- This topic has 4 replies, 5 voices, and was last updated 4:35 pm on 2015-12-08 by cfranks.
September 28, 2015 at 9:22 am #112171
Richard S. Wright Jr.Senior Moderator
While it is true that we have a very thorough users guide that details how to guide with a Paramount, recent events have led me (and others) to believe that a little tutorial or review of guiding best practices is in order.
Exhibit A to the right is my Veloce RH-200 with a Starlight Xpress CCD camera, off axis guider, and a Lodestar guide camera. Note that the cables are all contained and close to the equipment. Dangling cables get caught in the wind and can in fact pull on the OTA, or the cameras themselves creating deflections. An arc second or more is not unheard of. Cables down on the ground drag, and build up tension an then “pop” when they finally slide. This can result in a larger “jerk”, and will likely ruin a sub. I discovered this the hard way some years ago, and was blaming Raccoons (seriously) until I finally figured out what it was.
The stock Veloce does not have rings. The original configuration had a guide camera perched on top with a beautiful… but flexible… railing. I could not guide more than 5 minutes in that configuration without the flexure differences of the guide scope and the main imaging scope becoming large enough to start trailing the image. As an aside, ProTrack performance and TPoint accuracy were both greatly enhanced when I added the ring system as well. The whole OTA was flexing in all directions with the original mountings. Officina Stellare makes some of the best optics in the world, but I have to be honest, the original mounting mechanism left much to be desired. An even better ring configuration for larger scopes is to forgo the dovetail entirely and have rings that fasten directly onto the versaplate.
Off axis guiding is the gold standard (those things with the pick off mirrors). Guide scopes will flex, period. There will always be an upper limit to how long you can guide because of this difference in flexures between the two instruments. The characteristic symptom is elongation that increases with exposure time, and it may not be the same in all areas of the sky, and it may vary with temperature!
Guiding too everyone should understand is sometimes a necessary evil, but inherently it is a terrible thing to do. Really… it's like surgery with a chain saw, clumsy, but it can work. Look at the moon sometime with a high power eyepiece. See all that dancing and waving about the moon is doing? Your guide star is doing that too, and you've setup a system telling your mount so swing around 50 or so pounds to follow that bouncing about. Logically, as a thought experiment too… realize you are making a correction based on an error… after the error has occurred. The very best imaging results possible are always with unguided imaging. Using ProTrack and PEC helps you extend how long you can go unguided, but these have their challenges and limits as well, and this isn't the topic at hand here now. We often need to guide. We don't have time to model the mount, we have a lot of hysteresis (pretty much random flop or flexure) in our imaging system, you have non repeatable/non deterministic flexure (mirror flop for example, or a wooden tripod that swells and shrinks throughout the night) or well, maybe you don't have a Paramount. There's also narrowband imaging where you need to go for really long periods of time.
The first and most important thing you need for good guiding results is a good calibration. The image at the right shows two sets of calibration vectors after a “successful” calibration. If your guiding vectors look more like the top than the bottom, then something has gone horribly wrong. Most likely a hot pixel fooled the star search algorithm. Hot pixels can also cause the dreaded “star has not moved enough” message. This is counter intuitive to some users because they can actually see the star moving! The computer algorithm doesn't have the same neural net detection hardware that your brain does unfortunately. Sometimes too, another bright star can be moved into the field and create havoc. The easiest way to avoid rogue stars is to select a “lonely” calibration star. Put a large subframe around it away from other stars, and then don't move too much for your calibration steps. You may have to characterize your system to see just how much “too much” actually is, as it varies with focal length and guider chip size. Trial and error sounds tedious, but it's actually quicker than a Google search when your sitting scope side. Besides, each star field will be crowded differently, and you only need to calibrate once unless your removing and reattaching the guide camera between sessions.
Hot pixels can be dealt with easily with dark frames, either with a guider with a shutter (autodark) or by using a dark library and the “Full Calibration” option on the autoguider tab. The two most popular guide cameras without shutters (Lodestar and SSAG) have a hot pixel rejection algorithm built into the plug-ins. They are enabled by default, but you can toggle them on and off, or adjust their sensitivity in the camera settings dialog. This algorithm is being refined, and may bubble up into the camera add on at a later date so that it is available for all guide cameras.
The bottom graph shows the X and Y axes are orthogonal (90 degrees) from one another. It is okay that the X and Y seems swapped, in fact the vectors can have any rotated orientation as long as they are all orthogonal to each other.
Pick A Good Guide Star
TheSkyX will pick a guide star for you if you like, but I never let it. You are smarter than a computer. There is nothing wrong really with letting the software pick the guide star, but it will do the best it can with what it has. It will not tap you on the shoulder and say “Hey, I found something, it's the best in the current frame… but man, you should keep looking around a bit, maybe slew an arc minute to the left or right or something…” My own view is letting the computer do it can create a false sense of security that you have a really good star. Always inspect it yourself at least to make sure it's appropriate.
