Please help with hd 189733 b !

[alpha015]

Hi, I hope someone please can help me.

The other night, for the 12492394234 time, I tried to detect the transit of hd 189733 b, the easiest transit.

I used a pentacon 200 mm f2.8 of 50mm of aperture, cmos camera zwo asi 120 MM of 12 bits, a skywatcher eq3-2 mount and later the astroimageJ program to process the data.

i started recording at 1.20 (10 min before the beginning of the transit) and stopped at 3.40 (20 min after the end), because where I live is UTC+2, according to the ETD:

https://ibb.co/ceEshe

i took 147 photos, 1 per minute, each of them lasting 10 seconds

I used a gain (iso equivalent) of 17:

https://ibb.co/kxSnhe

the ingress is 0.48550 and the egress is 0.56065 according to the NASA explanet archivee:

https://ibb.co/bTkUvz

The apertures I choose are 5, 15 and 25

No matter what or how many reference stars I choose, this is the plot I get, which doesn't make any sense to me:

https://ibb.co/c3hnhe here you can download my data:

https://www.dropbox.com/s/4u4ndrqhubiqq56/data.zip?dl=0

when i turn on the fit mode, i get something like a light curve, but i believe the light curve should have already appeared without the fit mode:

https://ibb.co/mJH3Ne

I would appreciate if someone could please try with his laptop/program and tell me if he gets the same or different results

What am I doing wrong?

In another forum they said it might be because the photos are oversaturated, but they didn't look oversaturated when i was taking them, and I don't know how could they be oversaturated with 10 sec of exposure

Also some said that it may be because i may have been taking 8-bit photos, which I don't understand since the camera I use has a resolution of 12 bits; the program that came with the camera is SharpCap

[alpha015]

Anybody?

[robin_astro]

It is easy to check for saturation. Measure the highest individual pixel values in the star images. If any of them are equal (or close to) the value you get in an obviously over exposed image then the pixels are saturated and your magnitude measurements will be no good

Looking at the scatter of your individual values in your screen shot of the light curve, I would estimate your 2 sigma variation is around +-3% which is larger than the dip you are looking for so the dip will not be particularly obvious in the individual values.  (You can check how precise your measurement is by measuring one of your comparison stars instead of the exoplanet transit target. This should be a straight line  with points scattered around it).  The problem is you do not really have enough values before and after the transit to give a benchmark measurement to say for sure if you saw any dip.  Not sure what the fit plot is (I am not familiar with the software) but I suspect  it is just a plot of what you should expect to see. 

Why did you only take one image per minute ?  It would have been better to take as many images as possible. For example if you had taken 6 images per minute and averaged them, your variability would potentially have been  +-~1.2% making the dip easier to see.

Robin

[robin_astro]

Your images are 8 bit (The maximum value in your fits files is 255)  Producing 12 bit images will help but there are other problems.  Doing photometry to ~1% precision is not simple, particularly in a crowded field with a short focal length setup like this.  For example it is difficult to select an aperture to include just the star and your sky background anulus to exclude any background stars.  

Do you have a telescope ?  Things would be much easier with a smaller, more magnified field. Here is my observation of the egress of same star you are measuring  in 2005  using a modified web cam and telescope (I took 1800 3sec exposures and averaged them in blocks of 100)

http://www.threehillsobservatory.co.uk/astro/Photometry_HD189733b.htm

Robin

 

[robin_astro]

There are saturated pixels in your target and brightest comp stars (255 values).

This is going to be a challenge with your equipment. I would recommend refining your measurement technique outside transit.  Once you can get a straight line with points within say around +-0.5-1% in 5 min averages of exposures, you will be ready to try for a transit measurement.

Robin

[alpha015]

On 24/08/2018 at 17:41, robin_astro said:

Your images are 8 bit (The maximum value in your fits files is 255)  Producing 12 bit images will help but there are other problems.  Doing photometry to ~1% precision is not simple, particularly in a crowded field with a short focal length setup like this.  For example it is difficult to select an aperture to include just the star and your sky background anulus to exclude any background stars.  

