sky Planetarium was also added, a screen shot is here. So last night I got a chance to try it out, I can now image out to a minute at 1016mm FL without throwing away any frames basically. At ninty seconds star images were showing just a little RA bloating, but my tracking rate is a little slow and fixing that will help. This was my first try at using PECPrep to generate a tracking curve and to import it into the OnStep, it worked well. I actually measured the PE the night before, parked the mount, then last night I measured it again and had a very similar curve... The phase was correct too which means the steppers aren't missing many steps (hopefully none) and the parking code is indeed keeping track of the worm index. So I imported the curve into OnStep and turned on tracking with good results. Still room for improvement and I'm guessing that I can reach ±2 arc-sec or so tracking once everything is as it should be...

Before:


After PEC is applied:

Finally tested out the PEC feature again. I'd been waiting for a clear night, for quite a while, to do some more testing and this was it. After locating and fixing some timing related problems, etc. in the OnStep I was really looking forward to seeing how it would work. I'm happy to report that this definitely improved it's performance. After measuring periodic error at ±7" one round of training and PEC reduced this down to ±3.25" (4/3/13: most of this error disappeared when I moved the counter weights to east-bias the mount). I'll add PEC record averaging and fix a bug that kept me from writing the results to EEPROM and next time around I'll do a more comprehensive training session and see what I've got.

Had a fairly clear night and tested out the PEC routines, again. After correcting two minor bugs in the firmware, it appears to be working correctly now... I recorded the PEC data by using the autoguiding feature in my software. Playing back PEC, the tracking appears to be markedly improved. I still see random jumps on my images, but I'm betting that this is not due to my OnStep, but rather to the closely balance state that mount is in. I got a two minute exposure with round stars in one image (unguided 1032mm FL), and then the mount would do a big jump (like 20-30 arcsec) very quickly (as in a few seconds, max). I'll take a look at the G11 next time I'm up there and shift the balance so that it's east biased, this keeps the worm pushing the worm-wheel and results in better performance from what I've read. Next up, I want to add another PEC recording mode to the OnStep. A mode that averages the existing PEC data with the new data as it's recorded. Then, I'll do a couple of good rounds of training and take some before and after measurements to see what I've got.

Still fighting to get PEC to perform to it's full potential... I corrected a major timing related bug which should help. I also ended up changing the way that corrections are applied, they're now channeled into the routine that provides the sidereal tracking effect. This allows for smoother playback of the sidereal corrections, they are added to or subtracted from the tracking counter. I also went over the code to reclaim some flash memory, again.

I've changed the On-Step firmware so it writes PEC values to static memory instead of EEPROM during recording. This is for performance/stability reasons. The EEPROM writes are very slow and I feel they might be causing some problems. I also added a command to transfer the recorded values (and PEC state) to EEPROM so they are retained on the next restart. A nice side effect of this is that you can record and play a new PEC dataset and still easily go back to the old one if things don't work out. I still need to test these changes. Another shortcoming of the PEC code is in how it issues the corrections. They're played back in bursts: east corrections will halt stepping for a moment, while west corrections will be played out at 1.5x the sidereal rate. This could be coded so that the east and west corrections are applied evenly (one step at a time) over each one-second interval.

Finally, another clear night... First, I successfully tested Sky Planetarium's ability to submit jobs to and get results back from Astrometry.net's online plate solving services. Next, On-Step now remembers tracking rate adjustments so it doesn't need to be set each time I start an imaging session. Then, I recorded the PEC data again (while autoguiding). The session didn't go very well... the On-Step controller locked up at the end of the recording. Restarting things, it was playing the PEC data back, but tracking wasn't really improved. I have some ideas about what might be wrong with the On-Step's PEC code and I'll fix those and try again. Finally, I quickly imaged M57 in color just to have something to show for the night. This image could look better with more processing, but I'll wait for the MPCC and PEC improvements and image it again. The DSI-II Pro camera with my homebrew filter wheel worked wonderfully, it was trivial to capture this image. I haven't imaged in color much since starting these other projects and the magnetically indexed filter sensing I added to the camera hasn't been used in a while, it's nice to see that it's still working well.

As a side project, in the middle of this, I really want to get my Baader MPCC integrated into the optical train. Unfortunately, I didn't know the correct spacing/distance from the CCD with my custom camera. So I did a CAD drawing with the relevant elements of the assembly to work out the distance. Reading online, I also noticed that filters in the light path cause a reduction in the required distance of approximately 1/3 of their thickness. For my application, that subtracts 1.67mm from the distance leaving me with a required spacing of 22.39mm. I can get close to that using my 7.5mm + 15mm spacers giving 22.5mm. The final issue is that my 2" focuser needs to clamp down on that last spacer, and it's OD will have to be shimmed up by 0.19", I plan to use an aluminum tube that fits over it snugly. The images of M33 and NGC 7331 (below) show a fair amount of coma in the corners and this should help clean that up.

Last night everything was working very well... I opened the roof up and unparked the 'scope. It slewed directly to M27, which it placed on the (smallish) CCD no problem. I was just about to start imaging when the DSL connection was lost. Since the observatory is eight hours away (round trip) fixing little problems like not being able to close the roof can be a challange... The DSL finally came back on this afternoon (the ISP was having issues) and I parked the 'scope and closed in up. That's the first time it's been left open during the day, weather was good so no harm done. Someday I'd like to set things up to automatically park the 'scope and close the roof at dawn, this kind of thing happening causes too much stress!
Now I'm back to waiting for clear skies again.

