VintageBigBlue.org

 

RE: [Losmandy_users] PE as a function of number of cycles used by PemPro


Sep 17, 2008

 


----------------------------

#39769 Sep 17, 2008

As mentioned earlier, my data confirm the dependence of PE reported by

PemPro on the number of cycles used. It is more complicated than just

averaging data but here are the results from a run on 9-13-08 near meridian

at Dec about +10:







5 cycles RMS 2.84 +4/-5.1 peak-peak 9.1

RMS 2.6



15 cycles RMS 3.7 +3.4/-2.7 peak-peak 6.1

RMS 1.6



25 cycles RMS 4.32 +2.1/-2.1 peak-peak 4.1

RMS 1.31







All in arcsecs. Fitting of the drift using 5th order curve.







This is one way to make a mount look better although the RMS of the fit

increases with the number of cycles. The last two columns come from legend

on the PEC curve ready to be uploaded to Gemini. Here the RMS decreases with

the number of cycles (it has completely different meaning - it does not

pertain, I believe, to the curve fitting of the drift as the RMS in the

first column).







Best,



LN







[Non-text portions of this message have been removed]



----------------------------

#39770 Sep 17, 2008

Ladislav,



As the RMS error increases that means the accuracy of the data fit goes

down. As you increased cycles the RMS increased, probably because the worm

wheel teeth in your mount are very different (or that even 5th order drift

fitting couldn't correct the drift). Thus the least squares fitting can't do

much with your 15/25 cycle data (i.e., garbage in, garbage out).



The second RMS number, for the PE curve is the RMS error in arc-secs

compared to perfect PEC correction.



Thanks,



-Ray

> -----Original Message-----

> From: Losmandy_users@yahoogroups.com

> [mailto:Losmandy_users@yahoogroups.com] On Behalf Of Ladislav Nemec

> Sent: Wednesday, September 17, 2008 2:17 PM

> To: Losmandy_users@yahoogroups.com

> Subject: [Losmandy_users] PE as a function of number of

> cycles used by PemPro

>

> As mentioned earlier, my data confirm the dependence of PE

> reported by PemPro on the number of cycles used. It is more

> complicated than just averaging data but here are the results

> from a run on 9-13-08 near meridian at Dec about +10:

>

> 5 cycles RMS 2.84 +4/-5.1 peak-peak 9.1

> RMS 2.6

>

> 15 cycles RMS 3.7 +3.4/-2.7 peak-peak 6.1 RMS 1.6

>

> 25 cycles RMS 4.32 +2.1/-2.1 peak-peak 4.1 RMS 1.31

>

> All in arcsecs. Fitting of the drift using 5th order curve.

>

> This is one way to make a mount look better although the RMS

> of the fit increases with the number of cycles. The last two

> columns come from legend on the PEC curve ready to be

> uploaded to Gemini. Here the RMS decreases with the number of

> cycles (it has completely different meaning - it does not

> pertain, I believe, to the curve fitting of the drift as the

> RMS in the first column).

>

> Best,

>

> LN

>

> [Non-text portions of this message have been removed]

>

>

>

>

>



----------------------------

#39774 Sep 17, 2008

Ladislav,



In the spirit of openness, the PE numbers I have been quoting for the

Losmandy HP worm and the Ovision worm were both based on 10 cycles. I

will be interested in your results.



Frank

celestialwonders.com



--- In Losmandy_users@yahoogroups.com, "Ladislav Nemec" nemecl@...>

wrote: >

> It definitely looks that way. My point, however, was that while

reporting PE > errors based on PemPro, the number of cycles should also be stated. PE

> errors, not RMS of drift fitting, is what mostly concerns us and is

relevant > to the quality of the worm itself. The worm wheel (as you call it,

worm gear > is a possible synonym but it may get confused with the 'worm'

itself, worms, > after all, are kind of gear, I think - two very different parts of the

> mechanical assembly) is only tangentially handled by PemPro in the

sense you > describe.

>

>

>

> I would go as far as to state that reporting PE (or related

peak-peak of the > final PEC curve) is not very meaningful without specifying how many worm

> cycles were used.

>

>

>

> In my case the peak-peak variation of the PEC curve went from 9

arcsecs to > 4.3 arcsecs by increasing the number of cycles from 5 to 25, very

> significant decrease.

>

>

>

> The question is: should one upload the 5 cycle PEC to the mount, the 15

> cycle PEC or the 25 cycle PEC? I think only actual experiment will

tell us. > Very simple: uploading sequentially all three PEC curves and

measuring the > quality of the corresponding tracking with PEC on will decide. I have no

> idea what to expect.

>

>

>

> If things were only bit more simple!

>

>

>

> Best, LN

>

>

>

> -----Original Message-----

> From: Losmandy_users@yahoogroups.com

[mailto:Losmandy_users@yahoogroups.com] > On Behalf Of Ray Gralak

> Sent: Wednesday, September 17, 2008 2:28 PM

> To: Losmandy_users@yahoogroups.com

> Subject: RE: [Losmandy_users] PE as a function of number of cycles

used by > PemPro

>

>

>

> Ladislav,

>

> As the RMS error increases that means the accuracy of the data fit goes

> down. As you increased cycles the RMS increased, probably because

the worm > wheel teeth in your mount are very different (or that even 5th order

drift > fitting couldn't correct the drift). Thus the least squares fitting

can't do > much with your 15/25 cycle data (i.e., garbage in, garbage out).

>

> The second RMS number, for the PE curve is the RMS error in arc-secs

> compared to perfect PEC correction.

>

> Thanks,

>

> -Ray

>

> > -----Original Message-----

> > From: Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

> yahoogroups.com

> > [mailto:Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

> yahoogroups.com] On Behalf Of Ladislav Nemec

> > Sent: Wednesday, September 17, 2008 2:17 PM

> > To: Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

> yahoogroups.com

> > Subject: [Losmandy_users] PE as a function of number of

> > cycles used by PemPro

> >

> > As mentioned earlier, my data confirm the dependence of PE

> > reported by PemPro on the number of cycles used. It is more

> > complicated than just averaging data but here are the results

> > from a run on 9-13-08 near meridian at Dec about +10:

> >

> > 5 cycles RMS 2.84 +4/-5.1 peak-peak 9.1

> > RMS 2.6

> >

> > 15 cycles RMS 3.7 +3.4/-2.7 peak-peak 6.1 RMS 1.6

> >

> > 25 cycles RMS 4.32 +2.1/-2.1 peak-peak 4.1 RMS 1.31

> >

> > All in arcsecs. Fitting of the drift using 5th order curve.

> >

> > This is one way to make a mount look better although the RMS

> > of the fit increases with the number of cycles. The last two

> > columns come from legend on the PEC curve ready to be

> > uploaded to Gemini. Here the RMS decreases with the number of

> > cycles (it has completely different meaning - it does not

> > pertain, I believe, to the curve fitting of the drift as the

> > RMS in the first column).

> >

> > Best,

> >

> > LN

> >

> > [Non-text portions of this message have been removed]

> >

> >

> >

> >

> >

>

>

>

>

>

> [Non-text portions of this message have been removed]

>







----------------------------

#39777 Sep 17, 2008

It definitely looks that way. My point, however, was that

> while reporting PE errors based on PemPro, the number of

> cycles should also be stated. PE errors, not RMS of drift

> fitting, is what mostly concerns us and is relevant to the



Just to be clear the periodic error of the worm does NOT change if you do 1

cycle or 50 cycles.



What changes the result is the drift fitting accuracy and NON-periodic

errors caused by imperfections in each tooth of the worm wheel.



If you try to do 25 cycles of data the best fit (or by most other means)

will reduce the reported periodic error of the worm because it is averaging

non-periodic imperfections in the worm wheel. That's why the RMS error is

the true indication of the accuracy of the curve fit. In a good mount the

RMS error will be 0.2-0.3 arc-seconds. In your case you had RMS errors

10-15x as high. Note that an RMS of that magnitude means that there some of

the data is 2-5x in error with the curve fit (that could mean 15 arc-seconds

in your case).



So, the cycle count doesn't matter, just the RMS error. In your case with

4-5 arc-sec RMS the PEC curve cannot be trusted at all.



-Ray



----------------------------

#39779 Sep 17, 2008

As you probably know, I do not have and do not plan to have Ovision. The

results in my original e-mail are very comparable to the Ovision results - I

used 5, 15 and 25 cycles, 10 cycle results will be somewhere between 5 and

15, obviously.







I have some much data that I am unable to review them fast enough. I found,

for instance, that doing meridian flip on a star initially east of meridian

and going west all the way to altitude of some 35 degrees (refraction still

not serious), I have actually even better results - bit hard to believe for

the original Losmandy worm assembly (some 2 arcsecs peak to peak.). I think

that 10 cycles may be kind of standard. Nobody wants to waste hours (except

during a full Moon) to get some 30+ cycles as I have been doing for the past

week or so. It can be argued whether 5, 7 or 10 is the right number. As I

mentioned before, it is fairly easy to find out what PEC curve works best. I

may try it tonight.







Also, for the first time Paul clarified to me (others probably understood

that long time ago) how to use the PemPro feature of adding constant drift

correction to the uploaded PEC curve. Another thing to do during Large Moon.







Thanks for additional info - we may agree on a 'standard' number of cycles

for comparisons, 10 seems to be about right.







LN







-----Original Message-----

From: Losmandy_users@yahoogroups.com [mailto:Losmandy_users@yahoogroups.com]

On Behalf Of Frank Barrett

Sent: Wednesday, September 17, 2008 5:03 PM

To: Losmandy_users@yahoogroups.com

Subject: [Losmandy_users] Re: PE as a function of number of cycles used by

PemPro







Ladislav,



In the spirit of openness, the PE numbers I have been quoting for the

Losmandy HP worm and the Ovision worm were both based on 10 cycles. I

will be interested in your results.



