PHD Guiding

[widow18]

Could someone please tell me which variable affects OSC, and what it means when the OSC value is displayed in red. The value was less than .20 when the numbers turned red, and I am led to believe the value should be between.40-.50.
Using HEQ5 and Orion ST with Starshoot autoguider.
Thanks
Peter

[Garbz]

Try cranking up RA aggressiveness and hysteresis. The only time I've ever seen a really low OSC is when my graph almost flatlined and my guiding was quite crap as a result.

The other thing which has caused it for me before was having a really well balanced telescope mount. I used to think the closer to perfect the better but apparently you want the mount to be heavy on one side to help with backlash. At the time my guiding was just fine until my scope crossed the meridian and then it turned really poor.

[widow18]

Thanks Chris, that sort of sounds like what may have happened, shooting almost straight up as well so the balance would have been delicate.
Never seen the OSC numbers red?
Peter

[alistairsam]

Chris,

When you say flatlined, do you mean they're flat along the centre line or beyond the upper and lower limits?
I recently had one of my best guiding sessions where ra and dec were very close along the centre line throughout, but the images showed up elongated stars in the same direction, sort of star trail. I'm attributing it to a focuser tilt so will test for that by rotating my camera, but now am not too sure.
just wanted to check if ra and dec lines along the centre line was bad. I didn't think so.

[widow18]

Alistairsam
Same with me, best RA Dec lines ever, the only anomoly was the low OSC and Number displayed in red, and oval stars.
Peter

[jenchris]

My Guide graph looks like an earthquake - my stars are round - I'm on an OAG at F10 which may have something to do with the huge wobbles - the star in the pic stays within one correction of the cross though

[whzzz28]

Watching this thread with interest as i too have had oval stars with a good looking graph.

From my research the OSC-index represents the oscillation, or chance that guiding will switch directions during the next pulse. So if OSC is 0.3 then PHD predicts that it has a 30% chance of changing direction in the next pulse.
Why does it go red at low values? I don't know.

RMS is (i believe) the Root Mean Square, number of pixels needed for correction. Lower the better.

A while ago i ran some tests with PHD to see what i could get out of the settings. I took this: http://core-au.net/astro/phd.png when it was all working well (disregard terrible stars on the right side of the image, that is my refractor out of collimation).
For star elongation i normally increase DEC weight, i think the default is 100. Sometimes i need to use 200-300 on bad nights or if my polar alignment is out and guiding has a lot of work to do.

One thing i did do as well was turn off Star Mass (set it to 1.0) as it was causing me tracking grief as well.

[alistairsam]

Its a relief to know that oval stars even with good phd graphs can be guiding issues. I didn't really notice if the index value was red.

Could it have to do with the FL of the guide-scope vs main scope?
So not enough movement of the guidestar in phd to invoke aggressive correction, but shows up in main scope?
The graph basically represents movement of the guidestar. So if that hasn't moved much in phd but shows up as oval stars, I can only think of the FL relationship.
I'm using the 8x50 finder guider and an 8" F4.
Come to think of it, I've never had oval stars with good graphs with an ST80 guidescope and I never had such a good graph with the ST80 either.

I've been so focused on ccd tilt and focuser tilt, I was almost certain it was my focuser not square to the optical axis.
Will go through Nathan's explanation again.
now for some clear skies to test out the settings.

edit: attached are two raw file examples.

[whzzz28]

Your problem might not be guiding related yet.
No matter how good your guiding is, if your polar alignment is far off, your stars will be terrible.

A good test: don't do any guiding. Point the scope somewhere (globular clusters work well) and begin sidereal tracking. Take a 10sec shot. You should have minimal star elongation. If elongation is high, then it's obviously not guiding that is your issue. Maybe try a 20sec exposure as well - elongation shouldn't really be present if your polar alignment is good at this exposure time either.

[whzzz28]

Went rummaging through my first attempts at astrophotography;
Attached is a 30sec unguided exposure. This is with no polar alignment, i simply put the mount facing magnetic south (oops).
So if you see an image like that after 20-30seconds, i'd be taking a good look at your polar alignment.

[alistairsam]

That is quite likely to be the issue as well since I was struggling aligning with alignmaster
Every iteration was yielding large errors, over 10 deg.
But if polar alignment is the main issue then won't the guidestar drift considerably and show up in the graph? How does the guidestar not drift with bad pa?
When my pa was really bad, the guiding graph just kept going out, once I got pa better, the graph improved.
I've read that guiding can't correct pa more than a certain extent, hence the confusion with what the flat graphs and oval stars mean.
I'll focus on alignment first, then increase guiding parameters, then look at any tilt.
I have the cats eye auto collimator, I'm guessing that would show up focuser tilt.

