Jump to content


Photo

PHD guiding defies logic.

  • Please log in to reply
69 replies to this topic

#1 alpal

alpal

    Soyuz

  • -----
  • Posts: 3660
  • Joined: 15 Jun 2009
  • Loc: Melbourne Australia.

Posted 05 April 2011 - 07:53 PM

I have been trying to improve my guiding with PHD.
I attempted some ideas that I had - last night
to replace the tedious drift aligning process with
an illuminated reticule eyepiece.

I thought that if I was guiding on a star at the intersection of
the meridian & the equator - I could use PHD to help me
align the Azimuth & it did seem to yield some inconclusive results.
I was able to get it better by adjusting the azimuth screws.
However - I was unable to view a star on the Western or Eastern horizon
to check the elevation for polar alignment
but I looked at the display at the bottom of the screen & it was showing
North - 320, North 530, North 800 , North 350 etc.
All different values as it updated every second.

I thought - why do I need to look at a star on the horizon?
It is clearly showing a correction always North which is the polar axis.
I then played around with the elevation settings on my EQ6 mount.
It was set to 38 degrees - which is my latitude.
I tried setting it to 38.5 degrees, 39, 39.5, 40,
It was always still heading North on PHD.
I then tried to set it the other way:
37.5 degrees, 37, 36.5, 36, & guess what?
It still said on PHD that it was correcting for NORTH !!!
with similar values. 300 to 800.
(it was set to give a maximum excursion of 800)

Can anyone explain this as it seems to defy all logic?
What are the units of measurement for this 800 value that I get?

#2 Dan M

Dan M

    Vostok 1

  • *****
  • Posts: 152
  • Joined: 11 Oct 2008
  • Loc: Central Texas

Posted 06 April 2011 - 12:19 AM

I used PHD for helping to drift align but only to capture images of the star I want to drift align against. I use StarTarg on top of PHD to watch the drift. It really is a joy to use.

#3 alpal

alpal

    Soyuz

  • -----
  • Posts: 3660
  • Joined: 15 Jun 2009
  • Loc: Melbourne Australia.

Posted 06 April 2011 - 01:24 AM

I used PHD for helping to drift align but only to capture images of the star I want to drift align against. I use StarTarg on top of PHD to watch the drift. It really is a joy to use.


Thanks Dan - I'll check that out.
I'll also check out EQMOD which I hope allows you
to drift align & can give you more points
for pointing accuracy.

It is obvious that the numbers at the bottom of the PHD guiding screen are just garbage.
I also tried the graphs for Ra/Dec & dy/dx & they
were also misleading.
e.g - you can put the mount at different elevations &
the lines always end up at the top of the graph above the center line when they stabilise.
I am starting to think that there must be something fundamentally wrong with PHD guiding
unless someone can prove otherwise.

#4 shams42

shams42

    Apollo

  • *****
  • Posts: 1108
  • Joined: 05 Jan 2009
  • Loc: Kingsport, TN

Posted 06 April 2011 - 11:50 AM

I use PhD guiding for drift alignment all the time, and I've never seen what you describe.

The first thing you need to do is to physically rotate your camera in the focuser so that movement in RA and dec follow the horizontal and vertical axes of the display. If doesn't matter whether RA or dec end up on the horizontal sxis, just that moving in RA or dec doesn't result in diagonal star movement on the display.

Now, after doing an approximate polar alignment, begin the drift alignment procedure. Slew to a star near 0 dec and the meridian. Calibrate PhD and begin guiding. Now, open the graph, and switch the view from RA / dec to dx/dy. Also, select "off" from the dec guiding algorithm dropdown. You will see either dx or dy, whichever is the one corresponding with the dec axis, to begin drifting. Adjust the mount in azimuth until the drift goes away. You will most likely need to stop and restart guiding each time you move the mount.

Now slew to a star near 0 dec and the horizon. Calibrate PhD again (click on "brain", change dec guiding algorithm back to "resist switching" or "low pass filter", and check "force calibrate"). After PhD starts guiding, go to the graph and do the same thing you did before, this time making adjustments in alitude until drift stops.

