VintageBigBlue.org

 

Re: OVISION TEST, G11: Take 2


May 25, 2009

 


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

#42966 May 25, 2009

As per Bill's suggestions, I made sure my image train was tight, and ran another Ovision test. I did let PhD run for 12 minutes, which gave me three worm cycles. I ran the data through PECPrep. I input 600mm for the scope's focal length. The pixels of my Stellacam3 are 8.4um x 9.8um. PECPrep asked for pixel size, not dimension, so I input "9", given the pixel's dimensions.

The report indicated that my PE was a little over 10" (-5 to +5). Bill et.al, would you mind checking my work? I put the PHD log in my folder called "Ovision Test: Bob Hertel". Seems like my periodic error is large compared to others's reports. I would be interested in a critique of my data collection and, if I did it correctly, worm performance.

Thank you.

Bob Hertel



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

#42968 May 25, 2009

Hi, Bob -



I uploaded the PECPrep graph for your data - it's in the same folder as a jpg (000157.jpg). And, you know, I don't think it's all that bad. For one thing, PECPrep is reporting +/- 5, but looking at the graph, those are the extremes. I've seen this before and maybe PECPrep is not working the RA drift out of it altogether correctly. I think it's closer to +/- 4, which isn't too shabby. By the way, notice the Max Delta is something like +2/-2.51. This is in my opinion mainly due to seeing, with some, of course, attributable to roughness. The only other thing that could contribute to the "waviness", if you will, is if your gear mesh is a bit too loose.



I think this will get better over time as the gears get run in. My PE is around +/-3.3, but it didn't start out being that good.



I also think you're on the right track and that what you have is certainly guidable. Try doing some imaging with what you have and see how it turns out. If your primary image scale is something like 2 arc-seconds, you may not have any problem.



Regards,



Bill













-- In Losmandy_users@yahoogroups.com, "rjhertel2001" robert.hertel@...> wrote: >

> As per Bill's suggestions, I made sure my image train was tight, and ran another Ovision test. I did let PhD run for 12 minutes, which gave me three worm cycles. I ran the data through PECPrep. I input 600mm for the scope's focal length. The pixels of my Stellacam3 are 8.4um x 9.8um. PECPrep asked for pixel size, not dimension, so I input "9", given the pixel's dimensions.

> The report indicated that my PE was a little over 10" (-5 to +5). Bill et.al, would you mind checking my work? I put the PHD log in my folder called "Ovision Test: Bob Hertel". Seems like my periodic error is large compared to others's reports. I would be interested in a critique of my data collection and, if I did it correctly, worm performance.

> Thank you.

> Bob Hertel

>



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

#42986 May 26, 2009

Hi Bob,



The PHD log records pixel movements and to convert these accurately to arcsecs PECPrep needs to know not only the imaging resolution (determined form focal length and pixel size) but also the declination of the star being used (unless the star happens to be at declination 0). There should be a checkbox and entry field to allow you to do this (same place you set pixel size and resolution etc.)



With regard to Pixel size, I can see from the calibration data in your PHD log that your camera was mounted orthogonally to the mount i.e. X axis of the sensor is aligned with the RA axis. In this case the pixel size you enter is pixel width so you can use a value of 8.4



The PE trace you see is a combination of the error signals generated by the mechanical components in the RA axis drive and non periodic signals such as fluctuations caused by seeing, wind etc. To extract the mechanical performance from the background "noise" you need to apply PECPreps filters to the raw data. The "Autofilter" should do a reasonable job in isolating the main mechanical error but you can further tweak the individual sliders to ensure that the 'major peaks' remain unfiltered.



Here's what I get if I apply a simple lowpass filter (to take out high frequency data) and highpass filter (to remove frequencies lower then the worm) to your data (and using a pixels size of 8.4 um)



tech.groups.yahoo.com/group/Losmandy_users/files/OVISION%20%20TEST%20-Bob%20Hertel/pecprep_filtered.png



This screenshot still assumes declination=0 which is best case as far PE magnitude is concerned. The further away from 0 degrees you actually were the greater the scaling that needs to be applied to the PE.



Chris.

--- In Losmandy_users@yahoogroups.com, "rjhertel2001" robert.hertel@...> wrote:

>

> As per Bill's suggestions, I made sure my image train was tight, and ran another Ovision test. I did let PhD run for 12 minutes, which gave me three worm cycles. I ran the data through PECPrep. I input 600mm for the scope's focal length. The pixels of my Stellacam3 are 8.4um x 9.8um. PECPrep asked for pixel size, not dimension, so I input "9", given the pixel's dimensions.

> The report indicated that my PE was a little over 10" (-5 to +5). Bill et.al, would you mind checking my work? I put the PHD log in my folder called "Ovision Test: Bob Hertel". Seems like my periodic error is large compared to others's reports. I would be interested in a critique of my data collection and, if I did it correctly, worm performance.