Do you have a telescope ?  Things would be much easier with a smaller, more magnified field. Here is my observation of the egress of same star you are measuring  in 2005  using a modified web cam and telescope (I took 1800 3sec exposures and averaged them in blocks of 100)

http://www.threehillsobservatory.co.uk/astro/Photometry_HD189733b.htm

Robin

 

Thanks for your answer!

Yes, in other forums i have also been told the same, that i should change the settings to MONO16, so i can take images of 12 bits.

With respect to the equipment I have, it is more than enough. I'm in touch with someone who has previously detected many transits with the same equipment. 

Regards.

[robin_astro]

20 hours ago, alpha015 said:

With respect to the equipment I have, it is more than enough. I'm in touch with someone who has previously detected many transits with the same equipment. 

In that case I am not quite sure I understand why you are asking for help here. 

I did not say it could not be done but you are failing currently because your variability is too high.   It helps to have some knowledge of precision photometry  to get to this level of precision and at even a cursory glance I can see many areas where you could improve and I am no specialist in photometry.  It is difficult to have a sensible discussion when I see only part of it but here are a couple of references you might find useful

http://astrodennis.com/Guide.pdf

http://brucegary.net/book_EOA/x.htm

These are aimed at measuring much smaller depth transits than you are looking at but by following their recommendations your results will get better.

I wish you luck !

EDIT:  Wow on a reread I sound really curmudgeonly and condescending there, sorry for that !  I hope you manage to get a good transit. 

Robin

[robin_astro]

Here are a few more tips to help get your variability down

A stack of all your images (attached) shows significant drift (There is also some periodic error but that is a limitation of the mount.)  to reduce the variability due to changes in pixel  to pixel sensitivity, it is a good idea to image the stars at the same position as near as possible. Ideally you would auto guide but I am guessing your mount does not have this capability. You can improve your polar alignment though and with such a low magnification it should with care be able to minimise the movement of the stars during the run.

You should also dark and flat correct your images. The reason for this is you have a lot of warm pixels in your image and as your stars drift they will move in and out of the aperture and upset your measurement.  Similarly unless you flat field, any dust donuts will cause the brightness measurements to vary as the star drifts across them. (These can look vary similar to transits)

You should also try to centre your target to minimise any drift in brightness due to vignetting edge effects, though a good flat field should get of this.

These are all standard practise when doing any photometry but are especially important at this level of detection

Cheers

Robin

[alpha015]

4 hours ago, robin_astro said:

Here are a few more tips to help get your variability down

A stack of all your images (attached) shows significant drift (There is also some periodic error but that is a limitation of the mount.)  to reduce the variability due to changes in pixel  to pixel sensitivity, it is a good idea to image the stars at the same position as near as possible. Ideally you would auto guide but I am guessing your mount does not have this capability. You can improve your polar alignment though and with such a low magnification it should with care be able to minimise the movement of the stars during the run.

You should also dark and flat correct your images. The reason for this is you have a lot of warm pixels in your image and as your stars drift they will move in and out of the aperture and upset your measurement.  Similarly unless you flat field, any dust donuts will cause the brightness measurements to vary as the star drifts across them. (These can look vary similar to transits)

You should also try to centre your target to minimise any drift in brightness due to vignetting edge effects, though a good flat field should get of this.

These are all standard practise when doing any photometry but are especially important at this level of detection

Cheers

Robin

Thanks for your help ! 

[alpha015]

Btw quick question please, in sharpcap, should I select 'single frame' , unlimited or a specific number of frames? if so, how many?

https://ibb.co/dr4SNz

[robin_astro]

Not used Sharpcap but looking at the options, if sequence is ticked I would expect that would override the capture limit settings and it would capture a sequence of images of selected exposure length at selected intervals (not sure what happens if the interval is shorter than exposure+download time -hopefully it flags an error)

Otherwise it will either take a single image if ticked or a sequence as fast as possible (ie no delay between exposures) until either manually stopped, or until the the time or number of exposures limit is reached.