Frank

celestialwon celestialwonders.com> ders.com



--- In Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com, "Ladislav Nemec" nemecl@...>

wrote: >

> It definitely looks that way. My point, however, was that while

reporting PE > errors based on PemPro, the number of cycles should also be stated. PE

> errors, not RMS of drift fitting, is what mostly concerns us and is

relevant > to the quality of the worm itself. The worm wheel (as you call it,

worm gear > is a possible synonym but it may get confused with the 'worm'

itself, worms, > after all, are kind of gear, I think - two very different parts of the

> mechanical assembly) is only tangentially handled by PemPro in the

sense you > describe.

>

>

>

> I would go as far as to state that reporting PE (or related

peak-peak of the > final PEC curve) is not very meaningful without specifying how many worm

> cycles were used.

>

>

>

> In my case the peak-peak variation of the PEC curve went from 9

arcsecs to > 4.3 arcsecs by increasing the number of cycles from 5 to 25, very

> significant decrease.

>

>

>

> The question is: should one upload the 5 cycle PEC to the mount, the 15

> cycle PEC or the 25 cycle PEC? I think only actual experiment will

tell us. > Very simple: uploading sequentially all three PEC curves and

measuring the > quality of the corresponding tracking with PEC on will decide. I have no

> idea what to expect.

>

>

>

> If things were only bit more simple!

>

>

>

> Best, LN

>

>

>

> -----Original Message-----

> From: Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com

[mailto:Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com] > On Behalf Of Ray Gralak

> Sent: Wednesday, September 17, 2008 2:28 PM

> To: Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com > Subject: RE: [Losmandy_users] PE as a function of number of cycles

used by > PemPro

>

>

>

> Ladislav,

>

> As the RMS error increases that means the accuracy of the data fit goes

> down. As you increased cycles the RMS increased, probably because

the worm > wheel teeth in your mount are very different (or that even 5th order

drift > fitting couldn't correct the drift). Thus the least squares fitting

can't do > much with your 15/25 cycle data (i.e., garbage in, garbage out).

>

> The second RMS number, for the PE curve is the RMS error in arc-secs

> compared to perfect PEC correction.

>

> Thanks,

>

> -Ray

>

> > -----Original Message-----

> > From: Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

> yahoogroups.com

> > [mailto:Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

> yahoogroups.com] On Behalf Of Ladislav Nemec

> > Sent: Wednesday, September 17, 2008 2:17 PM

> > To: Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

> yahoogroups.com

> > Subject: [Losmandy_users] PE as a function of number of

> > cycles used by PemPro

> >

> > As mentioned earlier, my data confirm the dependence of PE

> > reported by PemPro on the number of cycles used. It is more

> > complicated than just averaging data but here are the results

> > from a run on 9-13-08 near meridian at Dec about +10:

> >

> > 5 cycles RMS 2.84 +4/-5.1 peak-peak 9.1

> > RMS 2.6

> >

> > 15 cycles RMS 3.7 +3.4/-2.7 peak-peak 6.1 RMS 1.6

> >

> > 25 cycles RMS 4.32 +2.1/-2.1 peak-peak 4.1 RMS 1.31

> >

> > All in arcsecs. Fitting of the drift using 5th order curve.

> >

> > This is one way to make a mount look better although the RMS

> > of the fit increases with the number of cycles. The last two

> > columns come from legend on the PEC curve ready to be

> > uploaded to Gemini. Here the RMS decreases with the number of

> > cycles (it has completely different meaning - it does not

> > pertain, I believe, to the curve fitting of the drift as the

> > RMS in the first column).

> >

> > Best,

> >

> > LN

> >

> > [Non-text portions of this message have been removed]

> >

> >

> >

> >

> >

>

>

>

>

>

> [Non-text portions of this message have been removed]

>











[Non-text portions of this message have been removed]







----------------------------

#39780 Sep 17, 2008

This is news to me. You are saying that a PEC curve with a legend of 4

arcsecs peak-peak should not be used for Gemini upload since it cannot be

trusted if the RMS of the drift fit was (previous step) more than, say, 2

arcsecs. Never occurred to me.







I think the time has come to put it for a test, hopefully tonight. As I

mentioned in a previous message, I seem to observe huge differences between

PemPro data obtained with the scope east or west of the meridian. I think

one should upload a PEC curve obtained in the east when working east of

meridian. I did not post my data west of meridian - I can hardly believe how

low RMS, PE errors and PEC peak-peak errors are. Maybe I am doing something

wrong. I will have to double-check (I have several thousands of point

obtained from about 1 hour before meridian, transiting and going down to

altitude of about 35 degrees to avoid any significant refraction effects).

Fascinating work, not very useful for actual corrections but illustrating

the behavior of my equipment.







My current idea is that the weight distribution and corresponding flexing of

the optical elements may be better when the scope is west of the meridian -

my setup is not east-west symmetric out of necessity.







I do not understand this part of your message:







'Note that an RMS of that magnitude means that there some of the data is

2-5x in error with the curve fit (that could mean 15 arc-seconds in your

case).'







When I load the pempro .txt data in the analysis section, a curve indicating

the PE errors using the n-th degree fitting of the drift is created. Two

quantities are reported: the RMS of the fit (that varies strongly with the

order of the fit) and the +/- values that I understood were (more or less)

the peak-to-peak values, somewhat averaged. These values (in my case) are

usually very low and depend only slightly on the order of the fit for the

drift.







Now you are saying that I have 15 arcsecs (peak-peak?) fast variations? I

have, indeed, individual points that are that are that far from each other

on the raw data. The process of creating uploadable PEC (next and separate

step) has to include some kind of smoothing - the uploadable PEC curves are

usually very smooth. Obviously, now and then the actual tracking errors will

not be corrected for by such PEC curve but, statistically, a significant

improvement should be seen.







Well, the best proof of a pudding is in eating. Planning to do just that

tonight. Still pretty bright here.







Best, LN











-----Original Message-----

From: Losmandy_users@yahoogroups.com [mailto:Losmandy_users@yahoogroups.com]

On Behalf Of Ray Gralak

Sent: Wednesday, September 17, 2008 5:13 PM

To: Losmandy_users@yahoogroups.com

Subject: RE: [Losmandy_users] PE as a function of number of cycles used by

PemPro







> It definitely looks that way. My point, however, was that

> while reporting PE errors based on PemPro, the number of

> cycles should also be stated. PE errors, not RMS of drift

> fitting, is what mostly concerns us and is relevant to the



Just to be clear the periodic error of the worm does NOT change if you do 1

cycle or 50 cycles.



What changes the result is the drift fitting accuracy and NON-periodic

errors caused by imperfections in each tooth of the worm wheel.



If you try to do 25 cycles of data the best fit (or by most other means)

will reduce the reported periodic error of the worm because it is averaging

non-periodic imperfections in the worm wheel. That's why the RMS error is

the true indication of the accuracy of the curve fit. In a good mount the

RMS error will be 0.2-0.3 arc-seconds. In your case you had RMS errors

10-15x as high. Note that an RMS of that magnitude means that there some of

the data is 2-5x in error with the curve fit (that could mean 15 arc-seconds

in your case).



So, the cycle count doesn't matter, just the RMS error. In your case with

4-5 arc-sec RMS the PEC curve cannot be trusted at all.



-Ray











[Non-text portions of this message have been removed]



----------------------------

#39781 Sep 17, 2008

I just looked at Frank's results with Losmandy and Ovision. I did not find

the values of his RMS that, according to your message, is essential for

decision whether the final (uploadable) PEC curve is useful or not. My

Losmandy PEC (by that I mean the uploadable) curve looks bit better than his

Losmandy curve but worse than his Ovision curve as one would expect.







It would be of some interest to know what his RMS data for curve fitting

(NOT the RMS data of the final PEC uploadable curve) were.







It's getting darker here.







LN







-----Original Message-----

From: Losmandy_users@yahoogroups.com [mailto:Losmandy_users@yahoogroups.com]

On Behalf Of Ray Gralak

Sent: Wednesday, September 17, 2008 5:13 PM

To: Losmandy_users@yahoogroups.com

Subject: RE: [Losmandy_users] PE as a function of number of cycles used by

PemPro







> It definitely looks that way. My point, however, was that

> while reporting PE errors based on PemPro, the number of

> cycles should also be stated. PE errors, not RMS of drift

> fitting, is what mostly concerns us and is relevant to the



Just to be clear the periodic error of the worm does NOT change if you do 1

cycle or 50 cycles.



What changes the result is the drift fitting accuracy and NON-periodic

errors caused by imperfections in each tooth of the worm wheel.



If you try to do 25 cycles of data the best fit (or by most other means)

will reduce the reported periodic error of the worm because it is averaging

non-periodic imperfections in the worm wheel. That's why the RMS error is

the true indication of the accuracy of the curve fit. In a good mount the

RMS error will be 0.2-0.3 arc-seconds. In your case you had RMS errors

10-15x as high. Note that an RMS of that magnitude means that there some of

the data is 2-5x in error with the curve fit (that could mean 15 arc-seconds

in your case).



So, the cycle count doesn't matter, just the RMS error. In your case with

4-5 arc-sec RMS the PEC curve cannot be trusted at all.



-Ray











[Non-text portions of this message have been removed]







----------------------------

#39782 Sep 17, 2008

I did not look at Frank's graphs, sorry about that. He used 10 cycles and

the RMS for the Losmandy worm was 2.52, for Ovision mere 0.78 (still

substantially higher than Ray's values of 0.2 - 0.3 arcsecs). Also Frank

used linear fitting for Losmandy worm, quartic fitting for Ovision worm.







My curves look quite different - with 25 cycles the highest entry on the sec

spectrum graph is at about 6000 second, followed by the evil 76 sec.

