Differential flexure or field rotation

[MartinB]

Steve, a guiding log would really help you sort out what is going on. I believe PHD will do this for you and I also think it will give you a live tracking error graph like Maxim. If you are getting egg shaped stars it could be flexion or the guide star oscillation. Oscillation typically occurs when aggression is set too high and the guiding starts chasing the seeing. PHD is able to calculate the centroid position to sub pixel accuracy so guide scope focal length shouldn't be an issue.

[Parallax]

I have just read the article that Peter suggested (many thanks) and there was one paragraph that has struck a chord :-

"If modest pressure on your

guide camera lets you move it relative to the

main scope, you will be limited in how long

you can expose without your stars trailing.

As the mount rotates, gravity acts on the

two setups and flex will result in it acting

on them differently. The guide star may

therefore remain stationary on the guide

camera while the image in the main camera slowly drifts."

I know that Ron alluded to this but I didn't put 2 & 2 together.

This may very well be the problem, as I have noticed movement in this department. I will try and make some modifications and will test it out at the earliest opportunity. I will also investigate the guiding log. Fingers crossed.

Steve

[blinky]

Can somebody just confirm my understanding of the balance bit..... The balance should be offset so the scope is acting 'against' gravity, i.e. the weights end of the scope is the heaviest? I had always made the scope side slightly heavier!

[RobH]

Hi Blinky.

The general rule is to balance the mount so it's very slightly east heavy, and very slightly camera end heavy.

However, you need to experiment with this while watching the way your guiding reacts...easy to do with PHD as you have a realtime graph.

I find that I get better guiding with my scope balanced fractionally west, so it's worth playing with things a bit.

I've been following this thread with interest, as there are obviously quite a few of us who have our guidescopes and cameras rock solidly attached, but are still getting egg shaped stars for some reason.

I was wondering if I was seeing periodic error which isn't being guided out?

I have tried guiding with PEC on, but had very little sucess with this.

Cheers

Rob

[Parallax]

As I understand it, if you are imaging in the west then the scope end should be the heaviest. Conversely for the east. But hey, who am I to talk!

Steve

[RobH]

The reason that it is generally regarded as correct for the scope to be slightly east heavy, is so the drive is always pushing 'up hill', and the gears are always meshed in that direction.

If it's weighted towards the west, the mount will be falling away from the gears. If it is perfectly balanced, there's more likely to be oscillation in the guiding in both east and west directions.

I've no idea at all why my setup likes it the 'wrong way' but it does

Cheers

Rob

[MartinB]

Steve, if you do like you say then the set up will be weighted to the East which is generally what is advised, as Rob says. An even bigger problem is dec corrections since they bring backlash fully into play. Provided you calibration speeds and guide corrections are slower than sidereal rate backlash doesn't affect RA. However backlash will affect guide calibration and guide corrections in dec. If backlash compensation doesn't appeal to you then you have 2 options, one is to slightly misalign the polar alignment so that dec corrections are always in the same direction and the other is to off balance dec so that the gears are always kept together by gravity thus eliminating backlash.

Thes problems would be very apparent in a live tracking graph as Rob suggsts.

[Parallax]

Thanks Martin, all points noted. Next time out I will try everything.

Steve

[Parallax]

Here we are again.

I was up until stupid o'clock last night/this morning logging PHD data and trying out various things. It maybe wishfull thinking but I seem to be getting somewhere.

I took onboard all of the suggestions and tightened things up, locked the focusers, reduced the movement of the webcam in the eyepiece holder, "off" balanced the scope and logged the PHD data.

On the last run the image was only drifting 1 pixel during a 5 minute exposure. I have collated the results from the log (attached is a zipped xls file) but I am not sure what it is I am looking for. I didn't chart all of the data as some of it was due to false starts. You should be able to see on the charts were I have altered the agressiveness. It seems that a value of 60% gives the best results (but I maybe wrong).

It could be that I have reached the limit with this setup - I knew I should have bought the EQ6.

Thanks again for your help everyone.

Steve

PHD_log_28May09.zip

[themos]

well, I have the EQ6 and I noticed exactly the same effect, pixel drift, consistent across 90 minutes. My set up is a 900mm fl reflector feeding a webcam, feeding PHD, feeding EQASCOM pulse guiding. The imaging camera is a piggyback 450D with a 135mm fl lens. Target was Cygnus in the east and every sub was shifted down (or was it up?) by a few pixels (leaving a telltale string of hot pixels). The point is, the drift was consistent in magnitude and direction. I was thinkind differential flexure but somebody at my AstroSoc mentioned field rotation. Surely, field rotation would not show up as uniform translation?

[blinky]

I wonder since a few of us are having the same problem if there is something eqmod related going on? I am sure in one of the mailings I saw that there had been a fix for some kind of drift in one of the later versions.

I wonder if somebody could maybe ask and give an example image - Thats why I am not doing it, I dont ATM have any samples with the effect.

[Psychobilly]

I used to get downward creep bewteen images when I was using the SCT as the guidescope . Since I have moved to using a 500mm tele with the DSI IIC fixed to the back using t-adaptors I no longer get this.

I guess when using the SCT it was mirror flop...

Peter...

[MartinB]

There's no obvious drift on the logs so it looks like a differential flex problem. The flimsiest part of your set up is likely to be the tube rings on your newt.

On a seperate note the graphs are pretty erratic. Look at the dec, it shouldn't be dancing around all over the place like that especially given that your polar alignment looks ok. What guide exposures were you using?

[Parallax]

Guide exposures - I think that they were 1s. I assumed that the PHD would log that info but it doesn't.

