bizpat72 if the guide star value is at 245 and stays there, we agree that corrections are negligeable. Then, the greater the change in the Peak Value, the bigger the corrections.

I, for one, do not agree.

You appear to be mixing up the difference between three things, (1) scintillation (changes in star brightness, as in "twinkle, twinkle little star," with apologies to Mozart) which is shown as Peak ADU (8-bit) by ASIAIR, (2) positional change of a star (shown as "Star(x,y)" in the bottom right of the ASIAIR Guide window), that is caused by atmospheric turbulence (colloquially called "seeing"), and (3) positional change of a star that is caused by the mount's imprecise tracking.

Scintillation (item 1) does not change the position of a star except for perhaps reducing the accuracy of the measurment of item 2, when the star's ADU dims by say 100 times of it maximum.

If you are critically sampled, a peak ADU of 60 is more than sufficient to get a decent estimate of the star's centroid.

Item 2 is what we need to try to keep low so that the centroid of the positional changes is stable, so that we can use it (or them, for multiple stars case) to accurately detect the error from item 3, and correct for mount errors (by issuing "correction pulses" to the mount, which tries to move the mount in the opposite direction of its own error).

The error from Item 2 can be reduced by using a longer exposure time. For example, doubling the guide exposure time (say from 1 second to 2 seconds) will reduce (statictically) the variance of the measured centroid by a factor of 2. I.e., the RMS centroid measurement error is reduced by a factor of 1.414. Using 4 second exposure instead of 1 seconds will half the RMS measurement error.

Similarly, assuming statistical ergodicity (and stationarity), one can also reduce this "seeing" error by using more than one star. Using two stars (both measured using 1 second exposure) is equivalent to using a 2-second exposure for a single star. When you have a mount that requires frequent adjustments (for example, every 1/2 second), this is your only recourse. However, there is a flaw in ASIAIR multi-star guiding where if you have a much brighter than average star, it will treat even a 12-star case as a 1-star case because ASIAIR is putting all the weight of the centroid averaging on the brightest star.

If you have a mount that requires 1/2 second guide exposures, you will not be able to use a single star case, unless you live very high up a mountain top when most of the pesky atmosphere is below you. Since you are unfortunate enough to choose ASIAIR for autoguiding, you will need to increase the guide camera gain so much that the brightest two or three star become saturated, and thus become excluded from being used as guide stars.

You definitely suffer from the 1 star "seeing" problem (even when using a 3 second exposure). Notice at the end of your guide graph how much the red (declination) graph has moved, without being directed to move (the red correction bars). Unless you have a particularly poorly machined mount, it is unlikely that the 2" error is caused by the gears themselves, so what is most likely is that it is cause by centroid measuring error.

Also, stop trying to polar align to better than 15". Unless you are trying to set up your mount to measure something critical, don't spend more than 4 minutes with polar alignment. You are throwing precious time away. You will see zero difference between polar aligment error of 0 and polar aignment of 1 arc minute unless you are taking subexposures of 1 hour each.

Have you tried using PHD2 on a laptop to grok autoguiding? If not, I would highly recommend that. You can't learn anything technical using ASIAIR (which itself has numerous technical flaws).

Chen

bizpat72 I once read that under good seeing conditions, guide exposure should be lowered to 1-2 sec while also increasing the RA and DEC agression, and when the seeing is "not so good" then increase exposure time and lower RA and DEC agression.

Aggressiveness is related to the traditional "loop gain" of a feedback loop, although PHD had used ad-hoc loops instead of text book loops.

I would recommend using as little loop gain (thus as little aggressiveness) as possible. Start near zero, and only increase the aggressiveness by a little at a time, and only if you find that the correction pulses are struggling to correct the errors. Each time you adjust aggressiveness, either restart autoguiding, or leave it alone for five or six times the exposure duration before deciding if it needs further adjustments.

But my GEM45 seems to prefer longer guide exposures and lower agression values no matter what.

Use multi-star guiding, and you can use shorter guide exposures.

So what you are seeing on my screenshot, is that the big star under the X50 button is probably messing up multistar?

No! No even probably.

That star is saturated and is ignored for guiding purposes when you choose multi-star guiding. It definitely is not a problem since you are guiding with just 1 star that you had selected -- that star is guaranteed to be ignored. Had you pick that bright star, you will find its max ADU clipping at 255 all the time.

Your biggest problem is not that star, but the star that you have picked shows an ADU of 245. This star is bound to occasionally saturate and cause the centroid to be poorly estimated and cause the guiding to go wacky. If you want to use 1-star guiding, pick a star nearer to ADU of 100, so that it has less chance of saturating when it scintillates.

Chen