What I do is take a full frame guider image and look for a nice star profile. Something with at least a few thousand ADU's, and something that is strong against the background. DO NOT PICK A SATURATED STAR. This is a star with a large central area all the same value (typically in the 60,000's). You'll see a broad flat peak on the 3D star display when this happens. Don't pick a guide star right on a vertical streak on the camera chip, or at the edge in the middle of a strong gradient. Don't pick a star right next to another star. One of the issues with auto guide star picking (besides hot pixels) is that you might grab a very close double star. I've seen this happen more than once. Sometimes you can tell by looking at the 3D graph peaks even if it's not obvious in the stretched star image on screen. Guide cameras with a shutter really do have some advantage here. Do auto-darks if you can. The star centroid algorithm (what figures out to sub pixel precision where the star is) has to estimate the background brightness. This is a bit more accurate if your dark subtracted, because now it's a smaller bias-like value.
Weak guide stars are often unavoidable. In the above scenario it's hard to go wrong and not get that perfect guider graph everybody wants. Unfortunately sometimes there just is no nice bright guide star near enough to what your imaging and we have to make due with a star that is only a few thousand or hundreds of ADU's above the background. To get the best results from a weak guide star, the next section on guider settings is more important.
Tweak Your Settings
To the right are some typical settings I use on my Paramount's. Direct guide gives finer grained and smoother adjustments than the relays do. The only reason to use relays is if you do not have a Paramount and you must. The small calibration distance is for my off axis guider, it does not need to be much, just enough to move the guide star several pixels.
The minimum move should rarely if ever be more than .1 arseconds. This is the minimum amount of the detected star movement before any correction is applied. It should also almost never really be less than 0.1 arcseconds either. Unless your on the moon, your seeing is not going to allow you to guide beyond this threshold anyway. There is really an easy way to determine what this is empirically based on your current seeing conditions (seeing is stability of the atmosphere, btw, not to be confused with transparency). Turn off the corrections (checkboxes “Relay Enabled”). and watch the guider graphs. The star will bounce around a little from seeing, even with good seeing. This is the “noise floor” for guiding, and you do not want your corrections to occur until the star drifts enough to fall outside this range. With PEC enabled, guiding should in the short term appear to be doing almost nothing to move the mount. If the seeing warrants, turn down aggressiveness too. The higher your minimum move, the lower your aggressivness should be… if your noise floor is 0.5 arcseconds, when you finally move outside that range, you do not REALLY want to make a 0.5 arcsecond change. This is the most valuable use of the aggressiveness setting, to keep from over adjusting based on seeing conditions.
Maximum move should also, almost always just be 1 arc second. If you have a smooth PEC loaded (or very little or smooth periodic error to begin with), you can sometimes crank this up a little to speed up the delay between dither operations. For most MYT's you should set it to 1 for reasons the mechanical engineers know and understand about how the control system works. Other than dithering however, there is almost no good reason you'd want the mount to move more than an arc second at a time anyway.
Aggressiveness can be any value during calibration. When the corrections are made however, they are scaled by the aggressiveness. Guiding rate (1/2 sidereal, etc.) has no meaning in the Direct Guide world. An aggressiveness of 10 applies 100% of the measured correction, 20 applies 200% of the correction, 5 for 50%, and so on. So, an aggressiveness of 1 then is just a 10% nudge in the direction and magnitude that the guide star was detected.
Don't Chase the Seeing!
Remember the analogy at the beginning of this about how the moon is wobbling all over the place? If you chase this wobbling your just going to get big fat bloated stars that look like they are out of focus. There are two ways to tame the seeing (remember too, seeing is going to have some blurring effect on your final image regardless!), one is by lengthening the exposure time, and the other is by turning down the aggressiveness. With a good PEC table loaded, my preference is to take longer guider exposures. You'll get a brighter guide star with plenty of signal, and the seeing will “average” out the stellar centroid (there's that fancy word for center again). You may still need to back off on the aggressiveness if your corrections are still bouncing around a bit. Without a PEC, you may need to shorten your guider exposures, and instead turn down the aggressiveness even more. Again, this is to keep your OTA from chasing the bouncing star! If you can go unguided for even a minute, then you know your guider should be making very small tiny nudges only to your position.
A very good question at this point is what does good guiding look like? Recently I had one of those rare clear nights in the Southeast this time of year and took to my dark sky site with a MYT and MX+. The MX+ had a dedicated guide scope that does not suffer from differential flexure for up to 5 minute exposures (actually, I've gone ten). The guider was a wide field guider and I always found nice bright perfect guide stars. Guiding was text book there. This is not nearly as instructive however to what was going on over on the MYT with an off axis guider.
The image at right shows what we generally want to see. The Y Axis was dec, and you see a slight push in one direction all the time. The RA was on the X axis (I had my camera sideways) and you can see some up and down motion and a couple of bigger bumps. The PEC had not been trained on this particular MYT I should point out, but when I measured it last year, it was coming in under 2 arcseconds. The scatter graph is also nice and tight in the middle. The two bumps are only about 1/10th of a pixel and temporary; it did not harm my 15 minute narrowband subs in the slightest.