Since my last post, I have invested yet more money into the problem and bought a QHY5. This is now securely fixed/screwed to the guide scope to eliminate movement there. I also noted that the lens cell of the GS was moving slightly and have fixed that. Due to the weather, I have been unable to try it out yet.

It does strike me as odd though, that if the guiding is keeping the guide star stationary, then the pixel drift means that the imaging scope is forever moving further away from the guide scope???

Steve

[Whippy]

I know I'm probably trying to get people's grandmothers to suck eggs here, but I have (probably still have!) this problem constantly and these things helped:

I found that having a really good polar alignment helps a great deal especially on longer subs (greater than 5 mins). I assume this is because the closer the mount is to the pole, then the less corrections the guiding software has to make therefore, better tracking. I've been using WCS and it's been doing the job nicely.

Another thing which helped is dovetails. I bought an chunky ADM 'tail and again, it seems to have made a difference from the stock Synta model.

WRT balance, my understanding is that you have your setup balanced one way for one half of the sky and the other way around for the erm other. Ie: imaging in the East requires the mount to be slightly counterweight heavy and camera heavy whilst it's the other way around imaging in the West.

I'm sure some of you already do all of this so I'm preaching to the converted (and I still do all that and I'll still get oval stars next time out!) and there are other factors like wind but I've found that these things do help, especially if you're going 'deep'.

Tony..

[MartinB]

All good points Tony.

Steve, I think it likely that the seeing wasn't too good during this run hence the erratic graph, the other alternative is a saturated guide star. There is no discernible drift due to poor polar alignment. You could therefore turn the aggressiveness way down to 0.1 or 0.2. increase the guidestar exposure to around 3 secs but make sure you don't take the guide star above 2/3 saturated.

That's all beside the point though. The drift can all be in a constant direction. If your main scope is well balanced the fore and aft flex is going to be minimal but the newt exerts significant lateral forces on those tube rings and the flex will progressively increase/decrease as the cross section of the scope moves away from or towards the horizontal if that makes sense. If your changes don't do the job then you should start looking at beefing up the tube rings and the dovetail. Those skywatcher rings are very bendy. People pay a fortune on imaging kit and then skimp on a couple of rings, doesn't make sense.

[Parallax]

Cheers Martin. I'll see how I get on when the weather permits. If there is no improvement, then, as you say, it's time to look at the rings and dovetail(s).

Thanks again for everyone's help. I am feeling confident that I'm almost there.

Steve

[Psychobilly]

Any comments on this guide graph please...

[ATTACH]23224[/ATTACH]

CP800XLT on Pier and Wedge guiding via DSI IIc on 500mm f8 the test target was Vega... 1.0s guide exposure...

Thanks Peter...

[blinky]

Looks a hell of a lot better than any of my graphs Peter!

[Psychobilly]

I'm getting my best star shapes ever at the moment...

The scopes on the fixed pier and wedge and I have started doing a 2 star alignment using a 12mm illuminated retucule EP in a x3 Barlow.

I forgot to check the reported error in polar alignment I know its a bit out but I am leaving well alone at least until the scope goes back onto the pier after Salisbury...

Peter...

[blinky]

It's still daylight here, no way I could see any stars!

[MartinB]

I don't know what the scale is on the graph Peter and I'm not sure what is meant by oscillation index but the dec has some pretty ugly spikes. I suspect the problem is using Vega as a guide star. It will almost certainly be saturated with a 1 second exposure which will totally destroy the accuracy of the centroid. This means that the software will be working on a disc rather than a point and will do a fair bit of wandering about. The result will be nice round stars but not as tight as they could be.

[Psychobilly]

Thanks martin...

I do have the log files..

PHD Analyzer is a tool for analysing log files i found a download link here...

Peter...

[barkis]

Since TJ does not use a guidescope, but relies upon an off axis guider which takes a guidestar from the main imaging scope,

has he experienced the same flexure indications Steve?

I know TJ's system still has possible filter wheel and camera

movement, so may still be prone to the same problem.

I have not seen a post from Tim himself in this thread, so it's possible he has missed it.

Ron.

[Psychobilly]

I dont know how PHD scales the graphs either Martin...

An PHD analyzer... well ... looks like i need to strip the guide logs down to a more managble size...

Here's soem inso from the web...

Here is what I got from PHD FAQ

Log file information

Q. Just what are the numbers in the log file? What units are they?

A. PHD deals in pixels (distance), seconds or milliseconds (time), and radians (angles). It never needs to know things like how long your focal length is because it only works in units that can be directly observed - how far the star moved in pixels in a given amount of time. Thus, we have:

Frame: Counter telling you what image it was, starting at 1

Time: Time since the start of the run, in seconds

dx: Distance the star was from the lock position along the camera’s X-axis, in pixels

dy: Distance the star was from the lock position along the camera’s Y-axis, in pixels

Theta: Angle formed by creating a line from the lock position to the star position, in radians

RA_Dist: Projection of the vector from lock to star position onto the RA axis (as determined in the calibration stage). Distance in pixels

RA_Dur: Duration of the RA pulse sent (in msec)

Dec_Dist: Projection of that vector along the Dec axis (as determined in the calibration stage)

Dec_Dur: Duration of the Dec pulse sent (in msec).

Some of these numbers (dx and dy) show up during guiding as well.

Q. Is there any good way to look at all those numbers in the log file?

A. There are two. First, you can import the numbers into a spreadsheet. This is just ASCII text with the data in “CSV” (comma separated variables) format. So tell your spreadsheet that commas are delimiters.

Second, Windows users can use Marcelo’s PHD Log Analyzer. This is available on the Yahoo Group.

Peter...