PHD2 settings

[Dealy]

I've been playing with a few of the settings in PHD2 to see if I can smooth out the trend line a bit but have only had marginal success.

Does anyone change the settings, or do you go with defaults?

Does anyone know any good tweaks that may work for me?

Mount is AZEQ6 with minimal backlash.

Screenshot is below.

Thanks

[troypiggo]

I'm no expert, but I think to assist analysis of what's going on it can be more helpful to have "corrections" ticked so you can see when a pulse is being sent to the mount. It will give an indication of whether you're over or under correcting. Also, bit more information on your imaging scales - guidescope focal length, guidecam pixel size, imaging scope, imaging cam FL. This will give a feel for what impact guiding errors will have on actual imaging rig.
If you enable the star profile graph, you can see if the star you're guiding on has a nice, pointy peaked profile, or a flat top which means it's saturated (you don't want that). The SNR of your guide star looks good at 16.

[codemonkey]

Troy's nailed it I think, but this post sparks a question for me. I've never dithered; would dithering impact your RMS?

Might help to change the graph to display in arcseconds rather than pixels as well.

Oh, and tweaking the settings is mandatory. Min move for instance should be adjusted based on current seeing conditions; if it's too small, you'll be bouncing around chasing the seeing, but if it's too big you'll be getting less precise guiding than you're capable of.

[troypiggo]

Dithering only temporarily puts a bump in the plot, because it shifts the star by some small, random amount, then the guiding centres on the shifted star within a short time usually (seconds, 10s maybe?). PHD2 labels dithers in the plot so you know a dither from other jumps in the plot (eg from backlash, PE, wind etc).

[Dealy]

Thanks Troy and Lee. The weather has turned cloudy here so next time I'm out I'll redo the graph and give some more info.

Since my first post I've had one imaging session that went reasonably well. The graph was fairly flat for most of the time, then it will have a spasm and go wild for a minute, then behave itself again.

There was no wind or any outside influence that would have caused it. I have heard that people have found grit in their SW mounts when they've re-greased them. I can only guess that mine may have a similar problem. It hasn't done that much work so I don't think it's wear and tear.

So I'm currently losing about 20% of my subs due to random guiding errors.

I read another thread here somewhere about guiding where it was suggested to change the graph to pixels. I will grab a screenshot of both.

Cheers

[gregbradley]

I am surprised you are losing subs with a graph that looks like that. That looks better than my graph with the AP1600! So there must be something else going on like flexure. No evidence in that graph other than stellar guiding.


I always use the guide assistant and accept the recommendations for min max move it comes up with after 120 seconds of running it.


I set star mass to 100 so that it does not stop guiding if a cloud goes overhead and now it just keeps going just like CCDsoft autoguiding does.


This recovers from a cloud normally especially if you have a bright guide star.

I put a GSO .5X reducer on the end of my SBIG STi so I get a high SNR of my guide stars and haven't had a problem getting a decent guide star in the last month since I fitted that. I was thinking of adding the infrared filter to help even more but that would require a quite bright guide star and I am not sure it will allow my STi to come to focus as it blocks it from going into the MMOAG deep enough to achieve focus.

I would watch the video of one of the developers who goes over the features of PHD2 to get a feel for it. I was amazed at how much tech has gone into this. Its way more advanced than any other guiding software I have used and superior to Sky X autoguiding which is essentially the same as CCDsoft.

There are different algorithims available for both RA and Dec and the video explains them and when to use them. The video is linked from the PHD2 website.

You can click on some stats that are helpful. One is the arc secs of error of guiding box. Also the FWHM of the guide star that gets updated.

[lazjen]

Well, that's an interesting idea and I happen to have both those items. How much spacing is required for the 0.5x reducer to focus?

[codemonkey]

Not a bad idea there. Something else to add to the shopping list ;-) Since a recent thread I questioned why I've not been binning my QHY5L, which I've started to do... helps a fair bit.

I've actually got an IR filter I've been meaning to try as well, but I've never gotten around to it. I found another bandpass graph for the one I have and found that it actually passed a fair bit of UV light, which concerned me... ordered another one to use instead but I've been waiting over two months now and it's still not been shipped.

Back on topic: the reason I suggested changing the graph from pixels to arcseconds is twofold: if you take a casual look at this graph it looks amazing... but then if you look at the RMS you realise it's quite poor, and it's just the sampling rate that makes the graph look nice.

