| The Right Mount for the Job
Over the last several years, I - like nearly every amateur astrophotographer out there - have been looking for ways to improve the quality of the data in my astro images. Moving from film to a digital SLR was the first major step and moving to a CCD camera - first one shot color, then monochrome - has yielded further improvements. My decision to go with a KAF-8300 CCD based camera in 2008 has allowed me to really get the most out of my optics on nights of good seeing, but the 8300 sensor's very small 5.4 micron pixels have presented some major hurdles for mounts I have used in the past and led me to eventually choose the Astro Systeme Austria DDM60 Pro mount.
The Losmandy G-11 mount I had in 2008 when I got the FLI ML8300 camera was a fairly average sample - measuring around +/- 5 arcseconds of periodic error without guiding. Unfortunately, the Gemini computer I had had older electronics (upgraded from version 3) and I could never get the mount's PEC to work correctly, so guiding was the only way to improve the tracking. I was the third owner of this mount, and while it worked well with shorter focal length instruments, I seemed to always have issues guiding at longer than 600 mm dus to an issue with the RA worm having non-periodic jumps of around +/- 1.5 arcsec. Had I chosen a CCD camera with larger pixels, I might not have noticed the error or I might have been able to process it out easier, but rather than overhall the mount or buy an aftermarket worm fix, I picked up a used Takahashi EM-200 Temma Jr. and sold the G-11 in early 2009. I have spoken to many others who are perfectly happy with the performance of their G-11 mounts, so I know that the results I saw were not typical for the design.
I admit I bought my way out of a problem rather than fixing it (new gear is ALWAYS the solution, right?), but the EM-200 really was a step up in performance so I did not feel like I was wasting my money. I measured the unguided tracking PE at +/- 4 arcsec, and with good guiding I would routinely get performance around +/- 0.7 arcsec. Images with my TSA-102 refractor now showed very tight round stars on most nights, but I noticed that the mount was sensitive to balance and wind more than the larger G-11, and I was concerned about future use at long focal lengths. In late 2009, a used Takahashi NJP Temma II came into my local shop and lasted about 2 days before I traded my EM-200 and Takahashi Epsilon 130 in to cover the upgrade cost.
The NJP was hands down an amazing mount. Unguided PE was +/- 3 arcsec, and guiding got the tracking down to +/- 0.5 arcsec. I was now able to image with the new Celestron 8" EdgeHD scope I had picked up for longer focal length galaxy and nebula imaging. Images with the 4" Apo were consistantly excellent, and despite the fact that the mount was a beast (55 lb eq head and nearly 85 lbs in the Pelican case made it hard to move around), I was pretty much content. With a 75 lb maximum payload capacity, I had plenty of headroom for heavy imaging systems. Any errors I saw in the mount's performance were usually attributed to seeing conditions, which rarely caused issues below 1000mm of focal length, but occasionally I noticed that the mount would "chase the seeing" and inadvertantly ruin some of the images taken with the 2000mm focal length EdgeHD. The atmosphere at my observing location can be unstable at times, and no matter how good the gears were in the NJP my guiding accuracy was at the mercy of the local conditions and the term "seeing limited" took on a very clear meaning. The solution was to jump into a mount that truly broke the mold on how amateur equipment operates and what it is capable of - the ASA DDM60 Direct Drive mount.
What is Direct Drive?
Almost all amateur equatorial telescope mounts made today and for the last 150 years have used some variant of a Worm/Wheel Gear configuration to drive the RA axis at the sidereal rate. Gears are imperfect things, and there will always be high spots and low spots plus some amount of surface roughness at the point of contact that causes errors. Errors on the Worm are largely periodic and repeat in the same place in every worm rotation (usually every 4-8 minutes), but the Wheel Gear meshing to the worm adds its own errors that repeat every 24 hours. Backlash, dust and balance issues add further error into the equation. Because only 2-3 teeth are in full contact with a Worm/Wheel drive, autoguiding is subject to backlash and occasional delay in addition to seeing concerns. One can make a very good Worm/Wheel system, but the gears must be very large to minimize error and the polar alignment must be flawless to avoid needing to autoguide over reasonable (10-30 minute) exposure lengths. This accuracy is almost never accomplished in non-permanent setups.
