Over the last couple of months I have been trying to sort out the pointing on my PME. With the help of Marcus I eventually decided to buy CCDsoft and augment it with The Sky6, Tpoint and Automapper II. Once I sorted a few minor settings I found that all this software works well together and actually performs the mapping quite well.
However several things have been concerning me.
First, I get quite a few errors with regard to either WSC not being used or pattern is not being recognised. Typically I am getting 48 errors for every 82 accepted on the plate solve. I am using 20 seconds per image and then auto dark subtraction. Do I need vary the gain, or is this to do with the number of stars in AutomapperII or something else completely?
Next, once I have completed a mapping run I am not able to get polar alignment really fantastic. I think it is good but not fantastic. See image below. For polar alignment I have done 5 runs with 20 map points per run. Then I did 2 at 30. Yet the previous run to this I was required to move 3 ticks in each axis. What am I doing wrong here? This polar alignment looks good but I have rejected 7 map points to get it looking like this at present. Two points were just terrible and obviously incorrectly mapped.
The of course is the final and most troubling problem. No matter what I do I just cannot get my RMS pointing accuracy below 100. I wonder if the pattern shown in the image is indicative of an orthagonality issue or something else? To be honest my pointing when I did a manual mapping run was near 100 and this is the best I can do with this data.
What I really need is some advise from guys in the know, I am really stuck (I normally can nut this sort of stuff out, but this time I just don't know what to do) and need help.
Well, im supprised you need 20secs per image to plate solve. I use 7 secs at bin 3 (lum filter) without a problem, with maybe 2 rejects per 40 images. If you get this failure rate with 20 images, thats 10 good solves, not good, look into that 1st (6 is the min for any result).
Does your map have a good spread across the sky?.
I use AAG Tpoint mapper BTW (free), ive heard its more user friendly and it shows a picture of your scan map.
The Tpoint polar align data (and adjustment recommend) is always bang on for me.
to be honest I could use 7 second images, but each image looks so noisey from the 20 second that I just did not think it would help the plate solves. Next run I will give it a go. Yes the hit miss ratio is quite high. I am using the QSI as the imager and I would have thought it would be fine. Interestingly I have to use the TSA to image with as the GSO could not plate solve at all. I put this down to the notion that the scope is probably not an f8 as reported and this affects the scale all the time.
The map is spread all over the sky from 40 degrees up to 80.
I was going to use Tpoint Mapper and had it installed but I had to pay for Pin point for that system to operate (at least I think this is what is required). I don't think there is a problem with the operation of CCDsoft, The Sky6, Tpoint and AutomapperII. So given I just paid for CCDsoft I would not be keen to use Tpoint Mapper.
I am also wondering if my image scale is not 100% correct. I am using the flattener also and I think that affects the focal length by 2mm. I will need to look this up to find out for sure.
Do you think the fit model data looks odd Fred? What are your thoughts on the RMS figure? Un adjusted everything is way of to the right. I suspect it is time related. I am using internet update and that could well be a problem too. I would welcome your thoughts on that issue. The modelling I can live with, but not getting good pointing is bothering me.
edit. sorry the flattener makes it 800mm for the TSA and that means a significant difference. I will try that tomorrow night.
The fit data looks a typical shape, but over quite a large area.
Yes, image scale is important and should be correct. I didnt adjust that, AAG seemed to extract that automatically from the sky (or FITS image data)
If you update every 10 mins or so from yr internet time scource, that would be OK, it wouldnt trash the model.
Its worth paying IMO for pinpoint for several apps, DL uses it to plate solve and calibrate "point here" nudging. Apparently Sky X has its own built-in version. I understand though that the CCDsoft has a cut down version build in, so that should do (or is that DL, you can platesolve, so thats OK I guess).
I had similar problems until I increased the star mag sensitivity (in AAG) to 20 mag, more stars to measure with I guess.
You really need to get the plate sovle success rate up, that seems wrong, and may indicate a different rpoblem.
PA isnt bad just the scatter is a concern.
With that amount of scatter your PA result is in the noise zone anyway.