> Thank you.

> Bob Hertel

>



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

#42991 May 26, 2009

My thanks to Chris for filling in the input/processing gaps in my PECPrep. I can see that the peak to peak PE is much better after filtering, but that is where my "interpretive ability" stops. I do know that a "smooth curve" is a good thing with respect to autoguiding, but I really don't know how to define that either quantitatively or qualitatively. I'm taking primarily NB wide field images from my crummy location, but what do people look at, e.g., in the PECPrep output Chris posted in my folder (Ovision test-Bob Hertel, filtered) when they see a curve like the one Chris produced for me, to decide if, in this case, the worm is doing a good job?

Thanks for any interpretive advice/examples.

Regards,

Bob Hertel

--- In Losmandy_users@yahoogroups.com, "Chris Shillito" chris@...> wrote:

>

> Hi Bob,

>

> The PHD log records pixel movements and to convert these accurately to arcsecs PECPrep needs to know not only the imaging resolution (determined form focal length and pixel size) but also the declination of the star being used (unless the star happens to be at declination 0). There should be a checkbox and entry field to allow you to do this (same place you set pixel size and resolution etc.)

>

> With regard to Pixel size, I can see from the calibration data in your PHD log that your camera was mounted orthogonally to the mount i.e. X axis of the sensor is aligned with the RA axis. In this case the pixel size you enter is pixel width so you can use a value of 8.4

>

> The PE trace you see is a combination of the error signals generated by the mechanical components in the RA axis drive and non periodic signals such as fluctuations caused by seeing, wind etc. To extract the mechanical performance from the background "noise" you need to apply PECPreps filters to the raw data. The "Autofilter" should do a reasonable job in isolating the main mechanical error but you can further tweak the individual sliders to ensure that the 'major peaks' remain unfiltered.

>

> Here's what I get if I apply a simple lowpass filter (to take out high frequency data) and highpass filter (to remove frequencies lower then the worm) to your data (and using a pixels size of 8.4 um)

>

> tech.groups.yahoo.com/group/Losmandy_users/files/OVISION%20%20TEST%20-Bob%20Hertel/pecprep_filtered.png

>

> This screenshot still assumes declination=0 which is best case as far PE magnitude is concerned. The further away from 0 degrees you actually were the greater the scaling that needs to be applied to the PE.

>

> Chris.

>

> --- In Losmandy_users@yahoogroups.com, "rjhertel2001" robert.hertel@> wrote:

> >

> > As per Bill's suggestions, I made sure my image train was tight, and ran another Ovision test. I did let PhD run for 12 minutes, which gave me three worm cycles. I ran the data through PECPrep. I input 600mm for the scope's focal length. The pixels of my Stellacam3 are 8.4um x 9.8um. PECPrep asked for pixel size, not dimension, so I input "9", given the pixel's dimensions.

> > The report indicated that my PE was a little over 10" (-5 to +5). Bill et.al, would you mind checking my work? I put the PHD log in my folder called "Ovision Test: Bob Hertel". Seems like my periodic error is large compared to others's reports. I would be interested in a critique of my data collection and, if I did it correctly, worm performance.

> > Thank you.

> > Bob Hertel

> >

>







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

#42992 May 27, 2009

Hi Bob,



The curve looks reasonably good to me. The overall amplitude of the error is not large, and I don't see any large sudden changes which would be difficult to guide out.



As long as you are guiding, a long exposure wide-field shot should not be a problem with this worm.



Regards,



-Paul

--- In Losmandy_users@yahoogroups.com, "rjhertel2001" robert.hertel@...> wrote:

>

> My thanks to Chris for filling in the input/processing gaps in my PECPrep. I can see that the peak to peak PE is much better after filtering, but that is where my "interpretive ability" stops. I do know that a "smooth curve" is a good thing with respect to autoguiding, but I really don't know how to define that either quantitatively or qualitatively. I'm taking primarily NB wide field images from my crummy location, but what do people look at, e.g., in the PECPrep output Chris posted in my folder (Ovision test-Bob Hertel, filtered) when they see a curve like the one Chris produced for me, to decide if, in this case, the worm is doing a good job?

> Thanks for any interpretive advice/examples.

> Regards,

> Bob Hertel

>



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

#42993 May 27, 2009

Hi Bob,



The PE trace you see is the sum of all periodic errors induced by the various mechanical components in your RA gear train. Each rotating component contributes a sinusoidal error signal of differing magnitude and possibly phase. The worm itself is just one component in the gear train but is the one that contributes the greatest amount of error. The point I'm trying to make here is that the PE curve you see in PECPrep does not simply reflect your worms performance.