The second reason I suggested it is because it gives you a visual indication that you can more easily marry to your imaging camera if it's in arcseconds, otherwise you have to work out what 1 pixel in your guide cam is vs 1 pixel in your imaging cam... just a bit easier to see how bad it is if you know your imaging cam's sampling rate and can see the graph in terms of arcseconds.

[peter_4059]

I'm interested to understand why binning the guide camera helps - I assume you get a brighter guide star so the centroid detection probably works better but don't you also reduce the guide system resolution by half? Sure - this will give you a nice smooth guide graph but does it make your actual guiding better?

[codemonkey]

Spot on Peter. I'm using an OAG, guide cam has pixels 3.75um, imaging 4.54, so I don't think I need the resolution, but the binning can help finding good guide stars. I won't always use it though, mostly just a convenience factor, and sometimes the available stars will be too bright to warrant it.

[Rod771]

Wow! Their website has improved out of sight! I was thinking, "video? What video!"

[gregbradley]

As far as binning goes with PHD2 I was under the impression it used RAW only which I take it means 1x1.

You can lower noise by using their noise reduction option (there are 2 options). This can suppress hot pixels as well.

With regards to the graph in arc secs, the RMS errors still look very low. That is 1.11 arc secs of error. That's tiny.

I think you are looking in the wrong area for your star problem. I would check for flexure and for non squareness of camera. Non squareness of camera can give elongated or odd shaped stars but not all over the image, usually one side or one corner.

Flexure will make all stars elongate like tracking. Its easy to confuse these 2 factors with autoguiding errors as they can look similar.

Flexure is handled by making sure adapters are strong and tight. Not using a guide scope or use the guide scope attached to the OTA tube not the mounting rings dovetail plates which is usual, making sure there is no mirror shift. A tpoint model will map all these flexure points and activating Tpoint if you are using the Sky X can reduce the effect of these slow flexures. Non squareness requires working out the orientation of the camera with respect to the displayed image (often reversed top and sides) and then packing the offending side out with thin metal until stars in all 4 corners are the same.

Greg.

[codemonkey]

I use the ASCOM driver for my QHY, which enables me to use 2x2 binning. Minor point is that it resets the pixel size every time I connect the camera, so I have to manually override it in PHD after connecting, otherwise it gets confused and shows me really awesome guiding (sometimes I forget and get excited briefly).

[gregbradley]

The advantage of 2x2 binning for autoguiding (which is usual practice) is higher SNR for the guide stars. As the centroid calculation is supposed to be very accurate I read there is no benefit to using 1x1.

Although I distinctly recall getting superior guiding on my Tak NJP mount using 1x1 guiding over 2x2 guiding.

Perhaps 1x1 with a reducer may be better than 2x2?

Greg.

[Dealy]

Thanks for the help so far.

Here are 3 single 240sec subs from the other night.

Note the star elongation in #1 & #2. The orientation is in different directions.

#3 is a good one, with reasonably round stars.

I was thinking it's a guiding error .... maybe it is flexure. But would I get flexure in different directions from 1 sub to the next?

[Dealy]

Derrrrrrr .... I think I found the problem!

The guide camera was loose in the guide scope :eek:

I don't know how it came loose :shrug:

Just goes to show ...... check the simple things first and don't assume anything.

Looks like it's going to be a nice night under the stars, so I'll give it a go tonight.

[Dealy]

I still found at times after some relatively smooth guiding something would suddenly send the graph of the scale, then several seconds later it would correct itself and start guiding smoothly again.

After I found the guide camera loose I checked the whole set-up. The only thing I can think of, except flexure, which is still possibly the main culprit, is that maybe I took too much backlash out of the drive system. I'm thinking there may be a few tight spots in the worm and wheel.

So I've re-adjusted and put a tiny bit of backlash back in. I'll see how that goes.

An OAG may still be on the wish list though.

[Dealy]

Yep, that fixed it.

There must be a couple of flaws in the worm or wheel. Adding tiny bit more backlash made all the difference. I also went back to PHD2's default settings.

I had a quick look at last night's subs and I don't think there's a bad one in there.

I love it when things go according to plan. :)

I've never used periodic error training - I'll have to try it. It may help even further.