Virtually no true periodic error. Based on that it is probably useless to

upload ANY PEC curve to the mount, no periodic errors, no PEC.







With 5 cycles, there is a 4 mins periodic error, obviously no 6000 second

error, a comparable error at about 800 seconds and, of course, the 76 second

error - all three about the same amplitude.







Sorry for introducing the issue of varying the number of cycles

significantly. It appears that it has some influence on both the analysis

and corrections using PemPro. It may be useful to upload the PEC curve from

the 5 cycle results (there is, after all, a periodic error). PEC curve

obtained with 25 cycles is of dubious value - may make things even worse.







Complications, complications.







LN







-----Original Message-----

From: Losmandy_users@yahoogroups.com [mailto:Losmandy_users@yahoogroups.com]

On Behalf Of Ladislav Nemec

Sent: Wednesday, September 17, 2008 6:25 PM

To: Losmandy_users@yahoogroups.com

Subject: RE: [Losmandy_users] PE as a function of number of cycles used by

PemPro







I just looked at Frank's results with Losmandy and Ovision. I did not find

the values of his RMS that, according to your message, is essential for

decision whether the final (uploadable) PEC curve is useful or not. My

Losmandy PEC (by that I mean the uploadable) curve looks bit better than his

Losmandy curve but worse than his Ovision curve as one would expect.



It would be of some interest to know what his RMS data for curve fitting

(NOT the RMS data of the final PEC uploadable curve) were.



It's getting darker here.



LN



-----Original Message-----

From: Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com [mailto:Losmandy_users@

mailto:Losmandy_users%40yahoogroups.com> yahoogroups.com]

On Behalf Of Ray Gralak

Sent: Wednesday, September 17, 2008 5:13 PM

To: Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com

Subject: RE: [Losmandy_users] PE as a function of number of cycles used by

PemPro

> It definitely looks that way. My point, however, was that

> while reporting PE errors based on PemPro, the number of

> cycles should also be stated. PE errors, not RMS of drift

> fitting, is what mostly concerns us and is relevant to the



Just to be clear the periodic error of the worm does NOT change if you do 1

cycle or 50 cycles.



What changes the result is the drift fitting accuracy and NON-periodic

errors caused by imperfections in each tooth of the worm wheel.



If you try to do 25 cycles of data the best fit (or by most other means)

will reduce the reported periodic error of the worm because it is averaging

non-periodic imperfections in the worm wheel. That's why the RMS error is

the true indication of the accuracy of the curve fit. In a good mount the

RMS error will be 0.2-0.3 arc-seconds. In your case you had RMS errors

10-15x as high. Note that an RMS of that magnitude means that there some of

the data is 2-5x in error with the curve fit (that could mean 15 arc-seconds

in your case).



So, the cycle count doesn't matter, just the RMS error. In your case with

4-5 arc-sec RMS the PEC curve cannot be trusted at all.



-Ray



[Non-text portions of this message have been removed]











[Non-text portions of this message have been removed]



----------------------------

#39783 Sep 17, 2008

This is news to me. You are saying that a PEC curve with a

> legend of 4 arcsecs peak-peak should not be used for Gemini

> upload since it cannot be trusted if the RMS of the drift fit

> was (previous step) more than, say, 2 arcsecs. Never occurred to me.



It's not a hard and fast number. For instance if the periodic error is 100

arc-seconds peak-peak, 4 arc-secs RMS is pretty good relative to the curve.

However, if you have a 1 arc-sec periodic error curve with 5 arc-sec RMS,

then the curve is not fitting the data very well.



That said a high RMS error might not be from an inaccurate PEC curve but

from poor drift fitting. 25 cycles of data is probably not going to get

fitted very well with even a 5th order polynomial.

> I do not understand this part of your message:

>

> 'Note that an RMS of that magnitude means that there some of

> the data is 2-5x in error with the curve fit (that could mean

> 15 arc-seconds in your case).'

>

> Now you are saying that I have 15 arcsecs (peak-peak?) fast

> variations?



Maybe. But more likely some data points are 15+ arc-secs from the fitted

curve. The reason could be poor drift fitting, worm wheel tooth variations,

optics moving, cables tugging, seeing over a roof, etc., etc., etc.



-Ray







----------------------------

#39786 Sep 17, 2008

I did not look at Frank's graphs, sorry about that. He used

> 10 cycles and the RMS for the Losmandy worm was 2.52, for

> Ovision mere 0.78 (still substantially higher than Ray's



0.78 is still VERY good. The best mounts I have ever seen are in the 0.2-0.3

arc-sec range. Poor seeing can also increase the RMS error. Rough gearing or

particles in the grease can cause high RMS. Each mount is different. You

need to look at the data to see how well it fits. Speaking of which, would

you mind uploading your raw log files to the files section here so I can

look at them?

> My curves look quite different - with 25 cycles the highest

> entry on the sec spectrum graph is at about 6000 second,

> followed by the evil 76 sec.

> Virtually no true periodic error. Based on that it is

> probably useless to upload ANY PEC curve to the mount, no

> periodic errors, no PEC.



Please upload your 25 cycle data and I'll take a look at it. It could be

that you had phase shifting occuring on your PC from time shifts on your PC.

> Sorry for introducing the issue of varying the number of

> cycles significantly. It appears that it has some influence

> on both the analysis and corrections using PemPro. It may be

> useful to upload the PEC curve from the 5 cycle results

> (there is, after all, a periodic error). PEC curve obtained

> with 25 cycles is of dubious value - may make things even worse.



As I said earlier you should look at the RMS error relative to periodic

error amplitude to determine if the curve is accurate. I have done 20-cycle

periodic error files on my AP1200GTO and got a relatively low RMS (less than

1 arc-sec). Your mount is obviously acting differently.



-Ray



----------------------------

#39790 Sep 18, 2008

Optics moving - definitely. Cables tugging, over roof seeing - absolutely

not. I am quite sorry that, at this late date I do not seem to understand

the basics. No reason to blame anything and anyone but myself. PemPro says

what it means: RMS for drift fitting and drift, apparently, means ANYTHING

that is not periodic with the worm. PE +/- values mean PERIODIC ERROR

variation. If there is very little of PERIODIC ERROR, the values are very

low. Consequently, in my opinion, PEC curve should not be even created -

message saying FORGET IT should be displayed. As Rolando writes in another

context on another forum, authors/manufacturers expect that users got

sufficient number of functioning brain cell. In my case, actually, recent

MRI of my brain has shown shrinking of the think with growing lacunae.







That's the way it is. Only now, around midnight, the clouds disappeared

completely and I may climb on top of my fire water tank and get going.







It's not easy to work with some of us (i.e. Mr. Ladislav Nemec, for

instance). Thanks for your patience.







LN







-----Original Message-----

From: Losmandy_users@yahoogroups.com [mailto:Losmandy_users@yahoogroups.com]

On Behalf Of Ray Gralak

Sent: Wednesday, September 17, 2008 7:27 PM

To: Losmandy_users@yahoogroups.com

Subject: RE: [Losmandy_users] PE as a function of number of cycles used by

PemPro





> This is news to me. You are saying that a PEC curve with a

> legend of 4 arcsecs peak-peak should not be used for Gemini

> upload since it cannot be trusted if the RMS of the drift fit

> was (previous step) more than, say, 2 arcsecs. Never occurred to me.



It's not a hard and fast number. For instance if the periodic error is 100

arc-seconds peak-peak, 4 arc-secs RMS is pretty good relative to the curve.

However, if you have a 1 arc-sec periodic error curve with 5 arc-sec RMS,

then the curve is not fitting the data very well.



That said a high RMS error might not be from an inaccurate PEC curve but

from poor drift fitting. 25 cycles of data is probably not going to get

fitted very well with even a 5th order polynomial.

> I do not understand this part of your message:

>

> 'Note that an RMS of that magnitude means that there some of

> the data is 2-5x in error with the curve fit (that could mean

> 15 arc-seconds in your case).'

>

> Now you are saying that I have 15 arcsecs (peak-peak?) fast

> variations?



Maybe. But more likely some data points are 15+ arc-secs from the fitted

curve. The reason could be poor drift fitting, worm wheel tooth variations,

optics moving, cables tugging, seeing over a roof, etc., etc., etc.



-Ray











[Non-text portions of this message have been removed]



----------------------------

#39791 Sep 18, 2008

AP people are the rich relatives of us mere Losmandy/Gemini people. I am

monitoring their forum and they have completely different issues to dicuss.







Mechanical precision is, I am afraid, the core of many of my problems.

That's why AP mounts are more than twice expensive than Losmandys. It may be

heroic (and futile) exercise to bring some of the Losmandys (not necessarily

all) to the AP standard.







LN







-----Original Message-----

From: Losmandy_users@yahoogroups.com [mailto:Losmandy_users@yahoogroups.com]

On Behalf Of Ray Gralak

Sent: Wednesday, September 17, 2008 7:47 PM

To: Losmandy_users@yahoogroups.com

Subject: RE: [Losmandy_users] PE as a function of number of cycles used by

PemPro





> I did not look at Frank's graphs, sorry about that. He used

> 10 cycles and the RMS for the Losmandy worm was 2.52, for

> Ovision mere 0.78 (still substantially higher than Ray's



0.78 is still VERY good. The best mounts I have ever seen are in the 0.2-0.3

arc-sec range. Poor seeing can also increase the RMS error. Rough gearing or

particles in the grease can cause high RMS. Each mount is different. You

need to look at the data to see how well it fits. Speaking of which, would

you mind uploading your raw log files to the files section here so I can

look at them?

> My curves look quite different - with 25 cycles the highest

> entry on the sec spectrum graph is at about 6000 second,

> followed by the evil 76 sec.

> Virtually no true periodic error. Based on that it is

> probably useless to upload ANY PEC curve to the mount, no

> periodic errors, no PEC.



