Fine tuning PHD2 settings for 1100 with Encoders #Guiding


Andrew J
 

Hi Roland/Bruce.

The clouds parted for awhile tonight and I was able to capture the following unguided graph from PHD2. I ran calibration and then set the MinMo to 20.0 as suggested and let it run for about 5 mins with 1sec exposures.

 

This may be a bit hard to read, so I also uploaded the orginal to a public folder on Google Drive. Click Link. I also included in the shared folder the results from running Guiding Assistant and the PHD2 Logs.

Let me know if there is anything else that would be helpful.

Andrew


Andrew J
 

Hi Roland.

I would really appreciate if you could show the annotation that was discussed to the unguided graphs provided.

Thank you.

Andrew


Peter Nagy
 

I may be wrong but I believe you should be using single star instead of multi star when running GA. GA automatically disables guide output so no need to set MinMo to 20.0.

Peter


 

>>> I may be wrong but I believe you should be using single star instead of multi star when running GA. GA automatically disables guide output so no need to set MinMo to 20.0.


FYI Guiding Assistant only uses one star regardless of that setting (there's no guiding happening during GA run)



On Wed, Oct 20, 2021 at 8:49 AM Peter Nagy <topboxman@...> wrote:
I may be wrong but I believe you should be using single star instead of multi star when running GA. GA automatically disables guide output so no need to set MinMo to 20.0.

Peter



--
Brian 



Brian Valente


Andrew J
 

Just a quick follow up. Roland was nice enough to annotate one of my unguided PHD2 graphs to illustrate how he measures Seeing. He explained that he uses this measurement to help determine the initial MinMo settings in PHD2. In my case, the maximum Peak-to-Peak values between two consecutive points on the graph were around 1.5 arc seconds, My guide camera has a resolution of around 2 arc seconds per pixel. This translates into an initial MinMo setting of around 0.8 - 0.9 pixels if you add a small buffer to the 0.75 pixel variation for Seeing. Initially I thought he was looking at the variation over the entire time period of the graph. It was a "light bulb moment" when he demonstrated he was just looking at 2 (maybe 3) consecutive points on the graph. 

Note that this was on a night with particularly bad Seeing. Normally, you would expect the variation between consecutive points to be much smaller.


Another interesting point came up during our discussion related to the MinMo for RA vs. DEC. I never really understood why PHD2 Guiding Assistant (GA) always recommended a more aggressive (smaller) MinMo for RA vs. DEC. Brian provided the following quote from Bruce (author of PHD2) to help explain why these numbers are different.

Initially, the GA used the high-frequency statistics you are talking about but we weren't happy with that. Those statistics rely on a high-pass filter implementation and it's hard to know how to tune that for a broad range of image scales. So a replacement algorithm was developed using the standard deviation of the drift-corrected Dec movement as the statistical basis.  Using Dec motion in this way makes sense because the Dec motor isn't running - the movement should arise from polar mis-alignment and seeing.  Once the standard deviation value is calculated, a scalar term is applied that should produce a guiding activity level of only 10-20% for Dec.  A different, smaller scalar value is used for RA because there are no potential direction-reversal/backlash problems present in RA and a higher level of guiding activity is often needed because of tracking errors in the RA drive system.

The last sentence is key. The reason GA recommends a more aggressive MinMo for RA is because "a higher level of guiding activity is often needed because of tracking errors in the RA drive system". Roland indicated that this is not the case for mounts with encoders as the tracking errors should be miminal and recommended using the same MinMo setting for both RA and DEC axis. 

I found both of these points enlightening and wanted to pass them along here for anyone else who might benefit from these explainations.

I want to thank both Roland and Brian for patiently taking the time to explain these points. 

Andrew

Edit: Sorry for the multiple post. Stupid groups.io deleted the image when i tried to edit for typos.