Easy to check though by experiment and see what happens before doing a run for real

For this exercise, I would probably select the sequence option, set the number of exposures to a v high value like 99999 so it keeps taking exposures until I tell it to stop and set the interval to be slightly longer than the exposure time to maximise the number of measurements but still keep them at even intervals for convenience when plotting. eg if the exposure is 8 sec set the interval to say 10 sec and run for as long as you need to, say 4 hours, giving ~1500 exposures.

[robin_astro]

Dont forget to take your darks and flats too.  Take at least 30 of each say and do a median average.

If you camera is not temperature controlled take a set of darks at the start and end of the run and average

Flats can be taken using the identical setup including focus setting, either at twilight while pointing at a region clear of any visible stars near the zenith, or during the day, again at the zenith using a diffuser (traditionally a white T shirt) in front of the aperture.  Expose so the average pixel count is ~1/2 of the maximum when saturated.

Robin

 

[alpha015]

57 minutes ago, robin_astro said:

Dont forget to take your darks and flats too.  Take at least 30 of each say and do a median average.

If you camera is not temperature controlled take a set of darks at the start and end of the run and average

Flats can be taken using the identical setup including focus setting, either at twilight while pointing at a region clear of any visible stars near the zenith, or during the day, again at the zenith using a diffuser (traditionally a white T shirt) in front of the aperture.  Expose so the average pixel count is ~1/2 of the maximum when saturated.

Robin

 

alright thanks  a lot!

[alpha015]

1 hour ago, robin_astro said:

Dont forget to take your darks and flats too.  Take at least 30 of each say and do a median average.

If you camera is not temperature controlled take a set of darks at the start and end of the run and average

Flats can be taken using the identical setup including focus setting, either at twilight while pointing at a region clear of any visible stars near the zenith, or during the day, again at the zenith using a diffuser (traditionally a white T shirt) in front of the aperture.  Expose so the average pixel count is ~1/2 of the maximum when saturated.

Robin

 

btw last night i tried to capture the transit but again failed

i used MONO16 but this error appeared with astroimagej: 'aperture peak count over saturation limit'

i believe this means that the images i took were oversaturated, which i don't understand since i was taking exposures of 10 seconds LOL

https://ibb.co/dWtVaK

also the photos i took yesterday looked on SharpCap far less bright than the one above on Astronimagej, why is this?

a friend used to the lens and camera i used suggested me to use a focal of 4 instead of 2.8 ( which i was using)

but selecting a focal of 4 would mean that i would only have 34 mm of aperture :S 

also what gain should i choose for an exposure of 10 seconds? i have been using 20 ( out of 100 ) 

 

[robin_astro]

Can you upload a single fits image that shows the error please

Stopping down the lens  to f4 can give sharper stars at the edges but the star images around your target looked OK in your original set so I dont see any obvious  reason to reduce the aperture.

Robin

[alpha015]

1 hour ago, robin_astro said:

Can you upload a single fits image that shows the error please

Stopping down the lens  to f4 can give sharper stars at the edges but the star images around your target looked OK in your original set so I dont see any obvious  reason to reduce the aperture.

Robin

Thanks for your answer.

Sure, I attach it.

I started recording 10 min after the beginning of the transit and finished 20 min after the end.

So just captured partial transit, but i think the light curve would show up.

Capture_00112.fits

It is true the star was high in the sky at the beginning and near the horizon at the end of the transit, so the overall luminosity of the photos went down with time. Not sure if this might be the reason why i didn't get the light curve. Or this together with the fact that i didn't take darks and flats last night and my photos have many hot pixels. 

The thing is that i didn't have problems with saturation before. Just last night, when i used mono16 for the first time.

[robin_astro]

Your target and some of reference stars are saturated in this 16 bit  image (several pixels with 65520 values).  I also told you earlier that you  had saturated stars in your 8 bit images (several pixels with 255 values). 