Please upload your 25 cycle data and I'll take a look at it. It could be

that you had phase shifting occuring on your PC from time shifts on your PC.

> Sorry for introducing the issue of varying the number of

> cycles significantly. It appears that it has some influence

> on both the analysis and corrections using PemPro. It may be

> useful to upload the PEC curve from the 5 cycle results

> (there is, after all, a periodic error). PEC curve obtained

> with 25 cycles is of dubious value - may make things even worse.



As I said earlier you should look at the RMS error relative to periodic

error amplitude to determine if the curve is accurate. I have done 20-cycle

periodic error files on my AP1200GTO and got a relatively low RMS (less than

1 arc-sec). Your mount is obviously acting differently.



-Ray











[Non-text portions of this message have been removed]







----------------------------

#39792 Sep 18, 2008

Shrinking of the THING. What can I add?



LN







-----Original Message-----

From: Losmandy_users@yahoogroups.com [mailto:Losmandy_users@yahoogroups.com]

On Behalf Of Ladislav Nemec

Sent: Thursday, September 18, 2008 12:17 AM

To: Losmandy_users@yahoogroups.com

Subject: RE: [Losmandy_users] PE as a function of number of cycles used by

PemPro







Optics moving - definitely. Cables tugging, over roof seeing - absolutely

not. I am quite sorry that, at this late date I do not seem to understand

the basics. No reason to blame anything and anyone but myself. PemPro says

what it means: RMS for drift fitting and drift, apparently, means ANYTHING

that is not periodic with the worm. PE +/- values mean PERIODIC ERROR

variation. If there is very little of PERIODIC ERROR, the values are very

low. Consequently, in my opinion, PEC curve should not be even created -

message saying FORGET IT should be displayed. As Rolando writes in another

context on another forum, authors/manufacturers expect that users got

sufficient number of functioning brain cell. In my case, actually, recent

MRI of my brain has shown shrinking of the think with growing lacunae.



That's the way it is. Only now, around midnight, the clouds disappeared

completely and I may climb on top of my fire water tank and get going.



It's not easy to work with some of us (i.e. Mr. Ladislav Nemec, for

instance). Thanks for your patience.



LN



-----Original Message-----

From: Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com [mailto:Losmandy_users@

mailto:Losmandy_users%40yahoogroups.com> yahoogroups.com]

On Behalf Of Ray Gralak

Sent: Wednesday, September 17, 2008 7:27 PM

To: Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com

Subject: RE: [Losmandy_users] PE as a function of number of cycles used by

PemPro

> This is news to me. You are saying that a PEC curve with a

> legend of 4 arcsecs peak-peak should not be used for Gemini

> upload since it cannot be trusted if the RMS of the drift fit

> was (previous step) more than, say, 2 arcsecs. Never occurred to me.



It's not a hard and fast number. For instance if the periodic error is 100

arc-seconds peak-peak, 4 arc-secs RMS is pretty good relative to the curve.

However, if you have a 1 arc-sec periodic error curve with 5 arc-sec RMS,

then the curve is not fitting the data very well.



That said a high RMS error might not be from an inaccurate PEC curve but

from poor drift fitting. 25 cycles of data is probably not going to get

fitted very well with even a 5th order polynomial.

> I do not understand this part of your message:

>

> 'Note that an RMS of that magnitude means that there some of

> the data is 2-5x in error with the curve fit (that could mean

> 15 arc-seconds in your case).'

>

> Now you are saying that I have 15 arcsecs (peak-peak?) fast

> variations?



Maybe. But more likely some data points are 15+ arc-secs from the fitted

curve. The reason could be poor drift fitting, worm wheel tooth variations,

optics moving, cables tugging, seeing over a roof, etc., etc., etc.



-Ray



[Non-text portions of this message have been removed]











[Non-text portions of this message have been removed]



----------------------------

#39796 Sep 18, 2008

Ladislav,



Moving optics, floppy focusers, etc. are all independent of the

mount. I bet your OTA mounted on an AP1200 or a Paramount ME would

have similar problems with drift. Sounds to me like you have some

mirror flop or possibly a sagging focuser.



To do a good PEC run (and you really only need to do this once in a

long while!) try to use a refractor. Doesn't matter if it's a good

one or not. A simple cheap achromat will do, just make sure you have

a solid focuser that can be tightened down, and a medium focal

length.



As I posted here earlier, my RMS fitting error with a Titan PE and a

refractor, using linear fit is about 0.7 arcseconds on a night of

reasonably bad seeing:



pk.darkhorizons.org/tmp/Titan-PE.jpg



On a night of good seeing, RMS error is about 0.5, and PE after

programming PEC is near 1 arcsecond.



I've tried to do PemPro runs with my 10" SCT in the past and gave up

rather quickly: it's got a very active, movable mirror :-(.



Regards,



-Paul





--- In Losmandy_users@yahoogroups.com, "Ladislav Nemec" nemecl@...>

wrote: >

> AP people are the rich relatives of us mere Losmandy/Gemini

people. I am > monitoring their forum and they have completely different issues

to dicuss. >

>

>

> Mechanical precision is, I am afraid, the core of many of my

problems. > That's why AP mounts are more than twice expensive than Losmandys.

It may be > heroic (and futile) exercise to bring some of the Losmandys (not

necessarily > all) to the AP standard.

>

>

>

> LN







----------------------------

#39802 Sep 18, 2008

Ladislav,



You wrote: > Optics moving - definitely. Cables tugging, over roof seeing

> - absolutely not. I am quite sorry that, at this late date I

> do not seem to understand the basics. No reason to blame

> anything and anyone but myself. PemPro says what it means:

> RMS for drift fitting and drift, apparently, means ANYTHING

> that is not periodic with the worm. PE +/- values mean

> PERIODIC ERROR variation.



Well, that's not completely correct...



First, PEMPro best-fits the data based on finding the least-squares fit to a

multi-frequency sine wave and drift curve. If you want to know something

about leasts square do a google search or look at this link:

en.wikipedia.org/wiki/Least_squares



After obtaining the best fit periodic and drift curves they are subtracted

from the data points. The remainder is called the RESIDUAL error.



KEY POINT--> The RMS value is calculated soley from the residual error.



The residual error could contain the sum of atmospheric seeing, unmodelled

periodic terms, unmodeled drift, mirror movements, worm wheel tooth defects,

etc.



The residual data may or may not have periodic components left in it. For

instance if you unchecked fundamental 3 but your mount has the 76 second

fundamental then that will be present in the residual and it contributes to

increasing the RMS error.



You wrote: > PEC curve should not be even created - message saying FORGET

> IT should be displayed. As Rolando writes in another context



There are so many posibilities that it is almost impossible to make a

decision on the quality of a particular PEC curve. That's why it is

recommended that after uploading a PEC curve you measure the residual

periodic error again and refine if needed (BTW, this is something I don't

think you did).



If you try to sample over 25 worm cycles and and there are all kinds of

influences that cannot be modelled well then you are not going to get good

(PEC) results. As I said garbage in, garbage out. You need to have clean

data.



So, that said, I want a chance to examine your data curves to see if there

is any way I can improve the modeling of the data. So please upload your raw

log files to the files section (if you haven't deleted the data.



-Ray



----------------------------

#39803 Sep 18, 2008

--- In Losmandy_users@yahoogroups.com, "Ray Gralak" rgr@...> wrote: > If you try to sample over 25 worm cycles and and there are all kinds

of > influences that cannot be modelled well then you are not going to

get good > (PEC) results.



I'd guess that splitting up Ladilav's 25 cycles into much shorter

runs, of say 5 cycles each, would help quite a bit by making the drift

fitting work better.



Regards,



-Paul



----------------------------

#39806 Sep 18, 2008

Hi Paul,



Yes,I think that is a good idea. I still hope Ladislav will post his log

files so I can get a chance to analyze what happened.



-Ray

> -----Original Message-----

> From: Losmandy_users@yahoogroups.com

> [mailto:Losmandy_users@yahoogroups.com] On Behalf Of Paul K

> Sent: Thursday, September 18, 2008 7:18 AM

> To: Losmandy_users@yahoogroups.com

> Subject: [Losmandy_users] Re: PE as a function of number of

> cycles used by PemPro

>

> --- In Losmandy_users@yahoogroups.com

> mailto:Losmandy_users%40yahoogroups.com> , "Ray Gralak"

> rgr@...> wrote:

> > If you try to sample over 25 worm cycles and and there are all kinds

> of

> > influences that cannot be modelled well then you are not going to

> get good

> > (PEC) results.

>

> I'd guess that splitting up Ladilav's 25 cycles into much

> shorter runs, of say 5 cycles each, would help quite a bit by

> making the drift fitting work better.

>

> Regards,

>

> -Paul

>

>

>

>

>



----------------------------

#39815 Sep 18, 2008

I suspect that much. My another long run last night, however, seems to

indicate that - at least during a part of the run - that the 'active

mirror' (is it similar to restless leg syndrome?) was probably not the

culprit. A strange wide 'valley' appeared on the continuous drift curves

from both the C11 main OTA and the piggybacked 4 inch Mak. It seems that the

whole thing sagged somewhat.







I think that a Mak is almost as good as a refractor - my model focuses by

moving the corrector plate/secondary using a the threaded enclosure. To make

sure, I added two set screws to fix its position after achieving reasonable

focus. In any case, I thing I have a 'refractor' (a spotting scope,

actually) that can be attached firmly to the OTA.







So far I have not observed significant differences between PemPro data from

my ST7 attached to the Mak and ST8 attached to the main OTA. I have not,

however, looked very closely - too much data, too little time.







Still, sagging of the whole thing (including the tripod) cannot be easily

(i.e. I did not figure out yet) distinguished from an imprecise worm wheel

(as Ray calls it).