You will be going back and forth each time you do a new run !
The screen shot you provided lists that you have 11 terms in use.
I would think that is too many for that level of uncertainty and that small number of points, the extra terms will probably be doing more harm than good, but not that bad that the RMS be at 164 - ie to say its not the main cause.
Print out the Fit Information Dialog Box so we can see what they are and what they are doing.
Try removing all but the standard terms and get the polar alignment sorted out first before adding all extra terms.
You really only need just a few mapped points more than the bare minimum (6) to get PA pretty accurate - 10 should be more than enough for the first physical PA adjustment. 6-8 for a first PA should work.
RMS (164 arcsec) is high because the scatter is widely spread.
You need to think about the cause of this and fix it in order to get the RMS lower, its not going to go away by itself just because you have a Tpoint model.
The scatter diagram shows a definite North/South spread so its not across all axis - flexure in your imaging train maybe ?
Wind, vibration, backlash, (PE ? not sure) etc could all be possibilities.
Are these mapped points across both meridians ?
Try mapping just on one side and see if the scatter is more localised before doing the second side.
Are you deleting the old model after manually adjusting PA ? - Just a thought !
What is the orientation of your camera with respect to North/South and East/West ?
Rally, I know about going back and forth, I have done this more often than not in the last couple of months.
The 11 terms actually are 5 extras, but each one makes it better. I think I started with 280 odd. I have read the instructions so give me credit here. 10% for each term or not at all. Right? Like you though I did ask the question about the RMS, I am trying to sort out what is going on there too.
Please see attached a copy of the dialogue box. Let me know if you can figure out what is going on.
I take your point about the points. I have done heaps of runs to get PA sorted but after 7 or 8 shortish runs I would have thought that PA is nailed? Don't you agree?
I agree that the Tpoint is not going to sort out things, but I was seeking an explanation as to why the RMS is so scattered. I knew it was odd but could not find any explanation over the net for this particular pattern.
I don't think it is flexure on the imaging train. You know what the Tak scopes are like, but I will check the screws holding the unit in place just in case. No wind, Clayton hardly ever has wind at night, lots during the day but not at night. Should not be vibration, you know the pier and I have fixed the connections with this adapter plate. Could be backlash, PEC not completely sorted yet. I bought Precision PEC to sort that. Tomorrow night I am going down again and I will do the PEC first, then work on PA.
Mapped points are scattered across the meridian but I reckon there was a lot more on the East side than the West. Maybe 15% on West. Not sure how to correct this as Automapper II does not seem to have an adjustment for that. I will take a look in the Sky6 and see if I can use a mapping model from that. If you have a suggestion here I would be interested.
Yes deleting old model after each adjustment.
Orientation to North is about 77 degrees according to WSC research. Good idea, I guess I need to shift that to North or near to it?
Appreciate the time you put into your reply but I have read pretty extensively on this and the instruction manuals can be a little thin. Suggestions are really appreciated.
Edit: sorry trying to put the fit info box up in text version
Last edited by Paul Haese; 25-04-2011 at 09:23 PM.
I'm under the impression that in addition to deleting the old Tpoint model after each adjustment, you also need to synchronize again (i.e., start from ground zero). I synchronize against an image link rather than a star.
Be sure to bin when auto mapping. It won't affect your accuracy. I bin 4x4 with a QSI-583 (on a C-14 @ f/11) and use 5 to 10 second exposures.
I'm using the latest TheSkyX with the Tpoint Add On. It works *much* better than TheSky6. For each 30 point run I get maybe one or two image link failures.
The Tpoint Add On supermodel feature works well, but requires lots of data to really truly shine (get down to 10 or so arc-seconds RMS) pointing. At least 150 points in my experience. But I never use more than 30 or so points per iteration when polar aligning.
If Tpoint polar alignment isn't working out for you, why don't you try using PEMPRO? PEMPRO has a drift alignment function function that many people swear by. PEMPRO also has a 60-day trial so that you can kick the tires before you buy.