You quickly see that, when you pass through ideal polar alignment, the reported direction of the drift will change. It sounds to me like you might have failed to recalibrate PhD when you slewed to a star near the horizon. You can't drift align in PhD with dec guiding active. Also, if you forget to switch from "RA/Dec" to "dx/dy" on the graph after turning off dec guiding, you will just see a flat line for dec corrections.

#5 alpal

alpal

    Soyuz

  • -----
  • Posts: 3660
  • Joined: 15 Jun 2009
  • Loc: Melbourne Australia.

Posted 06 April 2011 - 04:42 PM

I use PhD guiding for drift alignment all the time, and I've never seen what you describe.

The first thing you need to do is to physically rotate your camera in the focuser so that movement in RA and dec follow the horizontal and vertical axes of the display. If doesn't matter whether RA or dec end up on the horizontal sxis, just that moving in RA or dec doesn't result in diagonal star movement on the display.

Now, after doing an approximate polar alignment, begin the drift alignment procedure. Slew to a star near 0 dec and the meridian. Calibrate PhD and begin guiding. Now, open the graph, and switch the view from RA / dec to dx/dy. Also, select "off" from the dec guiding algorithm dropdown. You will see either dx or dy, whichever is the one corresponding with the dec axis, to begin drifting. Adjust the mount in azimuth until the drift goes away. You will most likely need to stop and restart guiding each time you move the mount.

Now slew to a star near 0 dec and the horizon. Calibrate PhD again (click on "brain", change dec guiding algorithm back to "resist switching" or "low pass filter", and check "force calibrate"). After PhD starts guiding, go to the graph and do the same thing you did before, this time making adjustments in alitude until drift stops.

You quickly see that, when you pass through ideal polar alignment, the reported direction of the drift will change. It sounds to me like you might have failed to recalibrate PhD when you slewed to a star near the horizon. You can't drift align in PhD with dec guiding active. Also, if you forget to switch from "RA/Dec" to "dx/dy" on the graph after turning off dec guiding, you will just see a flat line for dec corrections.


Thanks shams42.
I have sort of tried what you said above except for rotating the camera.
I was just wondering why it continues to read out a North error
& doesn't flip to South at some point when
adjusting the elevation?
I am not sure if I had resist switching or filter turned on.
I will have to revisit this.
I was trying to align the Dec without going near the horizon
as I couldn't see any stars close to the horizon from my location.
I wondered why that matters? - as the mount was still making North
adjustments which is the Dec axis.

This is my point:
PHD sees an error in elevation & adjusts for it -
no matter where your telescope is pointed therefore
why doesn't the error value FLIP at some stage
& go from North to South at the bottom of
the PHD guiding screen - when you change the elevation?
Have you ever looked at those numbers at the bottom of the screen?

#6 alpal

alpal

    Soyuz

  • -----
  • Posts: 3660
  • Joined: 15 Jun 2009
  • Loc: Melbourne Australia.

Posted 06 April 2011 - 04:48 PM

By the way - PHD guiding knows where North & South are
without rotating the camera.
It sends out a command to the mount to go North & then South
when calibrating - the result tells PHD guiding the direction for adjusting the Dec axis.

Also - I was using the info from this link here:

http://www.njstargaz...arAlignment.asp

#7 shams42

shams42

    Apollo

  • *****
  • Posts: 1108
  • Joined: 05 Jan 2009
  • Loc: Kingsport, TN

Posted 06 April 2011 - 06:54 PM

By the way - PHD guiding knows where North & South are
without rotating the camera.


Yes. For guiding, it doesn't matter how the camera is oriented. But for drift aligning according to the procedure I outlined above, the camera needs to be aligned with RA / dec. This is because you will be looking at the "dx / dy" plot, and if the camera isn't properly rotated, declination drift will appear as errors in both x and y.