Just for the fun I did PEC with a curve uploaded from Gemini and the only

result was that very small RA drift without PEC turned into major RA drift

with PEC. Did not try to play with the linear drift adjustment.







G8, G11 and Titan are progressively more expensive and, it seems, also

better, Titan probably quite close to AP mounts.







Well, it's time to finish this 'exploration', the Moon will soon be

cooperative.







Soon I will post a PemPro log (zipped?) with the data I reported recently.

Last night data look little bit different with about the same RMS errors for

drift fitting but - something new - with that strange wide 'valley'.







Thanks, you have been very helpful all the time.







LN







-----Original Message-----

From: Losmandy_users@yahoogroups.com [mailto:Losmandy_users@yahoogroups.com]

On Behalf Of Paul K

Sent: Thursday, September 18, 2008 5:08 AM

To: Losmandy_users@yahoogroups.com

Subject: [Losmandy_users] Re: PE as a function of number of cycles used by

PemPro







Ladislav,



Moving optics, floppy focusers, etc. are all independent of the

mount. I bet your OTA mounted on an AP1200 or a Paramount ME would

have similar problems with drift. Sounds to me like you have some

mirror flop or possibly a sagging focuser.



To do a good PEC run (and you really only need to do this once in a

long while!) try to use a refractor. Doesn't matter if it's a good

one or not. A simple cheap achromat will do, just make sure you have

a solid focuser that can be tightened down, and a medium focal

length.



As I posted here earlier, my RMS fitting error with a Titan PE and a

refractor, using linear fit is about 0.7 arcseconds on a night of

reasonably bad seeing:



pk.darkhoriz pk.darkhorizons.org/tmp/Titan-PE.jpg>

ons.org/tmp/Titan-PE.jpg



On a night of good seeing, RMS error is about 0.5, and PE after

programming PEC is near 1 arcsecond.



I've tried to do PemPro runs with my 10" SCT in the past and gave up

rather quickly: it's got a very active, movable mirror :-(.



Regards,



-Paul



--- In Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com, "Ladislav Nemec" nemecl@...>

wrote: >

> AP people are the rich relatives of us mere Losmandy/Gemini

people. I am > monitoring their forum and they have completely different issues

to dicuss. >

>

>

> Mechanical precision is, I am afraid, the core of many of my

problems. > That's why AP mounts are more than twice expensive than Losmandys.

It may be > heroic (and futile) exercise to bring some of the Losmandys (not

necessarily > all) to the AP standard.

>

>

>

> LN











[Non-text portions of this message have been removed]







----------------------------

#39817 Sep 18, 2008

I never delete anything, I just buy new disks. Well, I was not entirely

correct but close, I think. Basically I assumed that the RESIDUAL ERROR was

everything that could not be represented by a Fourier series. It is actually

what's left after fitting the 'long term' drift curve to an n-th order curve

and the periodic (based on the worm cycle period) Fourier representation.







Thanks for the correction (do I have it right now?). Years ago I used to do

some least square fitting - completely forgot the details, though. Prefer to

leave the details to you.







It is interesting to compare the frequency spectra of RA (x) and Dec (y)

data. No 76 s error is present (as one would expect) on the Dec data. The

'weight' of the 76 s error depends on the number of cycles used - I will

post several large sets of the logs in the file section, zipped they should

not take that much space. BTW, on virtually all sets of data I looked at,

the 76 s error has a 'value' of some 1.6 arcsecs associated with it - is it

a typical low or high value? Seems low to me but PemPro data are filled with

subtleties.







It should not take me too long to organize several sets of recent data and

add some reasonable descriptions and brief comments. Definitely today.







Thanks for your patience.







LN















-----Original Message-----

From: Losmandy_users@yahoogroups.com [mailto:Losmandy_users@yahoogroups.com]

On Behalf Of Ray Gralak

Sent: Thursday, September 18, 2008 6:55 AM

To: Losmandy_users@yahoogroups.com

Subject: RE: [Losmandy_users] PE as a function of number of cycles used by

PemPro







Ladislav,



You wrote: > Optics moving - definitely. Cables tugging, over roof seeing

> - absolutely not. I am quite sorry that, at this late date I

> do not seem to understand the basics. No reason to blame

> anything and anyone but myself. PemPro says what it means:

> RMS for drift fitting and drift, apparently, means ANYTHING

> that is not periodic with the worm. PE +/- values mean

> PERIODIC ERROR variation.



Well, that's not completely correct...



First, PEMPro best-fits the data based on finding the least-squares fit to a

multi-frequency sine wave and drift curve. If you want to know something

about leasts square do a google search or look at this link:

en.wikipedia en.wikipedia.org/wiki/Least_squares>

.org/wiki/Least_squares



After obtaining the best fit periodic and drift curves they are subtracted

from the data points. The remainder is called the RESIDUAL error.



KEY POINT--> The RMS value is calculated soley from the residual error.



The residual error could contain the sum of atmospheric seeing, unmodelled

periodic terms, unmodeled drift, mirror movements, worm wheel tooth defects,

etc.



The residual data may or may not have periodic components left in it. For

instance if you unchecked fundamental 3 but your mount has the 76 second

fundamental then that will be present in the residual and it contributes to

increasing the RMS error.



You wrote: > PEC curve should not be even created - message saying FORGET

> IT should be displayed. As Rolando writes in another context



There are so many posibilities that it is almost impossible to make a

decision on the quality of a particular PEC curve. That's why it is

recommended that after uploading a PEC curve you measure the residual

periodic error again and refine if needed (BTW, this is something I don't

think you did).



If you try to sample over 25 worm cycles and and there are all kinds of

influences that cannot be modelled well then you are not going to get good

(PEC) results. As I said garbage in, garbage out. You need to have clean

data.



So, that said, I want a chance to examine your data curves to see if there

is any way I can improve the modeling of the data. So please upload your raw

log files to the files section (if you haven't deleted the data.



-Ray











[Non-text portions of this message have been removed]



----------------------------

#39818 Sep 18, 2008

I did that - the splitting (by editing the original file) is bit tricky.

PemPro first says it cannot load the data - somewhere a total number of

points is there but I could not find it and edit it. Then it loads the

'truncated' set anyway.







As I reported yesterday, the RMS error of the drift fit increases, indeed,

with the number of cycles (I used 5, 15 and 25). Even with 5 cycles it is

still too high for creation of a 'useful' PEC curve - that's I understand

Ray's comments.







LN







-----Original Message-----

From: Losmandy_users@yahoogroups.com [mailto:Losmandy_users@yahoogroups.com]

On Behalf Of Paul K

Sent: Thursday, September 18, 2008 7:18 AM

To: Losmandy_users@yahoogroups.com

Subject: [Losmandy_users] Re: PE as a function of number of cycles used by

PemPro







--- In Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com, "Ray Gralak" rgr@...> wrote: > If you try to sample over 25 worm cycles and and there are all kinds

of > influences that cannot be modelled well then you are not going to

get good > (PEC) results.



I'd guess that splitting up Ladilav's 25 cycles into much shorter

runs, of say 5 cycles each, would help quite a bit by making the drift

fitting work better.



Regards,



-Paul











[Non-text portions of this message have been removed]







----------------------------

#39822 Sep 18, 2008

Ladislav,



I looked at your log files and most or all of them appear to NOT be original

PEMPro files. I know you mentioned you were going to do this but there seems

to be missing lots of informnation from the files (like gemini data, ascom

drivers, etc).



Please post the originals and don't try to cut/paste/truncate anything. I'm

going to have to put CRCs on each line (tied to a line number) to make sure

people are not changing these log files. :-)



Thanks,



-Ray

> -----Original Message-----

> From: Losmandy_users@yahoogroups.com

> [mailto:Losmandy_users@yahoogroups.com] On Behalf Of Ladislav Nemec

> Sent: Thursday, September 18, 2008 4:13 PM

> To: Losmandy_users@yahoogroups.com

> Subject: RE: [Losmandy_users] PE as a function of number of

> cycles used by PemPro

>

> I never delete anything, I just buy new disks. Well, I was

> not entirely correct but close, I think. Basically I assumed

> that the RESIDUAL ERROR was everything that could not be

> represented by a Fourier series. It is actually what's left

> after fitting the 'long term' drift curve to an n-th order

> curve and the periodic (based on the worm cycle period)

> Fourier representation.

>

> Thanks for the correction (do I have it right now?). Years

> ago I used to do some least square fitting - completely

> forgot the details, though. Prefer to leave the details to you.

>

> It is interesting to compare the frequency spectra of RA (x)

> and Dec (y) data. No 76 s error is present (as one would

> expect) on the Dec data. The 'weight' of the 76 s error

> depends on the number of cycles used - I will post several

> large sets of the logs in the file section, zipped they

> should not take that much space. BTW, on virtually all sets

> of data I looked at, the 76 s error has a 'value' of some 1.6

> arcsecs associated with it - is it a typical low or high

> value? Seems low to me but PemPro data are filled with subtleties.

>

> It should not take me too long to organize several sets of

> recent data and add some reasonable descriptions and brief

> comments. Definitely today.

>

> Thanks for your patience.

>

> LN

>

> -----Original Message-----

> From: Losmandy_users@yahoogroups.com

> mailto:Losmandy_users%40yahoogroups.com>

> [mailto:Losmandy_users@yahoogroups.com

> mailto:Losmandy_users%40yahoogroups.com> ] On Behalf Of Ray Gralak

> Sent: Thursday, September 18, 2008 6:55 AM

> To: Losmandy_users@yahoogroups.com

> mailto:Losmandy_users%40yahoogroups.com>

> Subject: RE: [Losmandy_users] PE as a function of number of

> cycles used by PemPro

>

> Ladislav,

>

> You wrote:

> > Optics moving - definitely. Cables tugging, over roof seeing

> > - absolutely not. I am quite sorry that, at this late date I do not

> > seem to understand the basics. No reason to blame anything

> and anyone

> > but myself. PemPro says what it means:

> > RMS for drift fitting and drift, apparently, means ANYTHING that is

> > not periodic with the worm. PE +/- values mean PERIODIC ERROR

> > variation.