Its pretty hard to sort out these problems with limited info and not being present to see how its occurring..
I can only suggest the obvious and the not so obvious - more info helps.
I note you have removed the CH term - Why did you remove one of the standard 6 geometric terms ?
I would add that term back in and remove TX to DNP in your list for starters and see if that improves anything.
The Fit Information indicates that TX is no good.
HCEC is possibly OK, DCES no good, DCEC no good, DAF & DNP definitely no good. Sigmas are much too high for their values.
This is all stuff you can follow through from the manual for adding new terms.
When you add the terms you also need to look at how they affect model and this needs to be done as you do it one by one - I can't tell from seeing it at the end of the run.
Best way to do this is create a spreadsheet and calculate out the values and see if they improve the result by enough to warrant inclusion - term by term.
The fact that you have two very distinct groups of clustered data points makes wonder if your camera has sagged either side of meridian, but I am not sure why that isnt reflected in TX
How does the camera attach to your OTA ?
Taks have focus tubes and if not adjusted tightly they can sag, the Tak CAA's always need tightening and adjustment to eliminate backlash, adapters can be loose in the threads due to faulty tolerancing in the threads or adjustment etc
The model shows something isnt right - so not much point saying it cant be this or that - try putting some finger pressure on your camera and see if you can detect any movement, I used a dial indicator on mine and was amazed at what I discovered
PA as I said isnt bad, and I am guessing it may be closer than Tpoint is reporting based on the model you have created.
But ME could be a lot better than 81.
How does it look on screen when viewed at say 400% - your monitor cross hairs should be touching each star.
But how good does it need to be ?
I guess it boils down to the question of what do you want to achieve - what limit do you set yourself ? - there are errors in the system because they are mechanical devices made by people ! so that has to be accepted.
If you are trying to get 0.0 arc secs - Its simply not possible.
So what is your acceptable value ?
If you set it too low you will be fighting the noise in your system and never be happy - Guiding and PEC will fix all of what you have got left there no problem at all !
Fred - Thanks mate - pester me till its right
I am no expert on this.
Last edited by rally; 26-04-2011 at 11:03 AM.
Reason: removed repeated sentence !
I'm under the impression that in addition to deleting the old Tpoint model after each adjustment, you also need to synchronize again (i.e., start from ground zero). I synchronize against an image link rather than a star.
Be sure to bin when auto mapping. It won't affect your accuracy. I bin 4x4 with a QSI-583 (on a C-14 @ f/11) and use 5 to 10 second exposures.
I'm using the latest TheSkyX with the Tpoint Add On. It works *much* better than TheSky6. For each 30 point run I get maybe one or two image link failures.
The Tpoint Add On supermodel feature works well, but requires lots of data to really truly shine (get down to 10 or so arc-seconds RMS) pointing. At least 150 points in my experience. But I never use more than 30 or so points per iteration when polar aligning.
If Tpoint polar alignment isn't working out for you, why don't you try using PEMPRO? PEMPRO has a drift alignment function function that many people swear by. PEMPRO also has a 60-day trial so that you can kick the tires before you buy.
Thanks Ernie,
Yes been synchonizing with a start before the model is started. Not done that with a image link every time though. Might try that.
I was binning at 3x3 but read in one of the manuals that 1x1 was most accurate. So I compromised and tried 2x2 and found accuracy went up a little from 3x3. Might just be an errant thing though.
I don't really want to buy SkyX if I can avoid it really, but will if I absolutely have to do so.
Can't use Pempro as I have already gone past the 60 days and would have to buy yet another piece of software. I had trouble getting it to talk to all my gear too on Vista.
Some things to think about and I will try 4x4 with 5-10 second images. Second person to recommend this so definitely seems the go.
Rally, sorry was really cranky last night after spending several weeks just doing all the reading and trying to get this working well but things were defeating me a little. I am really happy all the system works under Vista but getting better results is the issue for me.
I tried changing the 6 main terms to see what effect it had. So I am with you there, this produced the best pointing effect. I agree not necessarily the best result either.
