best sensor for photometry ?

Affiliation
American Association of Variable Star Observers (AAVSO)
Wed, 05/09/2018 - 21:27

 
Hello !

What will be the best performer to practice photometry?

A color CMOS sensor of 14 bits (IMX 193 for example with QHY247C)
A 16 bit black and white CCD sensor (ICX814 for examplewith QHY23)

What are the features to compare to make the right choice between 2 different sensors?

Thank you for your opinions.

Affiliation
American Association of Variable Star Observers (AAVSO)
first photometry camera

IMO either of those cameras would be a very poor choice for photometry.   

I  suggest that you read the first half of the AAVSO CCD Photometry Guide before you consider other cameras.  Pay particular attention to  matching the pixel size of the camera to the focal length of your telescope and your local seeing conditions.  You must also consider the Full Well Depth which is an important factor in the dynamic range your camera will have.  Small pixels give small FWD's.

For a first camera I'd suggest you look for a used CCD camera on Astromart or Cloudy nights, but read up on the important factors you need to consider before you pick a camera.

Phil

 

Affiliation
American Association of Variable Star Observers (AAVSO)
hi phil,

hi phil,
thank you for your reply.

can you develop a little why the choice of these cameras would be bad?
QHY247C 14-bit ADC has a welldepth of 36000e- 3.91 pixels and a large fields ... but a matrix of color bayer
QHY23 16 bit ADC a welldepth of 20000e- 3.69 micron pixels monochrome and requires a filter wheel + 2 filters V and B cousin
pixels are not very big, but look like efficient.

In addition to the welldepth, the size of the pixels, and the size of the sensor what are these important parameters to take into account, I do not find too much information on the guide of the aVSOO DSLR and CCD.

in the AAVSO CCD guide, unless I'm mistaken, they only mention the welldepth, the absence of antiblooming, the size of the field, and the sampling which should be between 2 and 3 pixels.

I have a newton tube of 200/800

 

thanks.

Affiliation
American Association of Variable Star Observers (AAVSO)
First camera for photometry

Fair enough.

First, I'm not saying you can't do photometry with these cameras.  I'm saying they are bad choices because there are better, cheaper cameras which  are much better suited to photometry: less processing, better SNR, greater dynamic range..

QHY247C:  I would reject any one shot color camera from the start, CMOS or CCD.  You're throwing away most of the detected photons.  Yes, you can do photometry with a OSC camera  if you jump through enough hoops, but if you're going to buy a camera for the purpose of doing photometry this camera would not make sense.  And yes, you would need a filter wheel and filters, but these would be filters designed for photometry rather the red, green, and blue tricolor filters of the bayer matrix.

QHY23:  This would be a better choice, but in my opinion, still not a good one.  For your scope this camera would give good sampling for the seeing at typical amateur sites but, at least for me, the FWD is a problem.  In my very limited experience with CMOS cameras, 20K e- just hasn't been enough to be able to attain an exceptable SNR (100) and still be able to measure stars in the same image which are separated by more than about 1.5 to 2 magnitudes.   I just think this is too limited.

As an alternative example, there is a used ST402me CCD camera with 9 micron pixels and 100Ke- FWD for sale now on astromart for $600.  This camera was designed to do photometry.  The photometric BVI filter wheel for this camera is still available new for $379. 

The FOV would be smaller, and the sampling would not be as good, (but this would be offset by less then perfect focusing).   ....less than $1000 total.  Also, this is a camera you could continue to use if you later get a telescope with a longer focal length.

Phil

Affiliation
American Association of Variable Star Observers (AAVSO)
ah. sorry, I did not

ah. sorry, I did not understand the nuance. My English is not very good.
I agree with you for the color camera. This choice is not the best, but for the moment offered the best compromise for me welldepth / sensor dimensions / ease of use.
 But actually the matrix of Bayer is damaging.

For the QHY23 it seems to me that it's a CCD sensor. but ..., actually 20000th -...

I do not know enough to buy a second hand camera so I would prefer to make my choice in a new camera.
That's why I ask what are the criteria to compare and make the "least worst" choice

st402M seems effectively exellent, but with a very small (too small ?) sensor.

Clearly from what I understand from the exchange note, the priority is:
Mono chrome sensor
welle depth important (often equal to large pixel size)

are there other things to consider?