>

> Well, that's not completely correct...

>

> First, PEMPro best-fits the data based on finding the

> least-squares fit to a multi-frequency sine wave and drift

> curve. If you want to know something about leasts square do a

> google search or look at this link:

> en.wikipedia

> en.wikipedia.org/wiki/Least_squares

> en.wikipedia.org/wiki/Least_squares> > .org/wiki/Least_squares

>

> After obtaining the best fit periodic and drift curves they

> are subtracted from the data points. The remainder is called

> the RESIDUAL error.

>

> KEY POINT--> The RMS value is calculated soley from the

> residual error.

>

> The residual error could contain the sum of atmospheric

> seeing, unmodelled periodic terms, unmodeled drift, mirror

> movements, worm wheel tooth defects, etc.

>

> The residual data may or may not have periodic components

> left in it. For instance if you unchecked fundamental 3 but

> your mount has the 76 second fundamental then that will be

> present in the residual and it contributes to increasing the

> RMS error.

>

> You wrote:

> > PEC curve should not be even created - message saying

> FORGET IT should

> > be displayed. As Rolando writes in another context

>

> There are so many posibilities that it is almost impossible

> to make a decision on the quality of a particular PEC curve.

> That's why it is recommended that after uploading a PEC curve

> you measure the residual periodic error again and refine if

> needed (BTW, this is something I don't think you did).

>

> If you try to sample over 25 worm cycles and and there are

> all kinds of influences that cannot be modelled well then you

> are not going to get good

> (PEC) results. As I said garbage in, garbage out. You need to

> have clean data.

>

> So, that said, I want a chance to examine your data curves to

> see if there is any way I can improve the modeling of the

> data. So please upload your raw log files to the files

> section (if you haven't deleted the data.

>

> -Ray

>

> [Non-text portions of this message have been removed]

>

>

>

>

>







----------------------------

#39823 Sep 18, 2008

Ladislav,



I'm going to have to write a utility to split the files.



-Ray

> -----Original Message-----

> From: Losmandy_users@yahoogroups.com

> [mailto:Losmandy_users@yahoogroups.com] On Behalf Of Ladislav Nemec

> Sent: Thursday, September 18, 2008 4:17 PM

> To: Losmandy_users@yahoogroups.com

> Subject: RE: [Losmandy_users] Re: PE as a function of number

> of cycles used by PemPro

>

> I did that - the splitting (by editing the original file) is

> bit tricky.

> PemPro first says it cannot load the data - somewhere a total

> number of points is there but I could not find it and edit

> it. Then it loads the 'truncated' set anyway.

>

> As I reported yesterday, the RMS error of the drift fit

> increases, indeed, with the number of cycles (I used 5, 15

> and 25). Even with 5 cycles it is still too high for creation

> of a 'useful' PEC curve - that's I understand Ray's comments.

>

> LN

>

> -----Original Message-----

> From: Losmandy_users@yahoogroups.com

> mailto:Losmandy_users%40yahoogroups.com>

> [mailto:Losmandy_users@yahoogroups.com

> mailto:Losmandy_users%40yahoogroups.com> ] On Behalf Of Paul K

> Sent: Thursday, September 18, 2008 7:18 AM

> To: Losmandy_users@yahoogroups.com

> mailto:Losmandy_users%40yahoogroups.com>

> Subject: [Losmandy_users] Re: PE as a function of number of

> cycles used by PemPro

>

> --- In Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

> yahoogroups.com, "Ray Gralak" rgr@...> wrote:

> > If you try to sample over 25 worm cycles and and there are all kinds

> of

> > influences that cannot be modelled well then you are not going to

> get good

> > (PEC) results.

>

> I'd guess that splitting up Ladilav's 25 cycles into much

> shorter runs, of say 5 cycles each, would help quite a bit by

> making the drift fitting work better.

>

> Regards,

>

> -Paul

>

> [Non-text portions of this message have been removed]

>

>

>

>

>



----------------------------

#39824 Sep 19, 2008

Ray, they ARE the originals. I did NOT 'connect' the program to the mount

during data acquisition - I would have been able to do so only for one copy

of PemPro (on one of two computers connected to two cameras) anyway. Hence

no data about the mount and/or ascom driver(s). As you know, one does NOT

have to connect the mount for data acquisition at the price of not having

complete set of data that, I am pretty sure, are quite irrelevant to the

task at hand.







LN







-----Original Message-----

From: Losmandy_users@yahoogroups.com [mailto:Losmandy_users@yahoogroups.com]

On Behalf Of Ray Gralak

Sent: Thursday, September 18, 2008 7:59 PM

To: Losmandy_users@yahoogroups.com

Subject: RE: [Losmandy_users] PE as a function of number of cycles used by

PemPro







Ladislav,



I looked at your log files and most or all of them appear to NOT be original

PEMPro files. I know you mentioned you were going to do this but there seems

to be missing lots of informnation from the files (like gemini data, ascom

drivers, etc).



Please post the originals and don't try to cut/paste/truncate anything. I'm

going to have to put CRCs on each line (tied to a line number) to make sure

people are not changing these log files. :-)



Thanks,



-Ray

> -----Original Message-----

> From: Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com > [mailto:Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com] On Behalf Of Ladislav Nemec > Sent: Thursday, September 18, 2008 4:13 PM

> To: Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com > Subject: RE: [Losmandy_users] PE as a function of number of

> cycles used by PemPro

>

> I never delete anything, I just buy new disks. Well, I was

> not entirely correct but close, I think. Basically I assumed

> that the RESIDUAL ERROR was everything that could not be

> represented by a Fourier series. It is actually what's left

> after fitting the 'long term' drift curve to an n-th order

> curve and the periodic (based on the worm cycle period)

> Fourier representation.

>

> Thanks for the correction (do I have it right now?). Years

> ago I used to do some least square fitting - completely

> forgot the details, though. Prefer to leave the details to you.

>

> It is interesting to compare the frequency spectra of RA (x)

> and Dec (y) data. No 76 s error is present (as one would

> expect) on the Dec data. The 'weight' of the 76 s error

> depends on the number of cycles used - I will post several

> large sets of the logs in the file section, zipped they

> should not take that much space. BTW, on virtually all sets

> of data I looked at, the 76 s error has a 'value' of some 1.6

> arcsecs associated with it - is it a typical low or high

> value? Seems low to me but PemPro data are filled with subtleties.

>

> It should not take me too long to organize several sets of

> recent data and add some reasonable descriptions and brief

> comments. Definitely today.

>

> Thanks for your patience.

>

> LN

>

> -----Original Message-----

> From: Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com > mailto:Losmandy_users%40yahoogroups.com>

> [mailto:Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com > mailto:Losmandy_users%40yahoogroups.com> ] On Behalf Of Ray Gralak

> Sent: Thursday, September 18, 2008 6:55 AM

> To: Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com > mailto:Losmandy_users%40yahoogroups.com>

> Subject: RE: [Losmandy_users] PE as a function of number of

> cycles used by PemPro

>

> Ladislav,

>

> You wrote:

> > Optics moving - definitely. Cables tugging, over roof seeing

> > - absolutely not. I am quite sorry that, at this late date I do not

> > seem to understand the basics. No reason to blame anything

> and anyone

> > but myself. PemPro says what it means:

> > RMS for drift fitting and drift, apparently, means ANYTHING that is

> > not periodic with the worm. PE +/- values mean PERIODIC ERROR

> > variation.

>

> Well, that's not completely correct...

>

> First, PEMPro best-fits the data based on finding the

> least-squares fit to a multi-frequency sine wave and drift

> curve. If you want to know something about leasts square do a

> google search or look at this link:

> en.wikipedia

> en.wikipedia en.wikipedia.org/wiki/Least_squares>

.org/wiki/Least_squares > en.wikipedia en.wikipedia.org/wiki/Least_squares>

.org/wiki/Least_squares> > .org/wiki/Least_squares >

> After obtaining the best fit periodic and drift curves they

> are subtracted from the data points. The remainder is called

> the RESIDUAL error.

>

> KEY POINT--> The RMS value is calculated soley from the

> residual error.

>

> The residual error could contain the sum of atmospheric

> seeing, unmodelled periodic terms, unmodeled drift, mirror

> movements, worm wheel tooth defects, etc.

>

> The residual data may or may not have periodic components

> left in it. For instance if you unchecked fundamental 3 but

> your mount has the 76 second fundamental then that will be

> present in the residual and it contributes to increasing the

> RMS error.

>

> You wrote:

> > PEC curve should not be even created - message saying

> FORGET IT should

> > be displayed. As Rolando writes in another context

>

> There are so many posibilities that it is almost impossible

> to make a decision on the quality of a particular PEC curve.

> That's why it is recommended that after uploading a PEC curve

> you measure the residual periodic error again and refine if

> needed (BTW, this is something I don't think you did).

>

> If you try to sample over 25 worm cycles and and there are

> all kinds of influences that cannot be modelled well then you

> are not going to get good

> (PEC) results. As I said garbage in, garbage out. You need to

> have clean data.

>

> So, that said, I want a chance to examine your data curves to

> see if there is any way I can improve the modeling of the

> data. So please upload your raw log files to the files

> section (if you haven't deleted the data.

>

> -Ray

>

> [Non-text portions of this message have been removed]

>

>

>

>

>











[Non-text portions of this message have been removed]







----------------------------

#39825 Sep 19, 2008

That may be useful. First I thought I can just edit some kind of 'index' but

it is, as usual, bit more complicated. For 'research' purposes it may be

desirable to use subsets of the acquired data - the number of cycles used

has some effect but I cannot really say what effect, exactly.







