Strange anomaly

[GeorgeCarey]

Can anyone explain this? From time to time I get an effect on some stars that looks as if the mount has suddenly moved in DEC and come back again. Autoslew does not report any errors.

The odd thing is that only some stars show the effect.

In this example two stars show 'trailing' but brighter stars in the image do not.

How can only two stars show this?

 

[GeorgeCarey]

Even stranger- the next frame shows nothing where the two stars were.

 

[GeorgeCarey]

I dug out a frame from May5th that had this effect. To my surprise it has the same two trails in almost the same place.

 

[GeorgeCarey]

If I align the stars it can be seen that the anomaly is not in exactly the same place.

 

[GeorgeCarey]

This is with both images aligned on the edges of the frames - this rules out strange camera effects because the streaks are not in the exact same place.

 

[Bernd_Eppinger]

Hi, George,

 

to be honest - no idea, because your posts 2...5 rule out everything that came into my mind when reading post 1. Everything that remains is approximately as unlikely as this:

 

You have discovered two nobel-price-worthy new stellar-like objects that emit infra-red light-flashes in irregular intervals and that are close enough to Earth to change their observed position over time. The infra-red light causes an RBI (Residual Bulk Image) effect in your CCD. If the chip is not cooled and the read-out speed is slow, a detectable amount of electrons that are released from the traps leak into the shift-register, while the image is read out. This could create weak trails that are exactly vertical. If you then slightly rotate the image to bring North up, you get the "almost" vertical trails of your images.

And forget this explanation quickly; I know it is nonsense.

 

Other explanations could use cosmic particles, sattelites or shooting stars, but it would be unlikely to find two of them in the same image and several of them in several images at almost exactly the same position.

 

Explanations using bad shift register columns or very bad pixels in your CCD are also unlikely, because you have two in one image, mysteriously nothing in the next image, and another event in a different column but magically associated with almost the same (real or pretended) object.

 

 

But can you provide more detail on the exposures?

What was the exposure time?

I assume the CCD chip was cooled down?

Did you post-process the images; especially, did you slightly rotate them e.g. to show north up? If so, are the trails exactly vertical in the original images?

What was the read-out speed (1sec or 10sec resp. 10MHz or 800kHz?)

Did you use a filter?

Did you look up the position in your planetary software?

 

 

Best regards,

 

Bernd

[GeorgeCarey]

Hi Bernd,

The camera is a QSI 683wsg. It was cooled to -30degreesC. I did nothing to the images - no calibration, no rotation.

Each exposure was 600 seconds, with a Luminance filter.

I have plate solved the images and determined the coordinates. They are slightly different on the two images from 5th May and 8th May.

I looked at the RA of the tops of the trails and that shows that the trails are almost exactly North-South.

 

[GeorgeCarey]

The original FITS images are here http://geoastro.co.uk/may2013/Fits/

[GeorgeCarey]

I measured the distance in pixels between the anomalies. 1600.9 pixels for both images.

[GeorgeCarey]

This is the two images stacked with alignment on the left anomaly. The right hand anomaly is perfectly aligned.

 

[Bernd_Eppinger]

Hi, George,

 

It seems easier to say what it is likely not.

 

Using your celestial coordinates and the geographic coordinates of Stonehenge (which seems to be in Wiltshire), and interpreting your time either as British Summer Time or as UTC, I was not able to find any object in StarryNight Pro that could match your trailing stars of 2013-05-05. Although this doesn't 100% rule out a celestial object or a sattelite, it means at least that I can't say which object it is.

 

Also, I can't imagine any CCD defect that creates trails that are not exactly vertical or horizontal. An RBI effect during just 10s readout of such strength can also be ruled out, because you cooled your camera. If it is an RBI effect, then it needs more time (see below).

 

Reflections of Earthbound objects in your optics could cause parallel trails, but not with a bright "star" and a weaker "tail" like yours.

 

Reflections of a brighter star outside the image area are also unlikely, because the star that creates it would have the same movement as the telescope and the other stars, so it wouldn't cause a trail.

 

Shooting stars - also unlikely. Why always 2 and always almost exactly at the same positions in the sky?