[Merlin66]

Alistair,
There's no reason that PHD would show any problems "guiding" on a poor polar alignment...it just does it's job...lock onto a star and keep it locked by issuing the appropriate signal..
The real issue is that all the other stars in the imaging scope are moving at dufferent rates, depending on the amount of PA error.
So, you can still have "good guiding" but poor imaging results.

[alistairsam]

Ken,
Pardon my ignorance but that's the bit I don't get.

It's a given that stars will drift in the main scope with poor polar alignment, not just drift but possibly result in field rotation.
The guide scope is fairly colinear with the main scope so is pointed at roughly the same area as the imaging scope, possibly 3 times wider if the fov of the guide scope is wider. There is the difference in pixel size of the two camera sensors as well.

Let's say scope is not well aligned, there is no guiding, won't stars at both the main scope and guide scope drift in the same direction? it'll drift a lot less in the guide scope but would surely drift if both scopes are parallel?

Now even if guiding is turned on, the ra and dec graphs merely represent how much the guide star has moved, the correction signals are the second phase, but the graph should show a steadily climbing ra and declining dec graph with poor pa. I've observed this several times where guiding could not correct enough, and the star was constantly lost with the graphs climbing outward.
Once pa was improved, the ra and dec lines did deviate, but a lot less.

The sinusoidal lines are a result of correction being applied and the resultant position of the guide star relative to its last position.
So am still a bit lost with the flat lines, perhaps phd's default recording of the movement is based on neighboring pixel value thresholds or settings, so is ignoring movement under a certain value?

[mithrandir]

PHD can keep the guide star in place even with fairly poor polar alignment by making both RA and Dec corrections.

What it doesn't do in this case is stop field rotation of other objects in the image. A properly polar aligned mount would not have that problem.

[alistairsam]

yep, field rotation due to poor PA but good guiding is easier to understand, oval stars in the same direction at all 4 corners with good guiding graphs is the odd bit, atleast for me.
But from what's been said, its all pointing to poor PA. So will need to improve on that,
The red osc index with low values in Peter's original query might be in phd's faq. will look it up.

Thanks.

[jenchris]

I can get 2 min exposures with no guiding - and round stars.
I'm on a wedge @F10.
I'm pretty sure from that that my PA is pretty good, my biggest worries seem to be from the poor PE and backlash so my balance is critical and guiding is variable according to the portion of sky I'm doing.
This was my setup test without guiding 2 mins single exposure - no tweaks

[alistairsam]

Thats pretty cool Jen, longest i've been able to go with my home built fork mount was 1 min or so with no guiding.

found some good explanations here
http://www.cloudynights.com/item.php?item_id=2755
in the section "I watch PHD's graph and display and all looks great but my stars are still oblong", one possibility suggested is differential flex.

another thread with changing phd settings
http://www.iceinspace.com.au/forum/showthread.php?t=48229

From one user "increase "RA agr" to max 100, and reduce "RA hys" to 5 in order to stop the "Osc-index" going further below "0.19"

[jenchris]

My RA agg is 101

[whzzz28]

Just remember, PHD settings vary based on the focal length of your guide scope. So some tweaking will need to be done for your mini guider.

Personally, i would be looking into your PA first. If alignmaster advises that you are greater than 10" off - then your mount is very far out of alignment.

When i first started i had huge issues getting PA right. Even now i have problems some times.
Just remember - you are after true south, not magnetic south.
The degree off magnetic south varies based on location and time. A calculator is here:
http://www.ngdc.noaa.gov/geomagmodels/Declination.jsp

And since you are using PHD, you can drift align with it.
http://njstargazer.org/PolarAlignment.asp
Basically: Get mount setup as close to true south as possible, set longitude as best as possible. I use a digital inclonometer, and this gets me almost perfect every time! The longitude markings on my HEQ5 are a few degrees out, so i'd recommend not using those unless that is all you have.

Manually move the scope to slightly off true south from park position (i go to the left as facing south) by a few degrees. Keep it as close to the horizon as possible (no more than 30degrees up). Turn on sidreal tracking, tell PHD to calibrate and begin guiding. Once calibrated, stop and go into the settings. Turn DEC guiding off (top right of settings page, should be set to Auto atm). Start guiding again on a star and turn on the PHD graph.
Hit the RA/DEC button which will turn the graph into graphing DX/DY. Look for the red line on the graph (disregard the blue).
If the red line goes up or down (from the middle) then you need to adjust the azimuth screws (move head clockwise/anticlockwise). If you are a long way off, you will actually be able to see the star drift up/down in the main window and the red line will basically shoot off the graph after 1 or 2 guide pulses, and PHD will complain that it lost the guide star. The further off you are, the faster it drifts.
One issue i had was that no matter how much i adjusted the azimuth screws, my line and the speed of drift never changed. I was confused and didn't know why. This occurs if you are a long way out. In my case i was pointing at magnetic south, so greater than 10degrees out. the PHD drift method is very sensitive so your graph will often veer off the screen very fast if you are more than a degree or so out, only when you are close does it stay within the graph screen.

