PHD choosing poor stars ASI120 mini

[david_taurus83]

I've been monitoring PHD closely since getting the new mount (CEM25P). The best guiding I have achieved so far is around 0.8" RMS and a usual average of 1.2" RMS. I've noticed when PHD auto selects a brighter star with good SNR that guiding seems to be smoother. More often than not however, it seems to auto select a star with SNR of around 20 and looking at the profile the star is sometimes barely distinguishable above the background noise with the appearance of a very faint blob. I'm using an ASI120 mini on a small 32mm guide scope with 128mm focal length. This may sound very small but there are always plenty of stars to choose from. I'm starting to wonder if these very faint stars are more affected by seeing conditions as they do appear to move around the profile a bit more. Is there a way to configure PHD so it auto selects a brighter star with better SNR? I can adjust the minimum HFD setting but it tends to pick a saturated star? Guiding improves but I'm aware this is likely to not detect mechanical errors and not issue correct guide commands.

Another issue which I'm not sure is related to the above but when looping exposures the background contrast changes with each exposure. Gets brighter and dimmer randomly. I thought this was normal as the change in contrast is usually subtle with 2/3 second exposures but last night I tried 4 and 5 second exposures and the difference also increased dramatically. At 5 seconds it also became evident that it wasnt just the sky background that was changing, it was also the star brightness as I got multiple lost star notifications. It would go almost completely black for an exposure, lost star ping, then next exposure would be back to normal again.

Using PHD 2.6.7 dev 1 and ZWO drivers.

Any ideas?

[david_taurus83]

I think I have sorted this now. Camera was set as 8 bit mode. Changed to 16 bit and now the SNR is around 80/90 on the guide star. Also slightly defocused the guide scope and tidied up the trailing cables a bit. Guiding much better this evening.

[tooth_dr]

Where do you change the bit level at?

[david_taurus83]

24 minutes ago, tooth_dr said:

Where do you change the bit level at?

Theres a little settings icon beside the connect camera button. If you select one of the Ascom ASI camera options you can control the gain and offset etc. If you select the proper ZWO native driver it only gives you the option to select either 8 bit mode or 16 bit mode. I believe I had selected 8 bit when I first set it up, hence it was giving me poor SNR stars.

[david_taurus83]

I know I shouldn't dwell on it but I've been doing a lot of graph watching lately. I've been kind of cheating recently and selecting saturated stars, not too saturated, just with slightly flat tops on the profile. SNR on these are up to 200 or so and the guiding graph always smoothed out. I've typically always used a high gain setting of around 80 or so and letting PHD select a star, it always tends to select a low SNR faint star and and watching the profile, the star wobbles and waves and naturally the centroid moves about and the mount follows suit. Typical guiding of around 1.3" RMS with some large peak to peak errors. Tonight I've experimented and dropped the gain down to 10 and let PHD select a star. Its selected a nice bright unsaturated star with SNR of 250 and the centroid stays exactly in the same spot and my guiding error dropping to 0.6" RMS. There is a lot of vertical banding now I'm guessing due to low gain but I'm not bothered. This has got me thinking. Are the problems we think we have with guiding actually to do with the mounts or a particular setting like above in PHD?

[kirkster501]

You can drive yourself mad obsessing over the guide graph.  Very often it is just the seeing that night.

[david_taurus83]

34 minutes ago, kirkster501 said:

You can drive yourself mad obsessing over the guide graph.  Very often it is just the seeing that night.

Possibly a lot of times, if you dont change settings but I dont think the seeing improved for me tonight as I changed gain. I'm used to seeing a SNR of 20 or so but I regularly see screenshots of other folks PHD screen with high SNR values. I'm starting to convince myself that low gain with this camera is forcing PHD to select better stars and therefore high SNR.

[david_taurus83]

Here is a screenshot of my initial guiding. Look how rough the SNR graph with an average value of 50. This was a star PHD selected.

 

I then lowered the gain and let PHD select another star. Much better performance all round and the star profile stays nice and steady.

[kirkster501]

11 hours ago, david_taurus83 said:

Possibly a lot of times, if you dont change settings but I dont think the seeing improved for me tonight as I changed gain. I'm used to seeing a SNR of 20 or so but I regularly see screenshots of other folks PHD screen with high SNR values. I'm starting to convince myself that low gain with this camera is forcing PHD to select better stars and therefore high SNR.

Yes true.  But don't make the mistake, on a rig that had previously been guiding well, that if you get a bad guide graph one night that you have to make wholesale changes.  I have learned this one the hard way.  After years of trial and error and me thinking that I know better,  I've found the defaults are indeed very good choices.

Edited May 16, 2020 by kirkster501

[david_taurus83]

53 minutes ago, kirkster501 said:

Yes true.  But don't make the mistake, on a rig that had previously been guiding well, that if you get a bad guide graph one night that you have to make wholesale changes.  I have learned this one the hard way.  After years of trial and error and me thinking that I know better,  I've found the defaults are indeed very good choices.

I agree, there are perhaps too many things the user can change. I wonder if in the Guiding Assistant they could employ an algorithm to determine the best gain and exposure to select a suitable star for your given sky conditions?

[Annehouw]

Here is something that you can try: If you are using the native ZWO driver in PHD, change to the ASCOM driver.

I had the same observation as you and changing the driver changed typical snr from 20 or so to well above 100.  What I noticed in the guide frame is that there were fewer stars to choose from, but the star shapes were much smoother. 

Edited May 16, 2020 by Annehouw

[david_taurus83]

47 minutes ago, Annehouw said:

Here is something that you can try: If you are using the native ZWO driver in PHD, change to the ASCOM driver.

I had the same observation as you and changing the driver changed typical snr from 20 or so to well above 100.  What I noticed in the guide frame is that there were fewer stars to choose from, but the star shapes were much smoother. 

I tried that and it made no difference initially. Using the ZWO driver you can still select 16 bit mode. Whichever you choose though, you need to input the correct ADU value in the Saturation by ADU box, so 255 for 8 bit mode or 65535 for 16 bit mode otherwise I find it was selecting staurated stars or incorrectly displaying the red SAT warning on non saturated stars. After changing to a low gain of 10, it worked perfectly last night despite seeing conditions. Even when high haze came in around 2am it carried on at around 0.6" RMS no problems. It was up there with my old AZEQ6 for performance last night.

[kirkster501]

1 hour ago, Annehouw said:

Here is something that you can try: If you are using the native ZWO driver in PHD, change to the ASCOM driver.

I had the same observation as you and changing the driver changed typical snr from 20 or so to well above 100.  What I noticed in the guide frame is that there were fewer stars to choose from, but the star shapes were much smoother. 

I use the ASCOM drivers for everything.