Skylab1 I just got notify 1.6 (7.75)beta is ready to install
Told you to be patient, eh? :-) :-)
What is the build number that you are seeing?
I don't think I got firmware 7.75 pushed to me. For build 3, I had noted down in my log as firmware 7.73 (perhaps by mistake?). Build 4 came out yesterday morning with firmware 7.77, and Build 5 appeared this morning with firmware 7.79.
The build number is usually denoted in parenthesis, after the "1.6". E.g., "1.6(5) 7.79"
The 1.6 is the app version, and the 7.7x is the firmware version. Back in the v1.3 days, the firmware version was an integer, e.g., 775 instead of 7.75.
All the actual multi-star mechanism is in the firmware. The app is just the GUI for it.
With Build 4 (yesterday), the Multistar switch had been removed from the Advanced Feature window (a sub window of the window with the ellipsis icon). I think they must consider the feature to be good enough (when all the remaining multi-star bugs are finally squashed) to be used by the regular folks.
Muti-star selection is now simplified, so that if you tap on a star (either to calibrate or to guide) it will use single star. If you ask it to calibrate or guide without tapping on a star, it will use multiple stars.
When you are testing multi-star, you might get the green circles to vanish one by one, never to return, until over time, you are left with just a single star for guiding.
Don't worry (this is beta!).
They are actively working on this problem, and I wouldn't be surprised if the Build 5 that I got this morning has already fixed it. They know why it is happening, so it is just a matter of finding a best approach.
Build 3 was dropping stars when the star size dropped below 2. With build 4 last night, it is stable down to at least a star size of 1.8.
You may not even have a star size ever go below 2 with your guide system. I use a Borg 55FL (a doublet which Borg advertises as apochromatic, but really between achro and apo) objective and ASI290MM. On top of that, I have a strong UV/IR cut filter to get rid of optical aberrations outside of the visible spectrum. If you have a smaller pixel size or if the optics don't have good MTF, you will not see a star size that small even when focus except when the intensity really drops. If you guide with a ASI174-mini, though, you might see small star sizes too.
Even though recommended by the unwashed crowd who pass information through hearsay, defocusing to get a larger star size is never a good idea because you lose signal to noise ratio. I always keep my guide scope as focused as possible and depend instead on good algorithms to figure out the centroid. An HFD of 2 is plenty good to find centroids since half of the photons are outside a diameter of 2; the centroid computation des not depend on just a 2x2 area. Viewed another way, it satisfies the sampling theorem.
With build 3, I was seeing the multi-star green circles disappear within a minute of staring to guide. With build 4 last night, at one point it took an hour for 12 stars to dwindle down to 8 stars; so much improved already.
The stars that temporarily are unguidable should return, but there is a bug that causes them to not come back. Again, don't worry, they are really actively working on it.
The key is if you can get improved guiding with multi star when there are many stars available. I get between 1.5x to 2.5x smaller total RMS error when I use multi-star. That is a whopping big improvement in the guiding world. But my mount has virtually no backlash. If the total RMS error is mostly caused by the mount, you might not see as big an improvement.
By the way, a 2.5x improvement is the equivalent of using 5.66 stars (2 to the power of 2.5), and if all the errors are caused by centroid estimation.
There are other errors obviously, like pulse latency, pixel quantization, and all the mechanical errors. So, getting a 2.5x is a very big improvement, since as I had described in past postings, multi-star only improves centroid estimation. It does not cure other errors, particularly mechanical problems. It also means that for my particular mount, their choice of 12 stars is actually quite good; I am at the point where something else is a limiting factor after the 5.66 stars. But the extra stars will help when seeing becomes worse.
Being able to get 2.5x improvement just from improving centroid estimation also says that "seeing" is one of the biggest culprit to good guiding, since all that multi-star does is to average out seeing errors.
Right now, you can actually see the RMS error climb as the guide stars disappear. Quite fascinating to watch.
Also, since I think you are one that complained about MinMo, it is now defaulted to 0.1 pixel (I don't know if it is mount specific), so that should be an improvement for you too, if MinMo used to be limiting your guiding before. With the improved MinMo, single star guiding on average for me has improved from perhaps 1.3" RMS to perhaps around 1.0" RMS (need longer term testing to be conclusive). But the RainbowAstro and Avalon mounts (large periodic error slopes, but very little backlash) are known to be finicky with the MinMo parameter. It may not matter as much to mounts with larger backlash.
Remember to feed back any anomaly (the TestFlight app has the link to feedback). Remember that Beta is a two way street. It is not a final product, where you are just the consumer. There is an implicit contract that you use it for testing and feed back test data.
Have fun. IMHO, Multi-star is a pretty big deal. It was for DONUTS and its predecessors, and it is for the current PHD2 beta testers, and I suspect it will also be a big "whoa" moment for ASIAIR users. I am pretty sure the ASIAIR multi-star centroiding part is different from the PHD2 multi-star centroiding part (thus the bugs with disappearing stars, star sizes and so on that don't exist in the PHD2 version of multi-star), but it still depends on PHD2 for the inner workings of autoguiding (thus, the PHD2 log part has not changed).
Chen