Keep adjusting until the red line is relativly stable near the middle. If you are after perfect polar alignment, then the red line should not go up/down significantly over 10mins or so. If it does, you are slightly out.
Expect a few variations, the graph won't be perfect. This is due to the seeing jumping the star around.
Whilst learning i would suggest just getting it as close as possible, don't wait 10mins and try and get it perfect first time. Getting it close (slight movement (decrease or increase) over 1min) will give you decent images, but not pinpoint stars. Just don't try 10min exposures.

Basically repeat the same thing for logitude (but use the longitude adjust screws) but on another star near the east or west horizon.

If you think you are close, you can always stop guiding and then run alignmaster and it will let you know how far off you are.

Expect to spend nights working on getting this right. It took me two months! but then again, i'm a bit slow with this. :shrug:

Don't forget to turn DEC guiding back to auto before you start guiding/imaging as well!

Once you know your PA is correct, i'd try imaging again. If you still have elongation, and your PHD is guiding OK, then start looking at flex/mirror slip etc.

[jenchris]

Thanks Nathan - I was wondering how that was done!
It's a bit of a brain f**t to work out in your head!

[whzzz28]

I only recently started using PoleAlignMax, i used to do it all via PHD drift alignment.

Instead my procedure is now:
Setup mount as perfect as i can get it manually, this normally is within 5 degrees.
Run a PoleAlignMax and follow the procedures (if you are a long way out, you will probably run out of screw space to change azimuth settings...).
This normally gets me VERY close. I set my CCD to take 1sec exposures continuesly and then turn on cross hairs on my image, move the azimuth/alt a bit and wait for the next image.
Takes about 10mins to do the PoleAlignMax run.
Then start up PHD and do a drift align to fix the last little bit of error.

First time i did this it took me under 20mins to get a good polar alignment. When i was using PHD to drift align it often took me 30+mins.

[widow18]

Thanks everyone, lots of good information there.
Still would like to know why the numbers turned red, I will try and contact the designer.
My oval stars were not field rotation, I was only doing 180 and 120 second exposures anyhow. My finder FL is 400mm and imaging scope is 1000mm, don't know what that does to the equation.
Anyhow, thanks again.
Peter

[Garbz]

Along the centre light. Really low OSC index. From the graph you'd think it's perfect, but the results showed quite clear lack of proper guiding with the stars bouncing around and trailing a bit.

I'm lead to believe that if you've got a very flat line then it's not actually good.


I believe terrible is a very strong word reserved for people who are at the level of doing many minute exposures. Certainly the only time I've ever had an issue with guiding was when I pointed my telescope the wrong side of magnetic south so I was 22deg off. PHD noticed it quite quickly when it failed the calibration steps. Some of my best sharpest and roundest stars have been the result of plonking the telescope down and going for it. Guiding was definitely poor as the stars were moving around quite a bit unguided, and yet despite the PHD graph looking like someone was playing drum and bass next to the scope I got some 3min exposures with perfectly round stars.

From my limited experience field rotation is small enough that if the guiding can keep up with a star using a 1second maximum step (this is dependant on mount, guiding method and scope) then field rotation is not relevant within 1-2 mins.

[widow18]

This just in from Mr Stark.

Regarding red Osc numbers:

"The red there just says "Hey, this is probably not ideal..." The Osc-index shows the odds that two adjacent guide commands (in RA) are in different directions. On a perfect worm gear with no periodic error, this would be 0.5 -- the last motion not telling you anything about the direction of the next one -- pure noise. As we have PE, the ideal value goes down. Let's say it's an ultra-smooth worm but has some real PE (just not a jagged curve). Well, as you go up the curve, the next one is in the same direction as the last. Same deal as you go down it. The only time you change direction is at the very top and bottom of the wave. The ideal Osc-index would be very low.

For most folks, the ideal is somewhere around 0.3-ish-or-so-kinda. I color code it to say "Hey, you may want to check this out." If your guiding was great - great! If not, it may be that you've done something like reduced the aggressiveness too much.


Craig"


thats the answer.
Regards
Peter

[mithrandir]

The other trick to improve guiding GEMs is "east heavy". You'll track better if the mount is slightly off balance to the east side, no matter if it is OTA or counterweights. This means the RA drive is always pushing the scope/weights.
Perfectly balanced means RA pushes, coasts and pulls. This makes guiding the hardest.
West heavy and the scope/weight pulls the RA drive.