#8 shams42

shams42

    Apollo

  • *****
  • Posts: 1108
  • Joined: 05 Jan 2009
  • Loc: Kingsport, TN

Posted 06 April 2011 - 07:04 PM

Thanks shams42.
I have sort of tried what you said above except for rotating the camera.
I was just wondering why it continues to read out a North error
& doesn't flip to South at some point when
adjusting the elevation?


I'm a bit puzzled by this as well; I've never seen that happen. At any rate, you should turn dec guiding off when you drift align, so you won't see ANY dec corrections.

Did you force a recalibration near the star that you are using near the horizon?

I am not sure if I had resist switching or filter turned on.
I will have to revisit this.


It probably doesn't matter.

I was trying to align the Dec without going near the horizon
as I couldn't see any stars close to the horizon from my location.
I wondered why that matters? - as the mount was still making North
adjustments which is the Dec axis.


It doesn't matter if you can see stars there or not, your guidescope can see stars where you cannot. At any rate, that's where the mount must be pointed for drift alignment to work properly. The logic of pointing near the horizon at zero declination is that azimuth errors will not induce declination drift there.

If you weren't pointed in the right spot and had an azimuth pointing error, that could cause the star to drift in declination -- perhaps consistently to the north.


This is my point:
PHD sees an error in elevation & adjusts for it -
no matter where your telescope is pointed therefore
why doesn't the error value FLIP at some stage
& go from North to South at the bottom of
the PHD guiding screen - when you change the elevation?
Have you ever looked at those numbers at the bottom of the screen?


Yes, my numbers behave as they should. If I have dec guiding active, I will see PhD begin issuing commands in the other direction as I pass through the point of perfect polar alignment. That's why I'm wondering if you always calibrate near the star you are using for drift alignment (otherwise PhD may confuse its directions, I think) and if you are pointed at a star that is in the right place in the sky.

However, this is a secondary concern; you should have dec guiding turned off for drift alignment.

#9 alpal

alpal

    Soyuz

  • -----
  • Posts: 3660
  • Joined: 15 Jun 2009
  • Loc: Melbourne Australia.

Posted 06 April 2011 - 08:43 PM

Thanks Shams - I'll have another go at this & report back.

#10 shams42

shams42

    Apollo

  • *****
  • Posts: 1108
  • Joined: 05 Jan 2009
  • Loc: Kingsport, TN

Posted 06 April 2011 - 09:46 PM

Good luck!

#11 bluedandelion

bluedandelion

    Surveyor 1

  • *****
  • Posts: 1857
  • Joined: 17 Aug 2007
  • Loc: Hazy Hollow, Western WA

Posted 07 April 2011 - 12:03 AM

Matthew,

Thank you for that quick tutorial for drift aligning using PHD.

This does assume that the guide scope axis is parallel to the main OTA doesn't it? Or in practice is that not a big factor since you are also autoguiding?

Ajay

#12 bardo

bardo

    Viking 1

  • *****
  • Posts: 795
  • Joined: 13 Sep 2009
  • Loc: US

Posted 07 April 2011 - 12:59 AM

Ive had it happen several times. when its doing that I make way to obviously large corrections to make sure Im going the right way with adjustments or something isnt astray.
sometimes I need to do like 3/4-1 turn. the lat gauge is known to not be terribly accurate anyway, just a jist to go by. Im visually oriented so I like watching the graph.

also, make sure you give it some time between intervals. often when i make adjustments it will go one way for the first few minutes then start a slow trend in the opposite direction. I cant explain it but it does it. so I give it 3-5 minutes or so each time I make an adjustment just so Im clear. plus it gives a longer plot on the graph to go by making the guess at adjustments more accurate.

#13 alpal

alpal

    Soyuz

  • -----
  • Posts: 3660
  • Joined: 15 Jun 2009
  • Loc: Melbourne Australia.

Posted 07 April 2011 - 03:31 AM

Shams

If you weren't pointed in the right spot and had an azimuth pointing error, that could cause the star to drift in declination -- perhaps consistently to the north.


