Appreciate help with PHD2

[ejp1684]

I have an HEQ5 Pro mount with Rowan belt drives added. I've adjusted the backlash (I can't feel any obvious backlash and the gears don't grind when the motors work) and went through the PHD2 steps of calibration and Drift align. Here's the result from the calibration:

Does this look ok?

The Azimuth from the Drift Align is below, and the Altitude was similar.

When I started guiding, these are two results:

I am surrounded by buildings including a block of flats so there must be a whole load of hot air rising up at night.

I've read through the article on the PHD website and felt overwhelmed by it all. I'd appreciate any comments folks can give, basically am I on the right track here, or is something seriously wrong.

Thanks in advance.

Eric.

[symmetal]

Your calibration and drift graphs look quite reasonable. I would say the aggression is set too high in RA as you get the RA corrections often overshooting needing a reverse correction which again overshoots. Expand the graph horizontally to see the corrections easier and aim for the correction to just bring the graph back to the centre line or overshoot slightly to help reduce the amplitude of the periodic error excursions. Try an aggression of 60 or 50 as a start and see what happens.

The Dec corrections are showing the effects of backlash where a series of corrections have no effect followed by one that overshoots causing the cycle to repeat. Reducing the Dec aggression may help here too and try running the phd2 backlash calibration routine which allows larger corrections when backlash occurs. I have my Dec aggression set to 35 to stop overshoots. Switching to Dec guiding in one direction only, by having a slight PA error leading to Dec drift in just one direction can help. This method fails if a Dec dither sends it over the backlash though I believe later release of phd2 recognise this and allow reverse Dec corrections after a dither even if single direction Dec guiding is enabled.

Finally getting a good guide star is the best way of successful guiding. You don't want a saturated star (clipped) where the star profile has a flat top. The profile should be a narrow spike without extra bumps in its shape. Using a high SNR figure also helps (over 100) but avoid it clipping. The seeing also affects it too. Use an exposure of 2 seconds to start with. In good seeing you can reduce it to 1 second. In bad seeing (also at low altitude) a longer exposure of 4 seconds may give a smoother graph.

You can try the new Z-filter guiding method using short exposures though I've had mixed success with that. Sometimes it's very good but other times the RA gives bursts of corrections in one direction followed by bursts in the other of 10 secs or so leading to stars looking like little Saturns. I then have to stop and restart guiding or switch to the standard guiding method.

Hope this helps a bit Eric. 

[ejp1684]

Very grateful, symmetal, many helpful suggestions. I'll try these the next chance I get.

Eric.

[symmetal]

No problem Eric.  You can also run the phd2 guiding assistant tool which will give recommendations for parameter settings. It will probably suggest to reduce the MnMo value (Miminum Movement) to something like 0.1, which sets how far the graph has to deviate from the centre line before corrections are applied. But I would reduce the aggression values first before running the guiding assistant.

Reducing the Dec aggression would make you think it would take even longer to take up the backlash but in the brain guiding setup there is an option which I believe is enabled by default to allow larger Dec corrections if it's off by a significant amount so the overall time to correct backlash won't be affected that much. You just want to avoid the final over correction which sets the whole process off again.

Setting the scope balance to be a bit camera heavy so that the Dec gears don't tend to sit in the backlash area but stay meshed at one side due to gravity can help with Dec backlash. This only works if the scope is not pointing towards the vertical. In this case I hang a little weight off one side of the scope to offset the balance point to keep the gears meshed.  

Alan

[kens]

Hi Eric, can you post the guide log from that session. Under the Help menu is an option to Open Log Folder. You can find the Guide Logs there with the Date/time of the session in the file name. The Guide Logs show a lot more about what was happening than you can get from a screen shot.

[ejp1684]

Thanks Ken. Here's the guide log.

PHD2_GuideLog_2019-03-29_201830.txt

Eric.

[kens]

Hi Eric - I see you are using the Z-filter algorithm this time.

Overall the results are as good as you could expect with RMS of 0.11 pixels on both axes. That is at the limit of what PHD2 can do so too improve you would need a smaller pixel scale for guiding.

The RA result could be optimistic because you were guiding at around 70 deg declination where any RA errors (e.g. from PE) would be greatly reduced.

Your exposure times are 3s and on the Z-filter you have an X-fac of 2. This gives a "virtual" exposure time of 6s which is quite sensible. The Z-filter is primarily designed to allow shorter exposure times (e.g 0.5s to 1.0s) but still works at longer exposures - its just the benefits are not as evident compared to the other algorithms. When using the Z-filter the MinMo values operate differently from the other algos in that it applies to the correction rather than the measured deviation of the guide star. This is because it can give small corrections that are less than the smallest movement possible in some mounts.

Your MinMo values are quite high given your pixel scale - equivalent to around 0.7". This sets a floor for your RMS values. For RA I would recommend starting at MinMo of 0 with the Z-filter. But you'd also need a session closer to Dec 0 to see what else may be needed.

On the Dec axis the MinMo value depends on your level of backlash. Your backlash seems to be under control with the settings you have so I would leave them as-is.

 

[ejp1684]

Ken, many thanks for your analysis and advice. Lots to do when UK skies are kind once again which doesn't look as if it's going to happen for at least another week!

Eric.