I will try to find my happiness in a new camera, if this is impossible in my budget (3000 € max) then I would turn to the occasion.

 

thanks for your help.

Affiliation
American Association of Variable Star Observers (AAVSO)
QHY23

Pardon me, Jean-Luis,  I was mistaken.  As you said, the QHY23 is a CCD camera, not CMOS.  This is an important difference, since the CCD chip can be binned (small pixels combined to make larger pixels) before they are read out.   This is called "on-chip" binning.   This is not possible with CMOS.  This binned CCD pixel has a much larger full well depth.  This would removed much of my objection to this camera.  (CMOS chips can be binned, but only after they are read out.)

There are two other problems with this camera which make it less than ideal for photometry.  It is "anti-blooming".   This is a way of keeping the pixels from saturating with electrons which can spill out into adjacent pixels and cause the star image to "bloom"- produce a long trail in nearby pixel.  Antiblooming causes the camera to lose its linear response well before the pixel's full well depth is reached. 

My other objection to this camera is that (it seems) that cooling is not regulated.  This means the camera temperature can change during the observing period.  Thus the "dark current", the electrons generated in the CCD as a result of heat rather than photons from the sky, will vary.  During the calibration of the image the dark current cannot be removed with the same accuracy as with a temperature regulated camera. 

Both of these last two issues can be managed, but there are compromises that must be made.  Since you are still shopping for a camera,  I think it would be better to find a camera that is better suited to photometry.

The AAVSO CCD Photometry Guide is available in French.

https://www.aavso.org/ccd-photometry-guide-french

Chapter 3 discusses most of these issues.

Phil

 

 

 

Affiliation
American Association of Variable Star Observers (AAVSO)
I think if one is willing to

I think if one is willing to spend in this price range, you should consider a modern, back-illuminated cooled monochrome CMOS sensor. I would always consider buying used astronomy gear, but *not* image sensors, because they keep improving with every new generation.

Let me explain a bit:monochrome sensors will allow you to use photometric filters (which are rather expensive tho, you need to fold in the cost for them to make the comparison apples-to-apples). For starters you will want to have at least the "V" and "B" band photometric filters. You can do photometry with RGB filters that are part of color sensors, but when you are in the 4 digit $ range, this might be a compromise that is perhaps not necessary. Even if you want to use the camera also for pretty pictures once in a while, the ability to use arbitrary filters (e.g. also narrow band filters like for H_alpha) will be useful.

CMOS vs CCD: a topic of its own ....  take some popcorn, google the discussions and enjoy the show :-). My personal opinion is that modern CMOS sensors offer more photons per second per $ investment. If you think about using the sensor for planetary imaging as well, you will appreciate a sensor that will  be able to record videos with lots of short exposure frames.

Resolution: consider fewer but larger pixels for photometry, for reasons already mentioned (effective dynamic range because of full well capacity, and limits on resolution from the telescope's specs (aperture, focal length) and atmospheric seeing.  Also processing needlessly large (in pixel count) images in photometry software slows down the processing and makes storage/archieval of images more cumbersome.

CS

HB

Affiliation
American Association of Variable Star Observers (AAVSO)
Bikeman, thank you for your

Bikeman, thank you for your opinion.
and I rather agree on the use of a monochrome sensor. I only have advantages

I own a tube newton 200/800. and indeed, my PC is not very powerful and I prefer to treat some heavy images that thousands ....
do you have a sensor monochrome backlit cmos? to give me an example

from what I found, it is difficult to find a CMOS sensor with a large well depth and with a large ADC, and with a large field

in general they are either 12-bit with a poor welldepth, and very small fields.

or better, 14 bits with interesting welldepth but in color sensor like the IMS071

with big pixel and big well depth I found the IMX174 with a 32000e welledpepth and pixels of 5.86microsn and ADC of 12bits, but a demonstration of R.PIERRI on this forum dissuaded me to talk about its weak dynamics!

for you ,what are themains criters of the camera to compare?
and coan we mathematically compare a CMOS sensor to a CCD sensor?

thanks for your answers

Affiliation
American Association of Variable Star Observers (AAVSO)
It depends what are your

It depends what are your targets. I use 14-bit monochrome CMOS (IMX178), higher gain (150/510) to reduce read noise and short exposures, which I stack later (what's more, I can filter them however I want, for example in "lucky" imaging). With such gain I have only ~2800 FW and I don't need more, because seeing has larger impact on it than noise. My information could even store in 12-bits, so I don't even use the advantage that it's actually ADC of 14 bits. Using 300mm f/2.8 lens (107mm of diameter only), for typical 10s frames, I bin measurements two 2-minute or 3-minute stacks (120/180s) for 17.5-18.0 mag and 15-minute stacks (900s) for fainter objects, which goes down to 19 mag, having saturation point at ~11 mag. In red photometric filter, to add.