 

 

A theoretical possibility: you imaged two very bright stars. Then, with the shutter open, you move the telescope away almost exactly in South direction. Immediately after that, you take an image at the new telescope position. In this case, you get the trailing stars as an RBI effect (i.e. a "ghost" image of your previous exposure). The 10 minutes exposure time should be sufficient to get most trapped RBI electrons into your image. In the resulting image, the two star trails would be absolutely parallel. Also, if you do the same thing twice, you get the same distance of the two stars and their trails each time.

 

Slightly North of your imaged area, I can see the bright stars "31 comae berenices" and "30 comae berenices". They are almost exactly north of your imaged area, which would fit to the direction of your trails. Their angular distance is 32'02", which seems approximately similar to the distances of your trailing stars. And they are almost exactly on an east-west line, with the right (western) star a bit more north, like your trailing image. Is it possible that you had the telescope pointing to these 2 stars and the camera shutter open, when you started to slew to the imaged area? That would have allowed the trails to appear as an RBI image on your next exposure.

 

 

Best regards,

 

Bernd

[GeorgeCarey]

I will check those two stars you mention. It certainly looks like a 'double image' of some sort, but I have checked the images taken before and after the 8th May image. They are both normal. The anomaly was number 4 in a sequence, so I am puzzled how the scope could have wandered off to look at 31 and 32 comae berenices. Autoslew seemed to be behaving OK. I had dithering set between images with a delay of 90 seconds to allow time to re-establish guiding.

On May the 5th the faulty image was the first in a sequence, so with that one there is more chance that the shutter was open at a different location. However, I still find this hard to accept because I use autoguiding, and set the scope guiding before any images are taken.

I may be able to inspect the guiding log from last night to see if somehow the scope did a sudden deviation.

 

George

Edited May 9, 2013 by GeorgeCarey

[GeorgeCarey]

The angular separation on 5th May was 1923.3 arc seconds.

On 8th May it was 1923.8 arc seconds.

 

The separation of 30 and 31 Comae Berenices is 1922.20, 1923.30, 1923.80 and 1923.4 depending on the source of the coordinates.

So agreement is better than 1 part in 2000.

 

Next clear night I will slew the scope North and see where these two stars are. Your solution to the problem has a lot going for it, but why were no stars in the ngc4725 area trailed?

Edited May 9, 2013 by GeorgeCarey

[GeorgeCarey]

I took the DSS images for ngc4725 and the two Coma Berenices stars and did some overlays. I had to rotate the Coma Berenices star image by 1.4 degrees anticlockwise.

 

The fit is very good.

 

All I need to figure out is how this could happen - twice!

 

[Bernd_Eppinger]

Hi, George,

 

My explanation obviously assumes that the faulty images are the first images in a series. If this is not the case, then I would have to assume that Autoslew suddenly jumps to 30 and 31 comae berenices and back, with the shutter open, and without storing that image to disk (otherwise you would have seen it). I agree that this is very unlikely.

 

What a pity; the direction and distance would have so well matched.

 

> "I still find this hard to accept because I use autoguiding, and set the scope guiding before any images are taken."

 

A delay of several seconds before taking the next image makes the RBI effect a bit weaker depending on the chip temperature, but it does not completely remove it. At least not, when the chip is cooled. Here is a link to a paper by Richard Crisp that explains RBI:

 

    http://www.narrowbandimaging.com/images/RBI_presentation_crisp_7249-22_with_speaker_comments_paper.pdf

 

I have once read a more detailed one, but I can't find it at the moment.

 

 

> "but why were no stars in the ngc4725 area trailed?"

 

Because the telescope was not moving when you imaged this area.

 

My theory, in a bit more detail, goes as follows:

 

1. You pointed the telescope at 30 and 31 comae berenices with the shutter of your camera open. This is a very weak point, because this seems not to have happened on May 8th.
2. You slewed towards NGC4725 with the shutter still open. That creates an image of north-pointing star-trails on the sensor. Areas with high infra-red contents, and according to some papers, also very saturated areas of this image, create trapped electrons below the photosensitive shift register of the CCD.
3. You close the shutter, and the camera flushes the sensor (i.e. reads out an image and possibly discards it). This removes all photo-electrons from the shift register. However, those that are trapped below the shift register structure are not removed.
4. After setting up dithering and waiting for the mount to stabilize, you make an exposure of NGC4724. As the mount is stable, you get no trailed stars from the NGC4725 area. But the trapped electrons from the previous exposure with the two bright trailed stars slowly leak into the photosensitive shift register of your CCD. After 10 minutes, when you finish the exposure, a considerable amount of trapped electrons has leaked into the shift register. When the CCD is read out, these electrons create a "ghost image" from the previous exposure.