The stacked cycle display in PECPrep shows how repeatable the error signal is over successive worm cylces. Obviously repeatability is a good thing (particularity is intending to use PEC) but a lack of repeatability is not necessarily down to the worm itself or its mountings. It could be due to the contribution of some other mechanical component that has a rotation frequency that is not harmonic with the worm and in consequence "slips in phase" with respect to the worm.



Now I should say here I have no experience of the levels of performance expected from an OVISION worm (or indeed G11) but as a general observation I would say your PE curve looks ok. It is roughly sinusoidal and appears quite repeatable (the more cycles of data you can feed into PECPrep the better). As well as the time domain PE graph PECPrep's frequency spectrum analysis can also be useful in diagnosing/interpreting performance issues. For instance if there were any restriction of movement of the worm then we would expect to see the frequency spectrum dominated by odd harmonics of the worm (for your data this doesn't appear to be the case).



For autoguiding you are correct that a 'smooth' curve is ideal - repeatability or Peak to Peak error isn't particularly an issue in that case. Of course much will depend on the autoguider and its settings. Generally sudden changes are bad as the autoguider needs time to play catchup and whilst its doing so you're image suffers. Make the autoguidng too aggressive to compensate and you may end up chasing seeing etc.



Chris.



--- In Losmandy_users@yahoogroups.com, "rjhertel2001" robert.hertel@...> wrote:

>

> My thanks to Chris for filling in the input/processing gaps in my PECPrep. I can see that the peak to peak PE is much better after filtering, but that is where my "interpretive ability" stops. I do know that a "smooth curve" is a good thing with respect to autoguiding, but I really don't know how to define that either quantitatively or qualitatively. I'm taking primarily NB wide field images from my crummy location, but what do people look at, e.g., in the PECPrep output Chris posted in my folder (Ovision test-Bob Hertel, filtered) when they see a curve like the one Chris produced for me, to decide if, in this case, the worm is doing a good job?

> Thanks for any interpretive advice/examples.

> Regards,

> Bob Hertel

>



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

#42996 May 27, 2009

Great information, Chris. Thanks. I have a much better understanding now of how the pieces of PE measurement and interpretation interrelate.

Regards,

Bob

--- In Losmandy_users@yahoogroups.com, "Chris Shillito" chris@...> wrote:

>

> Hi Bob,

>

> The PE trace you see is the sum of all periodic errors induced by the various mechanical components in your RA gear train. Each rotating component contributes a sinusoidal error signal of differing magnitude and possibly phase. The worm itself is just one component in the gear train but is the one that contributes the greatest amount of error. The point I'm trying to make here is that the PE curve you see in PECPrep does not simply reflect your worms performance.

>

> The stacked cycle display in PECPrep shows how repeatable the error signal is over successive worm cylces. Obviously repeatability is a good thing (particularity is intending to use PEC) but a lack of repeatability is not necessarily down to the worm itself or its mountings. It could be due to the contribution of some other mechanical component that has a rotation frequency that is not harmonic with the worm and in consequence "slips in phase" with respect to the worm.

>

> Now I should say here I have no experience of the levels of performance expected from an OVISION worm (or indeed G11) but as a general observation I would say your PE curve looks ok. It is roughly sinusoidal and appears quite repeatable (the more cycles of data you can feed into PECPrep the better). As well as the time domain PE graph PECPrep's frequency spectrum analysis can also be useful in diagnosing/interpreting performance issues. For instance if there were any restriction of movement of the worm then we would expect to see the frequency spectrum dominated by odd harmonics of the worm (for your data this doesn't appear to be the case).

>

> For autoguiding you are correct that a 'smooth' curve is ideal - repeatability or Peak to Peak error isn't particularly an issue in that case. Of course much will depend on the autoguider and its settings. Generally sudden changes are bad as the autoguider needs time to play catchup and whilst its doing so you're image suffers. Make the autoguidng too aggressive to compensate and you may end up chasing seeing etc.

>

> Chris.

>

>

> --- In Losmandy_users@yahoogroups.com, "rjhertel2001" robert.hertel@> wrote:

> >

> > My thanks to Chris for filling in the input/processing gaps in my PECPrep. I can see that the peak to peak PE is much better after filtering, but that is where my "interpretive ability" stops. I do know that a "smooth curve" is a good thing with respect to autoguiding, but I really don't know how to define that either quantitatively or qualitatively. I'm taking primarily NB wide field images from my crummy location, but what do people look at, e.g., in the PECPrep output Chris posted in my folder (Ovision test-Bob Hertel, filtered) when they see a curve like the one Chris produced for me, to decide if, in this case, the worm is doing a good job?

> > Thanks for any interpretive advice/examples.

> > Regards,

> > Bob Hertel

> >

>







Contact Us
This Site's Privacy Policy
Google's privacy policies

S
e
n
i
o
r
T
u
b
e
.
o
r
g