Yes looking through the fits file again last night and having read Brad Moores recommendations of sigma being less than half of the value. No matter what I do here the issue is not the model itself at present it is the way the data is gathered. Until that gets good results then the model manipulation is really a moot point.
The camera is attached using the standard Tak tension screws (yes it should have a screw in adapter but I have the flattener in place for both the tak and the RC and there is no screw thread on the flattener just an inserted end to the focusor) and could well as you point out be causing the problem. I will check this out tonight and use your suggestion of using some pressure to determine if there is some flex. I did this for eliminating flexure before when I was getting bad differential flexure so I know what you are suggesting here. It had not occurred to me that this might be an issue. Good tip this one. I will check it out.
PA results seem fairly easy to manipulate though. Removing a couple of stars can affect the model and suggest good PA. I am going to check with my own eyes tonight and see what drift I am getting. I take your point though about the ME. It ought to be a lot better than that for such PA.
How good does it need to be. Hey I would be happy with 50-70 RMS pointing. 40 would be great as far as I am concerned but 50 is realistic I think. That would put me near the center of the sensor each time. as 125 was putting me on the edge when I did this all by manual. I then simply nudged to get it center. I just want better point than I did have. Certainly have no illusions that this mount could be anything near the value you suggested. Yes the guiding and PEC do sort this but it is the pointing I am after.
Tpoint will help you identify physical/mechanical errors - use that info to physically fix these errors rather than try to use it to correct them in software.
Flexure and sagging with single screw or compression fittings and nose pieces is erratic - is doesnt shift at all until you get to a certain point and then it just goes (or not), as opposed to the more linear sort of bend with tube or bracket flexure.
Tpoint cant fix that sort of problem.
I am surprised you are using this sort of fitting in your rig and at the same time expecting precision results in the arc second level. Its a waste of time.
I thought all Tak flatteners had screw threads at each end ?
I am sure that you will be able to visibly see the camera moving with a light finger touch - try say 1/2kg equivalent of force - it will only be a small amount of movement but I feel sure it will be there. It may not be the same in all directions depending on where the screws are.
So you have multiple points for potential flexure
The Focusser, the CAA rotator (both should be able to be adjusted out with that lightweight camera), the 2" screw down adapter, the flattener itself maybe, and whatever method is used to attach to the camera itself
RTM
Read the Tpoint manual and follow it - its your best guide for using Tpoint - its states the same as you have mentioned ie 2x Sigma plus a lot of other stuff that needs to be followed if you want to add multiple extra terms.
TheSkyX will help you automate the extra terms much better.
Otherwise dont bother with them - if the system is stable and the PA good - its going to be well within your desired range.
You are much more likely to get yourself into deeper trouble if you dont understand what you are doing with Tpoint terms.
I am surprised you are using this sort of fitting in your rig and at the same time expecting precision results in the arc second level. Its a waste of time.
I thought all Tak flatteners had screw threads at each end ?
RTM
Read the Tpoint manual and follow it - .
Rally
Point taken and I suspected this might actually be part of the problem or it solely. I will need to get either a new focusor or a screw fixing adapter for the Tak focusor. The flattener for the TOA/TSA series is a slide in version. I do have a reducer though and might get the adapter made up for it to the Camera by Precise Parts. I had one made for the flattener to camera. However this is most likely the problem.
I will re read all the manuals again with particular emphasis on the Tpoint one.
Once I get this nutted I will report back what the problem was. It might help someone else down the track.
Paul, if you haven't already done so, then please go to the www.bisque.com website, register as a user and post concerning the problem in either the Paramount ME or the Tpoint for Windows support forums. They'll want to see your Tpoint model. Patrick Wallace (Mr Tpoint god himself) frequents the site. Software Bisque has quite a good support forum.
It's just all noise. There is something random going on. And not random
in a systematic way, like non-orthogonality. Random as in throw the dice random.
However, the good news is that the results are telling us that, so it is still good information.
Adding additional terms won't help at this stage.
Could you please show the results from the following fit commands? (i.e. the fits text files including as you did before).