The ASA direct drive mounts have no gears at all. The RA and Dec axes are actually the motor shafts and electromagnetic clutches are used to hold the mount in a certain spot with minimal friction and next to zero backlash or "contact error". Also, whereas most equatorial mounts have encoder resolution measured in the thousands or low tens of thousands of tics (the minimum resolution of a mount's position), the resolution of the DDM60 mount is 0.03 arcseconds on the axis, or approximately 43,200,000 positions, known as "tics" (the larger ASA DDM85 mount has an even higher resolution of nearly 67,000,000 tics). The exact encoder position is read out by the included Autoslew software over 100 times per second, and the motor movement of both axes is then adjusted to track to better than +/- 0.3 arcsecond and compensate for polar misalignment, wind gusts, and atmospheric refraction. Autoslew was written by Dr. Phillip Keller for professional observatories with 1-2+ meter scopes, so while it is still a work in progress meshing it to the ASA mounts, it is very powerful software and has proved itself capable of exceptional tracking for portable field use. Sequence, another included piece of software included with the ASA DDM series mounts, uses the Pinpoint plate solving engine to quickly align the mount to dozens of stars using my CCD camera through MaxIm DL and I can then load the models into Autoslew for use there. Autoguiding becomes effectively irrellevant, meaning less equipment and cables on the mount and no issues with "chasing the seeing" on marginal seeing nights.
Direct drives have usually found applications in high end CNC machines, airplane control systems, power generating turbines and robotics. Recently, many professional observatories have been turning to direct drive systems due to their increased accuracy and durability compared to traditional geared drive systems. Historically, direct drives have earned the reputation of being power hogs - drawing several amps or tens of amps during normal operation - but ASA's newly designed motors draw only 0.6 amps with the DDM60 Pro for sidereal tracking. A 10 amp AC power converter is a must with this mount, though, as it can briefly draw 6 amps or more while actively correcting for wind shear. A multi night trip away from home still means bringing several car batteries or planning ahead for an AC hookup at my destination.
The keys to making a mount like the DDM60 work are very accurate modeling of the night sky, extremely fine motor control, and very high encoder resolution. The encoder resolution of my mount itself must first be defined, and the the specific locations of several stars must be factored into the model. Points need to be limited to one hemisphere per model (East or West) when adjusting for polar alignment, but an all sky model can be run afterward in Autoslew or Sequence to allow all-sky coverage and tracking. Newer DDM60 mounts have factory set encoder resolution, but on my mount I need to be careful when adjusting or saving that setting.
Once the ASA mount initializes the drives (done immediately after Autoslew startup), I have the mount search for it's encoder home position on each axis. The mount needs to be pointed fairly close to the home position to locate it properly, so on a new startup I have to move the mount to a certain orientation before telling it to start looking. Once homed, I can sync the mount to TheSky (which I use as my planetarium program), load up my camera in MaxIm DL, and start up and sync the camera to TheSky in Sequence. Once that's done, I can set up a 20-30 point hemispherical pointing model in Sequence in one hemisphere and then solve the "plates" by matching the stars in each image (done mostly automatically in Sequence) and convert the data to an Autoslew compatible pointing file. Autoslew can then take that file and apply it to its encoder resolution to predict pointing position and tracking errors over that whole hemisphere of the sky. If this is the last step before imaging, Autoslew will also use this data to adjust the motor tracking rate in both RA and Dec to compensate for the predicted polar misalignment, cone error (where the scope and RA axis are out of parallel when aimed at the north pole), atmospheric refraction, optical tube and mount flexture (very little of the second), and it will correct for wind blowing the DDM60 off of its intended path. I have tested this myself by achieving round stars in unguided 10 minute exposures at RTMC in 2011 with a 30 mph crosswind. I was shocked at this high level of performance, as I have not heard of any other mount in the price range of the ASA DDM60 that can do this.
After a good model is applied to the sky, Autoslew can then predict and assist in correcting the DDM60's polar misalignment. I tell the mount to slew to a star high in the south (~30-60 degrees above the horizon) and in the same hemisphere that the mount was aligned on (so just east of the meridian if I aligned in the eastern sky), turn on the camera in focus mode, and make sure the star is centered and synced in the camera's field of view. Autoslew can then say how many arcminutes off from polar alignment the mount is is in altitude and elevation, move the star to the predicted correct position, and then I manually adjust the knobs on the mount to recenter the star. Once I have physically moved the mount, my Sequence/Autoslew pointing model is useless, so I clear the model in Autoslew, resync the mount in TheSky and Sequence, and shoot a new autopointing model in Sequence - usually with more stars this time. I will usually correct the polar position at least twice if I am setting up in the field, as refining this will get me more consistent performance. If I am only taking short exposures in the 10 minute range, I am done as soon as I have a polar alignment I like and one last 40-50+ star model loaded into Autoslew. When applying the final model, my calculated position error (the error in pointing that I would see were I to slew to a target at random) should be less than 0.25 arcminutes (15 arcsec) per axis. If I plan on shooting longer 15-20 minute exposures or setting up a 3-5 hour imaging run on a target, though, the next step is to set up an MLPT run.