The exposure you need will vary from night to night depending on atmospheric conditions and will also vary through the night as the star changes in height above the horizon.  I recommend checking some images before you start a run and adjusting your exposure  to limit the maximum pixel values to ~30000.  This should give enough room in case  the brightness increases during the run.   

I also suggest  reading all I have already recommended in this thread and applying it to the next run.

Robin

[alpha015]

9 minutes ago, robin_astro said:

Your target and some of reference stars are saturated in this 16 bit  image (several pixels with 65520 values).  I also told you earlier that you  had saturated stars in your 8 bit images (several pixels with 255 values). 

The exposure you need will vary from night to night depending on atmospheric conditions and will also vary through the night as the star changes in height above the horizon.  I recommend checking some images before you start a run and adjusting your exposure  to limit the maximum pixel values to ~30000.  This should give enough room in case  the brightness increases during the run.   

I also suggest  reading all I have already recommended in this thread and applying it to the next run.

Robin

Thank you, i will do that and post the results here.

Hopefully next sunday. 

[robin_astro]

14 hours ago, alpha015 said:

I started recording 10 min after the beginning of the transit and finished 20 min after the end.

I recommend running for as long as possible before and after the transit. The more data you collect the more obvious the transit will be. 10 min is too short to get a good reference level.

15 hours ago, alpha015 said:

what gain should i choose for an exposure of 10 seconds? i have been using 20 ( out of 100 )

Looking at the camera specification I would recommend running at a lower gain (say 10) so  you can use more of the dynamic range of the sensor 

The camera sensor is capable of recording up to ~14000 electrons (14.5k FWD) but the 12 bit ADC can only record 4096 counts.  A gain or 10 corresponds to a gain of  2.2e-/pixel which would allow you to record up to ~9000 electrons which should still be in the linear region of the sensor.  (The increased read noise does not matter. we are measuring bright stars with long exposures here, the read noise is insignificant)

  (Note the capture software you are using stretches the 12 bit camera output, multiplying it by 16 to make the image look like 16 bits.  The saturated 65520 counts you see in the image are really 4095 counts from the camera (2^12-1)

This should also help with your over exposure problem but you might still need to reduce the exposure time from your 10 sec to avoid any saturation problems

Don't forget, you dont need to wait for a transit to practice.  You can measure the star outside transit any time, trying to get a flat line with as little scatter in the measurements as possible.  

Robin

[alpha015]

9 hours ago, robin_astro said:

I recommend running for as long as possible before and after the transit. The more data you collect the more obvious the transit will be. 10 min is too short to get a good reference level.

Looking at the camera specification I would recommend running at a lower gain (say 10) so  you can use more of the dynamic range of the sensor 

The camera sensor is capable of recording up to ~14000 electrons (14.5k FWD) but the 12 bit ADC can only record 4096 counts.  A gain or 10 corresponds to a gain of  2.2e-/pixel which would allow you to record up to ~9000 electrons which should still be in the linear region of the sensor.  (The increased read noise does not matter. we are measuring bright stars with long exposures here, the read noise is insignificant)

  (Note the capture software you are using stretches the 12 bit camera output, multiplying it by 16 to make the image look like 16 bits.  The saturated 65520 counts you see in the image are really 4095 counts from the camera (2^12-1)

This should also help with your over exposure problem but you might still need to reduce the exposure time from your 10 sec to avoid any saturation problems

Don't forget, you dont need to wait for a transit to practice.  You can measure the star outside transit any time, trying to get a flat line with as little scatter in the measurements as possible.  

Robin

Thanks a lot!

Perfect, gain of 10 and more time before and after the transit. I will also check the results with less exposure time. 

Yes indeed, tomorrow or next day i will try to practice without a transit , if the weather is good.

I will post the results here.

[alpha015]

On 01/09/2018 at 13:00, robin_astro said:

I recommend running for as long as possible before and after the transit. The more data you collect the more obvious the transit will be. 10 min is too short to get a good reference level.