I was shooting the Helix Nebula, which is low in the south, even from south Florida. Anyone who has tried to shoot the Helix with an off axis guider can probably attest to the fact that guide stars in the area are pretty faint and hard to come by, and I had to fish around a bit.
This is not the best guider graph I've ever seen, but it is still a very very good result. If you get something like this, stop fiddling with things and get some images!
Just to show you by comparison, the MX+ had a big beautiful guide star and was doing pretty much the same thing. The X/Y orientation of the guider is reversed here however, with the dec nudges showing up on the X axis and a little periodic error wobbling around on the Y.
It is very important to point out for both of these graphs (yes, I took these with my cell phone… at the time I was sending them as texts to colleagues, I was not anticipating a blog “proving” that I could get guiding to work) the guide camera was either a straight up and down ra/dec, or turned 90 degrees to ra/dec. Our guiding algorithms do not require that this be the case, but only in such cases can you tease out the dec nudges and the RA cycles in the graph. In most cases however, regardless of the guider orientation, you may see an elongated scatter plot that is the magnitude of the uncorrected periodic error. The elongation may be at any orientation, up/down, left/right, or at a diagonal.
I tracked the Helix well past the meridian, and forgot to check the mount and lost about 1/2 hour as the mount stopped at the slew limit. When I did the meridian flip, the Helix was getting very low, and was getting into some muck to the south (the city of Okeechobee, and general atmospheric haze was enchroaching). Getting a good guide star was exceptionally difficult and so I have an example of “I hate it, but it actually did okay”. The really horrid bit to the far left was the dither operation from the last exposure.
First notice how horrible this looks at first glance, let's get that out of the way first. Then look at the Y, although it does bounce above the line a little, the dec corrections are still generally one way. Also notice that the Y corrections are mostly still well under a quarter pixel. The subs still came out okay, but in hind sight class, what should I have done here?
Correct. I should have backed off on the aggressiveness a little more to account for the fact that I'm now shooting through some thin clouds and my already weak guide star is being moved about not only by worse seeing, but some intervening thin clouds as well. A sane person might have quit here, but I was trying to get all the data I could on this before it was gone for the season.
Things That Go Bump In The Night
On of my biggest guiding mysteries is an interesting cautionary tail. I'd get good guiding in my back yard and then go into the house. For short subs, every other sub would be bad. For longer narrow band subs, most would be bad. Looking at the logs, I'd see good guiding and then the guide graph would go nuts for a few seconds, then settle back down. This is obviously a mount problem right?
Well, while sitting outside and staring at the thing one night daring it to act up, I noticed something by chance. You see, I live in Florida. It's warm in Florida most of the time, and most of the Fall/Spring months when everyone else is still cold and running their heaters, we are running our air conditioners all night. When my central AC unit would fire up, it would make an awful noise and would shake and rattle like it was coughing up a lung. When it did this… so did my guider graph! The vibrations from the AC was going through the ground to my patio, and up the legs of my tripod. Not only that, but I could tell by watching when my neighbors AC came on! Now, the neighbors AC was a much smaller “disturbance in the Force”, but I quickly learned to turn off the house AC while imaging. This caused more than a few “discussions” with my wife when I'd forget to turn it back on, but that's outside the scope of this article
Waking around, closing doors, plopping down in the observing chair. These can all cause momentary guiding problems too. Something else to watch for.
It's All Your Fault!
No, of course not! Yes, guiding can be a challenge, and it may take some practice. However we at Software Bisque are also continually striving to improve how our software guides, and yes we too are still learning a few things. As more and more customers with greater and wider varieties of equipment join our family we are looking carefully at their challenges and seeing what we can to to tighten things up, or work around common difficulties. For decades we have been a world leader in amateur and professional imaging, and you don't get here by being content with the way things are. You can be sure, we too miss a lot of good nights sleeps around here!
Good luck and clear skies!
RichardOctober 5, 2015 at 7:41 pm #220187
Nice article Richard. Just curious, is there a way you guys can do something that would allow us novices to keep the min. and max move settings you've suggested, yet speed up the dither operation. I'm only dithering 1-3 pixels and it's painfully slow. Thanks.October 9, 2015 at 9:09 pm #220458
Good blog Richard. Thanks.October 21, 2015 at 9:38 pm #221072
You mention “The small calibration distance is for my off axis guider, it does not need to be much, just enough to move the guide star several pixels.” Your diagram shows 50 seconds. Is this what you were referring to? 50 seconds does not seem small. Can you elaborate a little?
thanksDecember 8, 2015 at 4:35 pm #223506
Thanks for another very informative post! I use an ST-I plus the SBIG imaging kit, for my WideField rig on an MX. Are the 0.1 and 1 arcsec still valid for the min/max settings as the ST-I etc above has a resolution of 15arcsec/pixel. Secondly, when dithering (set to 1 pixel), it takes ~ 2 to 3 minutes for the guider to get back to where it should be (0.2 pixel) and for the next main camera image to begin. Do you have any guidance on settings that might help me improve on this? Thanks, Charles.
The topic ‘Guiding Best Practices’ is closed to new replies.