Friday night/Saturday morning I collimated the 'scope, did a polar (drift) alignment on the mount, and installed the Ruggeduino (see below.) I even got to image for a moment before calling it a night.
First, M33 (5 x 30 second subs, just stacked - no darks or flats, touchup in GIMP):



And, here is NGC 7331. I checked the limiting magnitude of this image against the USNO's NOMAD catalog, the faintist stars are around Magnitude 18 (again 5 x 30 second subs, as above), both of these images were taken before switching to the Ruggeduino:

The frequency accuracy/stability of the Arduino UNO I'm using for the On-Step is suspect in causing problems. So I could hack the board and add a crystal oscillator or just replace the board. I have an Arduino Duemillnova (they have a crystal oscillator) which would work, but the USB to serial converter on those always reset the device when I connect using the Misrosoft Serial Com API's, and that really slows down the debugging cycle while I work on Sky Planetarium. So I decided to spend the $40 bucks and get a Ruggeduino. This UNO clone has the form-factor, crystal oscillator, and debugging friendly USB to serial converter that I want. And the ESD protection is nice too.

In preparation for actually starting to image again, I've purchased a faster hard drive (a Western Digital 600GB VelocityRaptor 10K RPM.) This is mainly for improving the speed of plate-solves. I tried to use a Muskin Chronos SSD drive, but that lasted about a month before dropping dead. I just had it setup to handle the plate solving datasets... reading data only... I think I'll stick to mechanical hard drives for now.

I just finished testing the PEC feature again, and it seems to work although I believe I'm seeing the tracking rate drift some due to temperature changes affecting the resonator frequency. I'm not sure what the spec. is for the Ardunio UNO I'm using, but most resonators seem be at least a few 1/10's of a percent over their temperature range. And I'm using it over a fairly wide temperature range. When I get a chance, a simple hardware change should fix that problem. But anyway, I programmed the PEC again after making the changes indicated below, and it seemed to track much better. No crazy jumping around, just seemed to work. The mount tracks at about ±5 arc-seconds without PEC and I'll have to do a drift analysis to see if any improvement is made with the PEC turned on.

This morning, I finished adding the PEC controls to Sky Planetarium: better PEC status indication, better control layout, etc.

My first test of the PEC feature with my On-Step controller found some bugs. At the start of the night I added some temporary controls for PEC to Sky Planetarium and proceeded to gather "before" drift data in K3CCDTOOLS so I have a baseline to compare to. Once that was done, I turned on the autoguiding in Sky Planetarium and set the PEC to record mode. The PEC routines in the firmware fall right out of record mode and start playing (uninitalized) PEC data so the 'scope starts jumping around. Well, at least I know that it can try to play the PEC data! The problem seems to be with how the On-Step firmware keeps track of the worm zero index; It's getting falsely triggered by the guiding commands. The zero index (on the worm) marks the start of a timed sequence of recording/playback of PEC data. PEC recording starts with the firmware waiting for the zero index. Once it arrives recording of data actually starts... Once a second the sum of any guiding corrections (in steps, caused by :Mgdnnn#, etc.) are tallied up and stored in the controller's flash memory. When the zero index arrives again (in this case 240 seconds later, up to 884 seconds max) the recording stops, and playing begins. So it looks like the false triggering is causing it to fall right from waiting for the zero index to recording to playing. So, I've changed how it keeps track of the zero index and have my fingers crossed (fairly complex stuff, multiple variables and timing interplay), testing tonight or tomorrow night.

I'm getting close to having everything working. My On-Step telescope goto drive controller is coming along, everything except for PEC and two/three star alignment has had some basic testing done and appears to be working. I fixed several bugs related to: alignment, sidereal tracking speed, and the pulse-guiding commands.

So, my first remote imaging session with the On-Step G11/10" Netwonian/DSI II-PRO camera combination was mainly about getting it aligned with the sky and adjusting the sidereal tracking speed. I did manage to image M13 (in mono), but I need to spend some time getting the 'scope collimated and polar aligned before going further. Below is my first image with this setup. You can see some miscollimation, as the optical center of the telescope is in the top-left corner where the stars are nice and round. Also, I had to keep the exposures short (twenty, two second subs) since the mount is only roughly polar aligned. The round stars are a good sign that the stepper motors/driver are indeed tracking with enough precision.



The stars in this image measure about 3.2 arcsec (seeing limited), so I know my drive's minimum tracking precision is in that range.

I opened up the observatory and did some testing of my telescope GOTO drive system (On-Step). It Unparked/Parked fine and the whole session went without a hiccup. I'm still debugging On-Step and worked on a tracking rate problem it has. I need to make a simple hardware change to improve it's accuracy, but in the meantime I added code that allows the tracking rate to be adjusted on-the-fly to compensate for temperature changes/etc.

I had a bit of a scare when this camera stopped working. Turns out that one of the USB ports on the observatory computer was shot, switched to another and all is well. Now I'm back to waiting for good weather.