Looking at the camera specification I would recommend running at a lower gain (say 10) so  you can use more of the dynamic range of the sensor 

The camera sensor is capable of recording up to ~14000 electrons (14.5k FWD) but the 12 bit ADC can only record 4096 counts.  A gain or 10 corresponds to a gain of  2.2e-/pixel which would allow you to record up to ~9000 electrons which should still be in the linear region of the sensor.  (The increased read noise does not matter. we are measuring bright stars with long exposures here, the read noise is insignificant)

  (Note the capture software you are using stretches the 12 bit camera output, multiplying it by 16 to make the image look like 16 bits.  The saturated 65520 counts you see in the image are really 4095 counts from the camera (2^12-1)

This should also help with your over exposure problem but you might still need to reduce the exposure time from your 10 sec to avoid any saturation problems

Don't forget, you dont need to wait for a transit to practice.  You can measure the star outside transit any time, trying to get a flat line with as little scatter in the measurements as possible.  

Robin

Hi,

I did it tonight. These are the results, each one 5 minutes of data:

1. exposure 10 seconds - gain 10

https://ibb.co/jMgKEe

2. exposure 10 seconds - gain 5

https://ibb.co/bOmrSz

3. exposure 30 seconds - gain 5

https://ibb.co/iZhLLK

4. exposure 30 seconds - gain 10

https://ibb.co/fmYALK

All of them oversaturated lol

Maybe I should try exposure 10 seconds gain 0?

I will try again on thrusday night. 

Btw i didn't take darks and flats yet, but tomorrow yes.

 

In this post in cloudynights the exposure used was 15 seconds: https://www.cloudynights.com/topic/462161-detect-known-exoplanet-with-dslrtelephote-lens/

In this other, 50 seconds: https://spectrum.ieee.org/geek-life/hands-on/diy-exoplanet-detector

[Whirlwind]

The examples were done with DSLRs though (and probably unmoddified).  You have a much more sensitive camera.  The DSLRs also have filters to restrict certain wavelengths, you have a white light camera.  You can't really relate in the same way to exposure times.  Try shortening your exposure times to 5 secs or even less.  You want to aim for 16bit converted counts in the brightest pixel of no more than about 30,000 - 35,000 counts.  Alternatively you could try a filter of some type to reduce the flux to a manageable level (or at least knock out far red/IR light).

[robin_astro]

You can go to 5 gain but at 0 gain there is a risk that the sensor will saturate before the 12 bit ADC which is best avoided as it makes it difficult to see when you are saturated. 

I would keep reducing the exposure time rather than filtering as you collect more photons in total by taking more images and averaging them.  (The variability between individual  images increases with shorter exposures due to scintillation but reduces again after you combine the exposures.)

 Ignore other people's exposure times, this is totally dependent on many factors including the equipment, seeing, sky conditions at the time, how well you focus the altitude of the target, etc ,etc. 

The best exposure time will not necessarily even be the same for each run so you need to do exposure checks at the start of a run, remembering that if the target is rising during the run, that the counts can rise significantly so you need to allow for this .   Whirlwind's suggestion of ~30000 counts agrees  with my recommendation up the thread

Cheers

Robin

[robin_astro]

Can you post an example of your  image at gain  5, exposure 10 sec please ?

Thanks

 

[alpha015]

56 minutes ago, robin_astro said:

You can go to 5 gain but at 0 gain there is a risk that the sensor will saturate before the 12 bit ADC which is best avoided as it makes it difficult to see when you are saturated. 

I would keep reducing the exposure time rather than filtering as you collect more photons in total by taking more images and averaging them.  (The variability between individual  images increases with shorter exposures due to scintillation but reduces again after you combine the exposures.)

 Ignore other people's exposure times, this is totally dependent on many factors including the equipment, seeing, sky conditions at the time, how well you focus the altitude of the target, etc ,etc. 

The best exposure time will not necessarily even be the same for each run so you need to do exposure checks at the start of a run, remembering that if the target is rising during the run, that the counts can rise significantly so you need to allow for this .   Whirlwind's suggestion of ~30000 counts agrees  with my recommendation up the thread

Cheers

Robin

Thanks again for your help.

Perfect I will try again tonight.