I think you've answered the question here.
It seems that if you're pointed at any star,
other than one close to the East or West horizon
PHD is obtaining Azimuth movements
by moving North on the Declination!

That would explain what I'm seeing.
Therefore the fact that it reads Dec 500 or Dec 600
means nothing in regards to Dec.
It's actually compensating for Azimuth!

#14 bardo

bardo

    Viking 1

  • *****
  • Posts: 795
  • Joined: 13 Sep 2009
  • Loc: US

Posted 07 April 2011 - 04:00 AM

thats the same principle if you're not level, but it shouldnt be so drastic. you should still be mostly correcting the right direction. to combat this go back and forth between az/alt adjustments incrementally.

#15 shams42

shams42

    Apollo

  • *****
  • Posts: 1108
  • Joined: 05 Jan 2009
  • Loc: Kingsport, TN

Posted 07 April 2011 - 10:17 AM

This does assume that the guide scope axis is parallel to the main OTA doesn't it? Or in practice is that not a big factor since you are also autoguiding?
Ajay


I think it only assumes that that guide scope axis is roughly parallel to the RA axis, but I'm not sure about that. I don't think that the guide scope to main OTA axis is important at all at this stage.

#16 bluedandelion

bluedandelion

    Surveyor 1

  • *****
  • Posts: 1857
  • Joined: 17 Aug 2007
  • Loc: Hazy Hollow, Western WA

Posted 07 April 2011 - 10:56 AM

That makes sense. Drift alignment is about getting the RA axis pointing at the celestial pole. So having the guidescope parallel to the RA axis and then doing the PHD align procedure should do it.

Thanks again.

Ajay

#17 shams42

shams42

    Apollo

  • *****
  • Posts: 1108
  • Joined: 05 Jan 2009
  • Loc: Kingsport, TN

Posted 07 April 2011 - 02:01 PM

Actually, what I said makes no sense. I think as long as the guidescope is pointed at the appropriate region of the sky, that's all that matters. After all, as soon as you move the mount in declination, the guidescope is no longer aligned to the RA axis of the mount!

#18 alpal

alpal

    Soyuz

  • -----
  • Posts: 3660
  • Joined: 15 Jun 2009
  • Loc: Melbourne Australia.

Posted 07 April 2011 - 05:30 PM

Good news.
I finally got the mount to align with PHD guiding
by setting the Dec up on a star near the horizon.
The Azimuth was another story.
It ended up close but not perfect.
My mount Dec angle display was 0.5 degrees out.
Correct was 37.5 when it should have been 38 degrees.
I managed to get some subframes of Centaurus A _ NGC5128.
See pic - very quickly processed.

Attached Files



#19 shams42

shams42

    Apollo

  • *****
  • Posts: 1108
  • Joined: 05 Jan 2009
  • Loc: Kingsport, TN

Posted 07 April 2011 - 10:39 PM

Good news.
I finally got the mount to align with PHD guiding
by setting the Dec up on a star near the horizon.
The Azimuth was another story.
It ended up close but not perfect.
My mount Dec angle display was 0.5 degrees out.
Correct was 37.5 when it should have been 38 degrees.
I managed to get some subframes of Centaurus A _ NGC5128.
See pic - very quickly processed.


Hey, that's great! Looks like the stars are nice and round, which is all you can really ask for.

I wouldn't worry about the latitude scale being a bit off; it's not applied very precisely at the factory.

Did you do a complete drift alignment (for altitude and azimuth) or just for altitude?

#20 alpal

alpal

    Soyuz

  • -----
  • Posts: 3660
  • Joined: 15 Jun 2009
  • Loc: Melbourne Australia.