That is one of CMOS advantages, where you can make short exposures and bin for higher dynamic range (I believe TESS is doing that, but it has CCDs). To compare if CMOS or CCD give better results, I just test and compare to measurements in ETD (Exoplanet Transit Database), as observers upload more precise measurements than I usually find. I found out both have similar accuracy and that's why I will stick with CMOS. It has more possibilities (occultations etc.), but more limits. As you said, sensor size (lack of reference stars for bright targets, but more cameras with better sensors appear on astronomy market), small full well (if you want to use it for 180+ s subframes), no possibility of binning (it's software, not hardware, so better if you don't use it at all), smaller pixels (to be honest, it's actually an advantage for my use, but it requires more precise tracking - so that's why I use short exposures too) and probably require more space (if you plan to use short exposures, of course - my typical exoplanet transit observation takes 10-50 GB of data, for example 5000x7s, 4 MB each, which is even cropped to reduce the size).

Think about full well too. If you have 32000e on IMX174, how can you store the information in 12-bit ADC, which has only 4096 levels? That's why we have unity gain, to use maximum information from that. With smaller gain, you only drop data. For example, let's say you have 12345 e-, which in 0 gain gives you (12345/32000)*4096 ADU, which is 1580.16. No, it won't save as "1580.16", but "1580". It works like that.

IMX174, IMX290 and IMX224 have large ampglow, the last one has the largest. IMX178 is the cheapest CMOS sensor which doesn't have a problem with that. All sensors starting from ASI1600 and higher, don't have ampglow problem either.

Oh, and forgot to mention - the price! I'm happy that I'm using a camera which I bought for 500$ only, while having similar results to those CCDs for 1500$+. Magnitude accuracy test is the key. Some people have shown high quality light curves with ASI cameras. Yes, these are real.

Regards,

Gabriel

Affiliation
American Association of Variable Star Observers (AAVSO)
if i understand....

OK,

From what I understand to your answers, (and to the different readings on this subject on the forum).

it seems that the choice of an astro camera (for photometry) is done on the following points:

- WELLDEPTH the most important possible (2 questions:

                  * How much do you think it's enough for a 16-bit, 14-bit and 12-bit ADC?

                   * Is there a number beyond which it becomes useless (I saw some camera with well depth of 170000 !!!?

- largest pixel size possible. (but …. what about the ideal sample?)

- fields as large as possible (what about the vignetting this causes)

- ideally monochrome

- and of course the price ...

if that's right ???(but it is ??? ) it puts out most of the CMOS sensors, which have a very small well depth. (Except some cameras that use DSLR color sensors, but with a BAyer matrix)

on the characteristics of the cameras I often see the terms RON (between 5 e- and 15 e-) I imagine that  little is the noise better is ...

But also the term "Gain" in e- / ADU (0.8e- / ADU for a Moravian G3-11000 for example or 1.1e- / ADU for a QSI683 or  ATIK4000 at 0.61e-/adu) does this criterion also make it possible to compare the cameras between them?

Sorry for all these questions, but I need to see clearly to make the right choice and bring good results to AAVSO.

Thanks again for your opinion.

Affiliation
American Association of Variable Star Observers (AAVSO)
Read noise and Gain

Read noise is the uncertainty in the signal that results from the process of "reading" the pixel.  Smaller is better, but in most situations it is the least important of the major sources of noise in the  image.

In a CCD camera the gain is usually set by the manufacturer based on the full well depth, read noise, and the bit depth of the analog to digital converter (ADC) and is not adjustable.  It is not of much importance when deciding which camera to buy.

For your other questions, reading the CCD Photometry Guide will help you understand what is important.