Frank's usage of 10 cycles is, in my opinion, a good selection - 40 minutes

spent with PemPro is not that long and a lot of things can get averaged.

Recording more than 20 cycles is probably excessive - I have no plans to do

it when the Moon diminishes.







My secondary goal was to establish the difference between tracking using the

main OTA and the guidescope. So far I have no answer to that question.

Flying next week to Florida I can probably amuse myself with more detailed

analysis of the data on that 5 hour flight, battery allowing.







Regards, LN







-----Original Message-----

From: Losmandy_users@yahoogroups.com [mailto:Losmandy_users@yahoogroups.com]

On Behalf Of Ray Gralak

Sent: Thursday, September 18, 2008 8:13 PM

To: Losmandy_users@yahoogroups.com

Subject: RE: [Losmandy_users] Re: PE as a function of number of cycles used

by PemPro







Ladislav,



I'm going to have to write a utility to split the files.



-Ray

> -----Original Message-----

> From: Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com > [mailto:Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com] On Behalf Of Ladislav Nemec > Sent: Thursday, September 18, 2008 4:17 PM

> To: Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com > Subject: RE: [Losmandy_users] Re: PE as a function of number

> of cycles used by PemPro

>

> I did that - the splitting (by editing the original file) is

> bit tricky.

> PemPro first says it cannot load the data - somewhere a total

> number of points is there but I could not find it and edit

> it. Then it loads the 'truncated' set anyway.

>

> As I reported yesterday, the RMS error of the drift fit

> increases, indeed, with the number of cycles (I used 5, 15

> and 25). Even with 5 cycles it is still too high for creation

> of a 'useful' PEC curve - that's I understand Ray's comments.

>

> LN

>

> -----Original Message-----

> From: Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com > mailto:Losmandy_users%40yahoogroups.com>

> [mailto:Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com > mailto:Losmandy_users%40yahoogroups.com> ] On Behalf Of Paul K

> Sent: Thursday, September 18, 2008 7:18 AM

> To: Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com > mailto:Losmandy_users%40yahoogroups.com>

> Subject: [Losmandy_users] Re: PE as a function of number of

> cycles used by PemPro

>

> --- In Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

> yahoogroups.com, "Ray Gralak" rgr@...> wrote:

> > If you try to sample over 25 worm cycles and and there are all kinds

> of

> > influences that cannot be modelled well then you are not going to

> get good

> > (PEC) results.

>

> I'd guess that splitting up Ladilav's 25 cycles into much

> shorter runs, of say 5 cycles each, would help quite a bit by

> making the drift fitting work better.

>

> Regards,

>

> -Paul

>

> [Non-text portions of this message have been removed]

>

>

>

>

>











[Non-text portions of this message have been removed]



----------------------------

#39828 Sep 19, 2008

Ladislav,



You wrote: > Ray, they ARE the originals. I did NOT 'connect' the program

> to the mount during data acquisition - I would have been able



OK, sorry! However, the curves and frequency fundamentals in these logs

files do not match the earlier log files that you sent me. Your earlier logs

had fundamentals at 1x and 2x and your new logs do not. Since you did not

connect it's not possible to tell the PEC settings from the log file. So is

it possible you had PEC on by mistake?



It would be very helpful if you followed the normal procedure presented in

the help file just so that it matches what everyone else is doing. Note that

when you connect to the mount PEC phase information is then saved and the

PEC data is useful later. If you do not connect PEC phase is unknown.

> to do so only for one copy of PemPro (on one of two computers

> connected to two cameras) anyway. Hence no data about the



If you need to acquire data from one PC and analyze with another copy of

PEMPro on another PC that is fine.



You wrote (yesterday): > Just for the fun I did PEC with a curve uploaded from Gemini

> and the only result was that very small RA drift without PEC

> turned into major RA drift with PEC. Did not try to play with

> the linear drift adjustment.



If you did this then you had to connect to the mount! :-)



-Ray







----------------------------

#39843 Sep 19, 2008

Please disregard the previous (a few weeks old) logs. PEC was definitely off

and I can, of course, connect one of the computers to the mount. As you

noticed, I had to do it for that brief experiment with actual PEC curve

uploaded and PEC ON. I had no particular reason no to connect the mount on

the computer Gemini is connected all the time but used by TheSky6 for

pointing and adjustment of the position for both cameras. Most of the

longish runs were terminated when the star got 'too close to the edge'.







I have 3 - 4 nights before going to Florida and the clouds usually disappear

at night here, the site of one of few dedicated professional solar

observatories.







It is (a very minor) inconvenience to disconnecting and reconnecting the

ASCOM drivers when switching from TheSky, PemPro and (in another context)

PHD. I believe there is some trick to avoid it but I better do it patiently

and consistently.







Thanks for first look at the data - the main purpose of the exercise was to

see what is REALLY going on with my scope (shape of the long term drift) and

to compare two scopes - one to be used for guiding. The issues of PE and

using PEC on the mount are, at this time, secondary.







From the shape of the drift I made a preliminary conclusion that my entire

setup is kind of flexing or sagging. That's kind of fundamental and,

hopefully, I will be able to do something about it. On the other hand, the

11" mirror shifting does not seem to be minor but much more detailed

analysis (exporting data to Excel) is needed. It is conceivable that the

mirror shift (of the C11) is worse in some physical positions than in other.

Having quite unobstructed view and knowing the bad positions I may be able

to avoid them during real imaging.







PemPro has been an invaluable tool for collecting and quickly displaying the

data. There are, of course, other ways of doing it (disabling, for instance,

guiding in PHD and using the logs yielding essentially the same information

as PemPro logs) but none of them so convenient with abundance of analytical

data.







As for the absence of fundamentals, that seems to be one of the issues

dependent on the number of cycles. In my truncated files (I did not upload

them - easy enough to create from the large files) these fundamentals were

clearly present when only 5 cycles were used. For really long runs, as you

probably observed, the contribution of sub-prime (so to speak) components at

0.1, 0.2 multiple of the worm frequency prevails - I have no idea whether

this is real or just kind of 'artifact' of the Fourier representation.

Certainly not difficult to eliminate with guiding.







Best,



LN







The next results will be, hopefully, with the whole thing sitting more

firmly directly on top of my water tank. Bit unusual arrangement but most

convenient (I am above most of the trees that grow rather short on this side

of Big Bear Valley) to achieve an unobstructed view.







LN







-----Original Message-----

From: Losmandy_users@yahoogroups.com [mailto:Losmandy_users@yahoogroups.com]

On Behalf Of Ray Gralak

Sent: Friday, September 19, 2008 6:42 AM

To: Losmandy_users@yahoogroups.com

Subject: RE: [Losmandy_users] PE as a function of number of cycles used by

PemPro







Ladislav,



You wrote: > Ray, they ARE the originals. I did NOT 'connect' the program

> to the mount during data acquisition - I would have been able



OK, sorry! However, the curves and frequency fundamentals in these logs

files do not match the earlier log files that you sent me. Your earlier logs

had fundamentals at 1x and 2x and your new logs do not. Since you did not

connect it's not possible to tell the PEC settings from the log file. So is

it possible you had PEC on by mistake?



It would be very helpful if you followed the normal procedure presented in

the help file just so that it matches what everyone else is doing. Note that

when you connect to the mount PEC phase information is then saved and the

PEC data is useful later. If you do not connect PEC phase is unknown.

> to do so only for one copy of PemPro (on one of two computers

> connected to two cameras) anyway. Hence no data about the



If you need to acquire data from one PC and analyze with another copy of

PEMPro on another PC that is fine.



You wrote (yesterday): > Just for the fun I did PEC with a curve uploaded from Gemini

> and the only result was that very small RA drift without PEC

> turned into major RA drift with PEC. Did not try to play with

> the linear drift adjustment.



If you did this then you had to connect to the mount! :-)



-Ray











[Non-text portions of this message have been removed]







----------------------------

#39846 Sep 19, 2008

Ladislav,



After looking at all your log files I think that the drift is actual well

corrected with either 4th and 5th order polynomial drift fitting in your

logs. The high RMS seems to be coming from random errors from seeing or from

worm tooth anomalies.



Also, as I said previously there is no sign of fundamentals at 1x and 2x,

like in numerous instances of your older log files. However the infamous 76

second fundamental is present so my conclusion is that the drift fitting

works well with your data and you had PEC unknowingly enabled.



-Ray

> -----Original Message-----

> From: Losmandy_users@yahoogroups.com

> [mailto:Losmandy_users@yahoogroups.com] On Behalf Of Ladislav Nemec

> Sent: Friday, September 19, 2008 1:52 PM

> To: Losmandy_users@yahoogroups.com

> Subject: RE: [Losmandy_users] PE as a function of number of

> cycles used by PemPro

>

> Please disregard the previous (a few weeks old) logs. PEC was

> definitely off and I can, of course, connect one of the

> computers to the mount. As you noticed, I had to do it for

> that brief experiment with actual PEC curve uploaded and PEC

> ON. I had no particular reason no to connect the mount on the

> computer Gemini is connected all the time but used by TheSky6

> for pointing and adjustment of the position for both cameras.

> Most of the longish runs were terminated when the star got

> 'too close to the edge'.

>

> I have 3 - 4 nights before going to Florida and the clouds

> usually disappear at night here, the site of one of few

> dedicated professional solar observatories.

>

> It is (a very minor) inconvenience to disconnecting and

> reconnecting the ASCOM drivers when switching from TheSky,

> PemPro and (in another context) PHD. I believe there is some

> trick to avoid it but I better do it patiently and consistently.

>

> Thanks for first look at the data - the main purpose of the

> exercise was to see what is REALLY going on with my scope

> (shape of the long term drift) and to compare two scopes -

> one to be used for guiding. The issues of PE and using PEC on

> the mount are, at this time, secondary.