1) ID & IH only
2) ID, IH, MA, ME only
3) ID, IH, MA, ME, NP & CH only
Could you please also supply the scatter diagram that results from each of the above fits.
Knowing the results from the above might be able to tell me more.
However it could well be a case of 'garbage in, garbage out'.
For example, few things will throw out the results more than incorrectly identified/mapped stars.
What would also be useful is if you could do a File Text Export command and post the text file.
Based on the FIT text file you have already posted, the fact that the sigmas
are nearly as large as their corresponding terms tells me that it is noise.
The MA and ME terms are also giveaways.
These two terms have very distinctive mathematical signatures and are thus
normally not strongly correlated to other terms. But they too show that
something random is likely to be occurring.
Thanks Gary, I will try to get those requests sorted.
As to random looking data. I now think you are right.
Last night I did a drift polar alignment and it was out, drifting in just 2 minutes. So I drift aligned until it was 20 minutes on each axis. No more stuffing around with that. I will not be trusting Tpoint until I get the pointing issue sorted.
Then I checked for flex as Rally suggested. Non on the telescope, connection to mount or camera connection to the scope. All really rigid. I am really satisfied that this is ok too.
Next I took out the flattener and just tried mapping with the camera and scope at native focal length. Same results. North South separation and then something occurred to me.
On my mount if I stand with the RA pointed towards me and I gently grab the counter weight bar there is noticable slop in the RA axis (I am talking 10mm here; up and down perpendicular to the axis); so if you are holding the bar and the scope is say laying over in RA I can move the bar up and down by 10 mm. My camera is set up so that long axis of the sensor is perpendicular to the Dec axis. That should make the north at almost 90 to the RA. Would this slop account for the north south separation? When I took the face plate off some time ago I noted that the whole worm and gear move as one and that the assembly does not appear to be bolted to the body fo the mount, it sort of floats. Is this normal? I would have thought that this sort of slop would prevent good pointing. I think this is most likely the cause of the problem.
Gary, the odd thing is that many images are ok on the astrometric solution in CCDsoft but they get rejected in the sky or Tpoint (not sure which is rejecting, but I get the message in automapper that it has been rejected.. I get two main errors; insert WSC failure and cannot map. Last night I got 140 errors and 93 positives on one run. That means there is something really wrong physcially. Often the circled star in CCDsoft (the star being pointed toward) is all over the place. Sometimes near the cross hairs, but most often at random distances and places from the cross hairs. I think it might well be the mount work that was done before I bought it that is causing the problem.
Having read the posts, it definitely sounds like you have one or several mechanical issues to resolve. This could be slop in the mount as you mentioned, it could also be your adapters. Even the most rigid adapters have some degree of movement, its a matter of minimising it as you cannot remove it all.
In my opinion there is not much point doing 100+ tpoint runs until you have sorted the mechanical issues out first. Tpoint will not eliminate these for you alone.
When you are at the stage where you have minimised as many mechanical flaws as possible then go for it!
Adding terms is also not recommended until you have a substantial Tpoint model of about 200-300 points, remember this model is statistical, so adding or removing the wrong terms can be disastrous.
If you are using the tutorial, create yourself a spreadsheet and take the time in doing it. My first serious model with added terms took me a good 3 hours, and that was just doing the calculations and methodically checking the results. Remember that if you add a term and it has positive results it is likely that it has an adverse effect on another term and this is only apparent if you do the math in the spreadsheet.
Lastly, I personally had an 80% plate solve failure rate with CCDSoft and TheSky6 and it did not take me long to realise PinPoint was the only way to go. Now I have a 99% success rate in plate solving.
Brad Moore has come to the rescue and like me thinks there is a hardware problem. I have to do some investigation inside the mount. A PDF is on its way to me (incidently this is how software bisque fixes mount hardware problems - you get a comprehensive instruction on how to fix the problem and you fix it yourself. I like this because I will sort the problem) to check and tension bearings and fixings. I will have to take some video of the problem and send it off to Brad.
So I will get back to you all very soon and let you know how it works out.
.