Well, only a theory and with at least one very weak point, but I have nothing better at the moment.

 

 

Best regards,

 

Bernd.

Edited May 9, 2013 by Bernd_Eppinger

[GeorgeCarey]

I think we have to accept that the ghost stars were 30 and 31 Comae Berenices. How the scope pointed to them is a mystery. I have taken the guiding log from Maxim and very carefully isolated the part that goes with the errant image. I did a test image to confirm that the exposure time recorded in the FITS header is indeed the start of the exposure.

The guiding was without any sudden jumps. The previous image was also without any jumps.

 

What about this as a possibility? I use 5 seconds exposure with the guide camera. Autoslew makes a rapid Northward movement and returns, in less than 5 seconds. This might not show up on the guiding log. None of the stars around ngc4725 are bright enough to leave a trail. 30 and 31 Coma Berenices ARE bright enough so trails are made.

 

I will try and reproduce this next clear sky.

 

Edited May 9, 2013 by GeorgeCarey

[Bernd_Eppinger]

Hi, George,

 

> "The guiding was without any sudden jumps. The previous image was also without any jumps."

 

I assume this implies that there were also no sudden jumps between the two exposures, except for the expected dithering. If so, it would make RBI as an explanation really unlikely.

 

> "What about this as a possibility? [...] Autoslew makes a rapid Northward movement and returns, in less than 5 seconds. [...]"

 

I think that could work as an explanation. Not only would the stars around NGC4725 leave dimmer trails, but these trails would also be partly masked by the telescope spikes which become visible due to the much longer exposure time of these stars.

 

Concerning the brightnesses: according to StarryNight, the brightest star to be expected in your exposure is HIP62778 with a brightness of 7.53mag. The darker of the 2 suspected ghost stars is 30 Comae Berenices, allegedly with magnitude 5.75. Assuming those numbers are approximately correct, the intensity ratio between the two would be a factor of 5.15. This together with the bigger size and the spikes of HIP62778 could explain why its trail is not as well visible.

 

Your theory would also lead to the same shape of the star-trails that is visible in your images.

 

I also think that a DDM mount could be fast enough to do such a slew forth and back in 5 seconds or less.

 

My only problem is: I have never seen Autoslew doing this. What could have caused it?

 

Could you do one last test? Stack your two images that show the anomaly so that the stars are aligned, magnify the part around HIP62778 to pixel scale 1:1, and stretch it so that the sky background becomes a noisy gray. If your theory is right, then this star should have a very faint trail downwards, which might look just like a more prominent telescope spike. Perhaps it becomes visible after stacking. With some guessing, I think I might already see something like this in your post from 09.05.2013 04:32 PM, but it is not clear.

 

 

Best regards,

 

Bernd

[GeorgeCarey]

I had a look at the relative intensities of the two stars earlier today. Unfortunately the left one is fully saturated with many pixels at 64000 so accurate values are not possible. If the right one is taken to be mag 5.75 the left came at near mag 6. The fact that the left one is saturated gives us a clue as to how long Autoslew stayed on it before going back to ngc4725. I can do some test exposures to find the minimum time to reach saturation. Also if the star was saturated then RBI is less likely.

In my experience Autoslew can make sudden and unpredictable excursions. However, usually it reports an error. This was not the case last night.

My RA axis never misbehaves, but the DEC can give problems. Some weeks ago I was getting 'over current' errors that could not be corrected. I seem to have cured this by doing two things. First I tweaked the tuning and altered the values slightly from those found by auto tuning.

Secondly, and this may not really be doing anything, I swing the DEC axis through about 80 degrees by hand before switching on the power. I do this about 8 times. This is because the DEC axis seems to have some mechanical stiffness, and there maybe some tight spots that benefit from 'warming up exercises'.

Since I started doing this I have not had any over current errors, and position errors are much fewer.

I will have a look tomorrow at your later suggestions to see if the are any spikes on HIP62778.

 

PS - the scope is a refractor so there should not be any diffraction spikes.