MLPT - Multi Local Point Tracking
MLPT is a new feature in Sequence (as of late 2011), where the software takes a series of images along the exact path it will be tracking a target for the night. After centering the object I am going to shoot that night (using TheSky and MaxIm DL to check framing), I pull up the MLPT tab in Sequence and tell it how many minutes I expect to be shooting the target (always guess long, to have a safety margin) and how many images I want shot along the path (not necessarily the same # as my actual frames in MaxIm DL). For example, on a planned 4 hour exposure run on M51, I told MLPT to shoot along a 300 minute curve (5 hours, or 75 degrees of motion) and to take 24 images along that path. Sequence then moves the mount all the way to the end of the run to make sure there will be no mount position conflicts (like running into my pier by going too far past the meridian) and then back to the start position. Next, the DDM60 is moved to equally spaced points along the path corresponding to the number of exposures I have asked MLPT to take (in this example, every 3.125 degrees or 12.5 minutes of exposure time) and an image is taken with the camera to add to a new custom pointing file. As each picture is taken, a new point is added to a graph at the top of the screen for both the RA and Dec positions and the calculated tracking error from the start to that position is recorded.
Once all images have been taken and the scope is moved automatically to the object I plan on shooting again, MLPT applies a plate solve for those points and then calculates how applying a new tracking vector will improve performance using a Fourier Transform (hey, I only did 2 years of Calculus in college - the math is above my head so I'm not sure what is actually being done with the data here except that it works!). While I might have had a calculated 20-30 arcseconds of drift over 4 hours resulting from an imperfect polar alignment, once MLPT is done I often have residual errors of no more than ~1-2 arcseconds and an RMS (Root Mean Square calculation) error over any period shot (in this case, 12.5 minutes) of only around 0.1-0.2 arcseconds. To put that in perspective, that means a single pixel on my camera (around 1.4-1.9 arcsec/pixel depending on the camera used on my Pentax 125 SDP) is ~7-19x larger than the RMS error over any given 12.5 minute period of time! Jupiter, on average, covers a little more than 40 arcseconds seen from earth, so the mount is basically able to track reliably on ~1/200 of Jupiter's diameter and might drift a total of ~1/12 Jupiter's diameter over a 4 hour run barring outside interference.
That's not to say that every image I take is always perfect. Poor seeing, focus shift due to temperature changes, very slow flexture of the optical system (which MLPT can't always catch), and vibration from people walking or my roommates's dogs running on the concrete slab my scope is set up on can and do occasionaly cause elongated or imperfect stars in my exposures. Residual polar alignment error is also a factor, as field rotation is often what limits my maximum exposure length to ~20 minutes in a field setup. My long term plans call for investing in a permanent roll-off roof observatory, a very rigid computerized autofocuser, and large diameter machined connectors between my camera and scope to prevent as much flexture as possible and spending a full night or two dialing in my polar alignment as accurately as I can. I am confident that once this is accomplished, shooting 30 minute long unguided exposures will yield consitently excellent performance.
I am very satisfied with the performance of the ASA DDM60 Pro mount to date and plan on using it for many, many years to come as my ultimate imaging platform.
|ASA DDM60 Pro - Direct Drive Mount|
|First Light shot of M76.
12 x 10 min exposures unguided at full 1663x1252 resolution (binned 2x2 due to seeing).
Real resolution is 1.1 arcsec per pixel.
|100% crop of IC 1848 (lower left corner).
12 x 20 min exposures unguided.
Real resolution is 1.8 arcsec per pixel.
|NGC 7000 taken at RTMC 2011 w/ 30 mph winds
15 x10 min exposures unguided.
Not full resolution, but stars looked good at 1.65 arcsec per pixel
|Screen shot of Autoslew showing interface and menu for handling mount modelling and polar correction.|
|Screen shot of Autoslew showing calculation of a pointing and tracking file from plate solve data acquired in Sequence.|
|Screen shot of MLPT data generated from a 12 point model overlaid onto a larger pointing file. Notice the large decrease in residual and RM errors after the fit.|