Affiliation
American Association of Variable Star Observers (AAVSO)
Welldepth an binning

I think the point about binning that was made by others here and in a related thread are valid, and somewhat diminish the significance of the well depth as a limiting factor. If a sensor you consider has (following the fashion of the time) very high resolution, therefore small pixel size and low well depth, you can compensate for that by adding neighboring pixels together e.g. by 2x2 binning , or by making more but shorter exposures. There's no free lunch tho, you pay a price, because if the binning is done in software (as, I understand,  with CMOS sensors, or if you are binning "in time" by stacking exposures), you accumulate more read-out noise in the final image. Then again modern CMOS sensors have very low read-out noise, so it's a compromise worth considering. And binning pixels togther early in the processing pipeline makes processing and storage of images cheaper for the following steps. 

I wasn't aware until now that many of the 14bit color CMOS chips which otherwise would look promising for photometry do not have monochrome variants :-(. That's a shame.

I also use an IMX178 monchrome based camera. I wish the chip was bigger, but your scope is very fast (good) with a focal length of "only" 800 mm, which means you will get a field of  ca 32 x 21 arcmin even on this small chip (IMX178). 

You can use this nice online tool http://www.skyatnightmagazine.com/astronomy-field-view-calculator to not just calculate the FOV, but also take a virtual look at different astronomy objects (actually any field)  to simulate what they would look like on a particular chip if you type in the specs of your scope and chip. You can also use the AAVSO finder chart generator and select the respective field of view to get an idea about comparison star density for objects you would like to do photometry on. Just to get a feel whether you are OK with a certain FOV.

>on the characteristics of the cameras I often see the terms RON (between 5 e- and 15 e-)

>I imagine that  little is the noise better is ...

Read-out-noise , usually the number is "e- rms", (root mean square error in electrons). Yes, smaller is better. So if you have a sensor with tons of small pixels where binning (see above)  might be useful, they should better have a small RON . e.g.  IMX178 based cameras have specs claiming ca 2.2 .. 1.4 electrons rms, even doubling it via binning still gives a good value. 

>- - fields as large as possible (what about the vignetting this causes)

Vignetting (and some other instrumental artifacts!) are compensated for with flat field images (see the tutorials on CCD and DSLR photometry). I would not worry about "too big" a field per se.

CS HB

Affiliation
American Association of Variable Star Observers (AAVSO)
Binning...

Hello Heinz-Bernd,

In fact in photometry we are making a very large soft binning in any case, this is just the number of pixels that are in our inner circle, easily 100 (5x5 to 15x15 of 3.7 microns pixel in my case, at 800 mm focal lenght, 8", depending the defocus). Then in photometry the SNR is, for the most of it, defined by the shot noise of the signal itself. In case of a weak signal like 10000 e- in the inner aperture, the shot noise is 100 e- and then the SNR is 100. Now let say you have 2 e- Nyquist-Johnson noise per pixel due to the sensor itself (like yours), for the 100 pixels of the inner aperture it's 20 e- noise (quadratic combine). What the contribution of it ? A new time quadratic, the total noise is then 102 e- instead 100 ! An incease of 2% of the noise, and a reduction of 2% of the SNR, totally negligible...

In fact we also have to consider other sources of shot noise (sky, dark current...), overall the NJ noise of the sensor (RON of commercial datasheet) is no more a parameter of interest for us. 

Clear Skies !

Roger

Affiliation
American Association of Variable Star Observers (AAVSO)
quid about ASI294 MC ?

Hello Roger,

After having read carefully different comments about CMOS sensors, and our different e-mail exchanges.
I have a question about using the following camera:

ZWO ASI294MC. https://astronomy-imaging-camera.com/product/asi294mc-pro-color

apart from the fact that it is not monochrome, it meets all the requirements for photometry. dynamic; 14-bit ADC and fullwell of 63700 e-, large sensor size. low read noise (1.2e-), pixels of 4.63microns

and has the immense advantage of being cooled, and having a good sensitivity in the red for the photo of the classical sky.

Do you think that we can consider that it can be used for photometry as a DSLR and usable as such?

or ,  to choose material for photometry, it must definitvely use a monochrome sensor?

I think you've already answered this question, but would like to be clear. because I read often (certainly by ignorance of this practice) that to obtain serious results in photometry,It is illusory with a matrix RGB

thank you very much for your reply.