Edited May 9, 2013 by GeorgeCarey

[Bernd_Eppinger]

Hello, George,

 

> "Also if the star was saturated then RBI is less likely."

 

I agree. I would even say this excludes RBI. I cannot see how RBI could create a ghost image with more than some 100 electrons per pixel, while saturation for the KAF8300 is at 25500 electrons.

 

> "the DEC axis seems to have some mechanical stiffness, and there maybe some tight spots"

 

I remember there were discussions in the Yahoo ASA forum about position errors. Some of the problems seemed to be caused by mechanical problems, especially for older DDM60 mounts. See here: http://tech.groups.yahoo.com/group/ASA_AstroSystemeAustria/message/7459. The post also says that ASA can repair it.

 

If we assume that your DDM60 really has a point where it "hangs" in the DE axis, then I think we could construct an explanation for your effect.

 

Assume the DE axis is really stopped at a tight spot during an exposure. Then the controller will increase the DE motor current in an attempt to overcome the tight spot. If it overcomes the friction without a position error, then the DE-axis will suddenly start moving quickly, because it is accelerated with high torque. The controller will try to counteract this, but it will have a nonzero reaction time which depends on the PID setting. Therefore, the telescope would make an overshoot in north- or south- direction and then come back to the original position. In your case, it would probably be the north direction. The whole thing would look similar to the first derivate of the step response which can be seen in the tuning dialog. If your refractor is not too heavy and the PID settings are good, it can happen within a second altogether.

 

One consequence of this explanation: At least with my PID settings, I remember a slight overshoot when the mount reaches the required position. Hence I would expect to see a trail on HIP62778 in south- and in north direction.

 

 

> "the scope is a refractor so there should not be any diffraction spikes."

 

This is interesting. I think I can see some weak "diffraction spikes" to north and south in your image posted on May 9th 04:32PM at HIP62778. If they can't be diffraction spikes, then they may be the trails we are looking for.

 

 

Best regards,

 

Bernd

[GeorgeCarey]

I think we have a clear picture of what happened. This is the picture of HIP62778

 

 

 

[GeorgeCarey]

Strangely there is a 10.92 magnitude star over on the right that has a beautiful straight line showing above and below it. The star is colour index 0.57 which may be why it has produced a bright trace,

 

Edited May 10, 2013 by GeorgeCarey

[Bernd_Eppinger]

Hi, George,

 

> "Strangely there is a 10.92 magnitude star over on the right that has a beautiful straight line showing above and below it."

 

Another miracle. I think it is likely that the trail is really caused by this star (I think TYC1990-2169-1), because the trail passes exactly through the star, and because I think I can see where the trail stops near the upper edge of the image. That would have been the maximum excursion of the overshoot.

 

The only explanation I have why colour could make a difference: HIP62778, according to StarryNight, has an extreme B-V colour index of 1.46, a surface temperature of 3567K and is 7.53mag bright. It seems to be a red giant. TYC1990-2169-1 has a B-V colour index of 0.73, a surface temperature of 5630K (i.e. almost sun-like) and a brightness of 10.84 mag.

 

I could imagine that the extreme red colour and perhaps infra-red of HIP62778 falls into an area where your refractor is not so well corrected and spreads the central part of the light over more pixels. This would make the area around the star trail brighter and the trail itself a bit darker and more difficult to detect. But I still find it hard to believe that this effect could make a difference of 3.31mag.

 

 

Best regards,

 

Bernd

Edited May 10, 2013 by Bernd_Eppinger

[GeorgeCarey]

Thanks for you help with this, Bernd. I will be watching my mount closely to see if it does it again.

[GeorgeCarey]

I managed to reproduce almost exactly the same effect. I started a 10 minute exposure on ngc4725 and then did a rapid slew northwards and then back to ngc4725.

30 and 31 Coma Berenices showed up in the right places. The trace was not as clean as in the previous examples and it looks like there was a small wobble  -  the return trace was slightly displaced. However, the guiding was not affected and no deviation showed on the graph. The slew was easily completed in 5 seconds.

Conclusion: it is possible for Autoslew to do a rapid excursion and return!

 

 

Edited May 11, 2013 by GeorgeCarey

[GeorgeCarey]

The clouds started coming over but I left a second exposure of 10 minutes running. Low and behold - it did it again!