>

> From the shape of the drift I made a preliminary conclusion

> that my entire setup is kind of flexing or sagging. That's

> kind of fundamental and, hopefully, I will be able to do

> something about it. On the other hand, the 11" mirror

> shifting does not seem to be minor but much more detailed

> analysis (exporting data to Excel) is needed. It is

> conceivable that the mirror shift (of the C11) is worse in

> some physical positions than in other.

> Having quite unobstructed view and knowing the bad positions

> I may be able to avoid them during real imaging.

>

> PemPro has been an invaluable tool for collecting and quickly

> displaying the data. There are, of course, other ways of

> doing it (disabling, for instance, guiding in PHD and using

> the logs yielding essentially the same information as PemPro

> logs) but none of them so convenient with abundance of

> analytical data.

>

> As for the absence of fundamentals, that seems to be one of

> the issues dependent on the number of cycles. In my truncated

> files (I did not upload them - easy enough to create from the

> large files) these fundamentals were clearly present when

> only 5 cycles were used. For really long runs, as you

> probably observed, the contribution of sub-prime (so to

> speak) components at 0.1, 0.2 multiple of the worm frequency

> prevails - I have no idea whether this is real or just kind

> of 'artifact' of the Fourier representation.

> Certainly not difficult to eliminate with guiding.

>

> Best,

>

> LN

>

> The next results will be, hopefully, with the whole thing

> sitting more firmly directly on top of my water tank. Bit

> unusual arrangement but most convenient (I am above most of

> the trees that grow rather short on this side of Big Bear

> Valley) to achieve an unobstructed view.

>

> LN

>

> -----Original Message-----

> From: Losmandy_users@yahoogroups.com

> mailto:Losmandy_users%40yahoogroups.com>

> [mailto:Losmandy_users@yahoogroups.com

> mailto:Losmandy_users%40yahoogroups.com> ] On Behalf Of Ray Gralak

> Sent: Friday, September 19, 2008 6:42 AM

> To: Losmandy_users@yahoogroups.com

> mailto:Losmandy_users%40yahoogroups.com>

> Subject: RE: [Losmandy_users] PE as a function of number of

> cycles used by PemPro

>

> Ladislav,

>

> You wrote:

> > Ray, they ARE the originals. I did NOT 'connect' the program to the

> > mount during data acquisition - I would have been able

>

> OK, sorry! However, the curves and frequency fundamentals in

> these logs files do not match the earlier log files that you

> sent me. Your earlier logs had fundamentals at 1x and 2x and

> your new logs do not. Since you did not connect it's not

> possible to tell the PEC settings from the log file. So is it

> possible you had PEC on by mistake?

>

> It would be very helpful if you followed the normal procedure

> presented in the help file just so that it matches what

> everyone else is doing. Note that when you connect to the

> mount PEC phase information is then saved and the PEC data is

> useful later. If you do not connect PEC phase is unknown.

>

> > to do so only for one copy of PemPro (on one of two computers

> > connected to two cameras) anyway. Hence no data about the

>

> If you need to acquire data from one PC and analyze with

> another copy of PEMPro on another PC that is fine.

>

> You wrote (yesterday):

> > Just for the fun I did PEC with a curve uploaded from

> Gemini and the

> > only result was that very small RA drift without PEC turned

> into major

> > RA drift with PEC. Did not try to play with the linear drift

> > adjustment.

>

> If you did this then you had to connect to the mount! :-)

>

> -Ray

>

> [Non-text portions of this message have been removed]

>

>

>

>

>







----------------------------

#39847 Sep 19, 2008

Thanks again for such a close attention to my data. I am just going up to my

'observatory', the tripod standing now directly on somewhat bouncy plastic

water tank. Obviously, I am blazing a brand new trail of 'observatories'

here in the West sitting on the top of water tanks.







If I can align and operate my scope under the NEW and (possibly) IMPROVED

circumstances, I'll see what I can accomplish.







Isn't it exciting? Sorry, I just finished one diluted Ballantine shot.







Driving under influence is a crime. Posting here may be inappropriate but,

not yet, a crime.





Apologizes to more stable members.







LN







-----Original Message-----

From: Losmandy_users@yahoogroups.com [mailto:Losmandy_users@yahoogroups.com]

On Behalf Of Ray Gralak

Sent: Friday, September 19, 2008 6:33 PM

To: Losmandy_users@yahoogroups.com

Subject: RE: [Losmandy_users] PE as a function of number of cycles used by

PemPro







Ladislav,



After looking at all your log files I think that the drift is actual well

corrected with either 4th and 5th order polynomial drift fitting in your

logs. The high RMS seems to be coming from random errors from seeing or from

worm tooth anomalies.



Also, as I said previously there is no sign of fundamentals at 1x and 2x,

like in numerous instances of your older log files. However the infamous 76

second fundamental is present so my conclusion is that the drift fitting

works well with your data and you had PEC unknowingly enabled.



-Ray

> -----Original Message-----

> From: Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com > [mailto:Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com] On Behalf Of Ladislav Nemec > Sent: Friday, September 19, 2008 1:52 PM

> To: Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com > Subject: RE: [Losmandy_users] PE as a function of number of

> cycles used by PemPro

>

> Please disregard the previous (a few weeks old) logs. PEC was

> definitely off and I can, of course, connect one of the

> computers to the mount. As you noticed, I had to do it for

> that brief experiment with actual PEC curve uploaded and PEC

> ON. I had no particular reason no to connect the mount on the

> computer Gemini is connected all the time but used by TheSky6

> for pointing and adjustment of the position for both cameras.

> Most of the longish runs were terminated when the star got

> 'too close to the edge'.

>

> I have 3 - 4 nights before going to Florida and the clouds

> usually disappear at night here, the site of one of few

> dedicated professional solar observatories.

>

> It is (a very minor) inconvenience to disconnecting and

> reconnecting the ASCOM drivers when switching from TheSky,

> PemPro and (in another context) PHD. I believe there is some

> trick to avoid it but I better do it patiently and consistently.

>

> Thanks for first look at the data - the main purpose of the

> exercise was to see what is REALLY going on with my scope

> (shape of the long term drift) and to compare two scopes -

> one to be used for guiding. The issues of PE and using PEC on

> the mount are, at this time, secondary.

>

> From the shape of the drift I made a preliminary conclusion

> that my entire setup is kind of flexing or sagging. That's

> kind of fundamental and, hopefully, I will be able to do

> something about it. On the other hand, the 11" mirror

> shifting does not seem to be minor but much more detailed

> analysis (exporting data to Excel) is needed. It is

> conceivable that the mirror shift (of the C11) is worse in

> some physical positions than in other.

> Having quite unobstructed view and knowing the bad positions

> I may be able to avoid them during real imaging.

>

> PemPro has been an invaluable tool for collecting and quickly

> displaying the data. There are, of course, other ways of

> doing it (disabling, for instance, guiding in PHD and using

> the logs yielding essentially the same information as PemPro

> logs) but none of them so convenient with abundance of

> analytical data.

>

> As for the absence of fundamentals, that seems to be one of

> the issues dependent on the number of cycles. In my truncated

> files (I did not upload them - easy enough to create from the

> large files) these fundamentals were clearly present when

> only 5 cycles were used. For really long runs, as you

> probably observed, the contribution of sub-prime (so to

> speak) components at 0.1, 0.2 multiple of the worm frequency

> prevails - I have no idea whether this is real or just kind

> of 'artifact' of the Fourier representation.

> Certainly not difficult to eliminate with guiding.

>

> Best,

>

> LN

>

> The next results will be, hopefully, with the whole thing

> sitting more firmly directly on top of my water tank. Bit

> unusual arrangement but most convenient (I am above most of

> the trees that grow rather short on this side of Big Bear

> Valley) to achieve an unobstructed view.

>

> LN

>

> -----Original Message-----

> From: Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com > mailto:Losmandy_users%40yahoogroups.com>

> [mailto:Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com > mailto:Losmandy_users%40yahoogroups.com> ] On Behalf Of Ray Gralak

> Sent: Friday, September 19, 2008 6:42 AM

> To: Losmandy_users@ mailto:Losmandy_users%40yahoogroups.com>

yahoogroups.com > mailto:Losmandy_users%40yahoogroups.com>

> Subject: RE: [Losmandy_users] PE as a function of number of

> cycles used by PemPro

>

> Ladislav,

>

> You wrote:

> > Ray, they ARE the originals. I did NOT 'connect' the program to the

> > mount during data acquisition - I would have been able

>

> OK, sorry! However, the curves and frequency fundamentals in

> these logs files do not match the earlier log files that you

> sent me. Your earlier logs had fundamentals at 1x and 2x and

> your new logs do not. Since you did not connect it's not

> possible to tell the PEC settings from the log file. So is it

> possible you had PEC on by mistake?

>

> It would be very helpful if you followed the normal procedure

> presented in the help file just so that it matches what

> everyone else is doing. Note that when you connect to the

> mount PEC phase information is then saved and the PEC data is

> useful later. If you do not connect PEC phase is unknown.

>

> > to do so only for one copy of PemPro (on one of two computers

> > connected to two cameras) anyway. Hence no data about the

>

> If you need to acquire data from one PC and analyze with

> another copy of PEMPro on another PC that is fine.

>

> You wrote (yesterday):

> > Just for the fun I did PEC with a curve uploaded from

> Gemini and the

> > only result was that very small RA drift without PEC turned

> into major

> > RA drift with PEC. Did not try to play with the linear drift

> > adjustment.

>

> If you did this then you had to connect to the mount! :-)

>

> -Ray

>

> [Non-text portions of this message have been removed]

>

>

>

>

>











[Non-text portions of this message have been removed]







Contact Us
This Site's Privacy Policy
Google's privacy policies

S
e
n
i
o
r
T
u
b
e
.
o
r
g