Another good night with beta 12 with one exception


#1

Ran all last night with beta 12 and the only issue was failure of focus when it moved to a target in a star-poor area when seeing had also deteriorated. I had the AF sample set to max so reduced that to around 30 as well as switching to Lum focus and offsets and that fixed the “issue”.

It did make me wonder about better notifications on failed focus as it seemed to “not know it had failed”. The only way I knew that had happened is that PhD started dinging it’s “lost guide star” tone. I also wonder why it did not just return to previous focus point when it failed. Of course, I am not sure it “knew” it had failed, maybe that is the issue.

So if there is a problem, it is not in the focus itself as that was a user adjustment issue and easily dealt with. Rather it was in the way it failed and did not let me know. If I had the sound off on the PC, it would have wasted the images - which kinda sucked anyway in this case due to the bad seeing :smile:

Log With Focus Issues

Also, and this is a academic issue, how much effect would you expect seeing to have on the focus with the SGP system? It seemed (and maybe this is ad hoc thinking) that the problem was greater when the seeing went downhill. Just for comparison, her is the seeing monitor graph from last night:


#2

So…

There were at least 2 warning notifications sent to GNS during this run, but I do see one (failure / warning) area in AF that was just blatantly missed. I will patch that up this evening.

Also, it seems like you might be selling yourself a bit short for AF (especially in less dense areas of the sky). Looks like your exposure settings are 4 sec at 1x1. Because we are trying to present and measure round things (stars) with square things (pixels), I cannot recommend any AF binning above 2x2. The increase in sensitivity and decreased download time are just bonuses. The real benefit is that it spreads the same stars out over more pixels which makes it a much better candidate for precise measurement (HFR).


#3

Hmm, something does not make sense to me about the above:

I cannot recommend any AF binning above 2x2

To me, “above” 2 x 2 has always meant 3x3, 4x4, etc. Do you mean 1x1 as binning “above” 2 x 2? Technically, 1x1 is not binning at all, I have always seen 1x1 referred to as “unbinned”. I am confused by what you mean. Unbinned (1x1) would actually spread the star of a given arcsec size over more pixels, binning would spread them over fewer, larger pixels. Given that, why would binned “superpixels” be better for focus (assuming enough signal in both)?


#4

I do not get this one ken, more binning will put less pixel per stars, isn’it ?

edit, cross posted with ccdman, same question.


#5

Yes. unbinned… whatever you prefer. I figured you guys would get the gist from the statement above. “Unbinned” pixels are relatively terrible for HFR measurement.


#6

Just does not make sense to me, seems like better sampling = more data = more precision = better focus. Why is that not right?


#7

Sorry, you are right. Let me rephrase:

Unbinned focus does indeed give you the highest number of pixels per star, BUT it comes at the cost of sensitivity (especially in less dense areas of the sky). We have found that 2x2 is “typically” superior (especially at shorter exposure lengths) because the stars cover enough pixel area to provide precise measurement, the frames download much faster, and, in terms of “whole-image” measurements, it is considerably faster. At 1x1, despite the high sampling, you would require ~4 times the exposure length to get the data found on a 2x2 image. It’s about compromise. If 1x1 works for you, please continue.


#8

OK, that is pretty much what I have always thought. To summarize what I think you are saying is that from a strictly math point of view, assuming you have enough stars that are bright enough, say in the Milky Way, unbinned (aka 1x1) will be potentially more accurate.

OTOH, (especially) in star poor areas, because of the lack of enough stars that are bright enough, 2x2 is often superior on a practical level.

It seems to me that there would somewhere be a “crossover point” where unbinned 1x1 would become less effective than binned but this would also depend on the sample size (number of stars slider) setting as well so could get complex.

I guess the real answer is whatever works! :smile:


#9

Also… looking at your logs a little deeper. Good Night System sent you 9 warnings about your auto focus behavior. I am not sure exactly what GNS does on a warning (meaning… not sure if it beeps or buzzes). Was your GNS working well throughout the night?


#10

Interesting. I did not get any. I will have to look at my settings but it did not “hoot” like it did in testing and setup. It was running all night any giving me current exposure status and such.


#11

Ah. Never mind. Looking at the code I see that “Warning” is an SGPro construct and does not map to anything specific in GNS. To GNS, warnings are just regular messages (with a “Warning” prefix) so no alarm will sound. I’m wondering if that’s how warnings should behave. They are designed to indicate that something may be wrong. Does that warrant an alarm?


#12

Indeed. I did just check all the settings and everything was checked in the GNS notifications section and both sound and vibrate were on in the Android app and it was getting exposure updates. I would have heard at least one of the errors if they had given me an audible because, as I mentioned, I noticed a couple times because PhD issued an audible “star lost” ding from my PC which is about 15 feet away (phone was right next to me).

Argues for an SGP branded monitor! :wink:


www.mainsequencesoftware.com