Posted 08 April 2011 - 01:50 AM

Hi Shams,
The pic is not too bad considering that it's a tiny crop
of the total frame.
I did a complete drift algnment for altitude & azimuth.
I got the Dec exact.
I had no frame rotation after 150 minutes of subframes.
PHD was just hovering around +/- half a divison on the dy graph
when I guided.
The azimuth was mostly 1 division positive on dx & it actually
guided worse than before (when it wasn't properly aligned.)
I had to chuck out 19 of 30 x 5 minute subframes
leaving me only 11 resonable frames to stack.
That is another issue - I think.
I only have homemade tube rings which have too much
play in them - in my opinion.
I also need to tune the closed loop system by playing around
with the PHD settings.
I will try & get it right.

One thing I need to do is install a Canon program
to view more closely the frames as I take them.
(instead of just the preview on Canon EOS remote)
I didn't realise that the stars weren't round on all frames
until later when I magnified them up.
I will get this sorted one way or another.

#21 shams42

shams42

    Apollo

  • *****
  • Posts: 1108
  • Joined: 05 Jan 2009
  • Loc: Kingsport, TN

Posted 08 April 2011 - 10:15 AM

Wow, that's too bad. Can you post the details of your setup and your settings in PhD?

#22 alpal

alpal

    Soyuz

  • -----
  • Posts: 3660
  • Joined: 15 Jun 2009
  • Loc: Melbourne Australia.

Posted 08 April 2011 - 04:59 PM

Wow, that's too bad. Can you post the details of your setup and your settings in PhD?

Sure - I just clicked on the "Brain" function.
These are the settings & although I tried 2 second
capture frames half way through, PHD lost track
altogether so I changed to 1 seconds frames.
The settings below were not altered:

RA Aggressiveness 36
RA Hysteresis 10
Dec guide mode Auto
Dec Algorithm Resist swit
Max Dec duration 520
Dec slope weight 5.00
Calibration (ms) 1540
Min. motion (pixels) 0.55
Search region (pixels) 15
Noise Reduction None
Time Lapse (ms) 0
Camera gain (%) 100
LE port Port 358
LE read delay 0
Force calibration tick
Use subframes no tick
Log info no tick
Disable guide output no tick


I'd be interested to know a good way of tuning my setup.
PHD has never given me 100% good frames.
Re-checking I actually got only 9 reasonable frames out of 30
when taking shots of that Centaurus A galaxy.
My previous time on the Southern Pinwheel galaxy
I got about 20 good frames out of 35.
None of the frames are absolutely perfect.

Is there a way of tuning this double axis closed loop
by perhaps putting in a command from the hand controller on say a x1 forced movement &
looking at how PHD handles the disturbance on the graphs?
(and adjusting accordingly)

#23 shams42

shams42

    Apollo

  • *****
  • Posts: 1108
  • Joined: 05 Jan 2009
  • Loc: Kingsport, TN

Posted 08 April 2011 - 08:07 PM

Your settings look kind of off. With RA aggressivenes that low, you are hardly making any corrections in RA. Try these settings and see what happens:


RA Aggressiveness 85
RA Hysteresis 15
Dec guide mode Auto
Dec Algorithm Resist switching
Max Dec duration 1500
Dec slope weight 5.00
Calibration (ms) 5000
Min. motion (pixels) .2
Search region (pixels) 15

#24 alpal

alpal

    Soyuz

  • -----
  • Posts: 3660
  • Joined: 15 Jun 2009
  • Loc: Melbourne Australia.

Posted 09 April 2011 - 01:03 AM

Thanks Shams,
I'll try those figures you've given me.
I have played aroud with RA Aggressiveness.
100% was too much & caused worse guiding.
I know I'm close & I hope it just needs a bit of tweaking.
cheers
Alpal

#25 alpal

alpal

    Soyuz

  • -----
  • Posts: 3660
  • Joined: 15 Jun 2009
  • Loc: Melbourne Australia.

Posted 09 April 2011 - 06:19 AM

There might be a clue to the problem.
Look at this tiny crop of one of the worst frames.
You can actually see 2 stars instead of one.
Surely that must be caused by the external guiding scope
shifting it's position?

Attached Files








Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics