Imaging / Photometry for Variable Star observing - how to go fainter

[Mr Spock]

Being new to this imaging malarky, my knowledge is very limited. So, I would be grateful for some assistance. My main aim is to do photometry of variable stars, not deep sky imaging.

So far I've determined the best settings for my Nikon D500 with Sigma 150mm f2.8 lens to go faint is 30s @ f4, ISO 400. This gives the best SNR for the camera, least aberrations for the lens, and is just on the verge of my horrendous light pollution showing though. With this I can see on my RAW file, comparison star DD on the chart for T Uma which is mag 14.1. It's actually too faint to be useable, but AA at 13.2 is easy. 

I should add, for the software to work, I have to use an unprocessed RAW file. Any kind of processing corrupts the data.

Now I've reached 14.2, with 13.2-13.5 useable, the question is, how do I get below that. Maybe as far as 15? I can't increase the exposure due to light pollution. So am I stuck at this, which does allow me to follow Mira variables, or is there a solution?

Would using a scope work? My FS60 with reducer would be 256mm f4.2 for example. Obviously I don't want to spend huge amounts of money on a reducer for that if it is of no benefit. 

Thanks for any assistance.

[vlaiv]

What about stacking?

With stacking you can go as deep as you want - but not sure if that qualifies as "processing".

Personally, I don't see any reason why not to stack the data - if one take scare of how it's done. The least intrusive method for stacking would be good old "shift and add".

That is literally equivalent of taking longer exposure. no interpolation is done and only difference is level of read noise that is present in such "exposure".

Drawback is that you'll get lower resolution data - but given that you are using small lens, and that you are measuring brightness - you don't really care about resolution.

5 minutes ago, Mr Spock said:

Would using a scope work? My FS60 with reducer would be 256mm f4.2 for example. Obviously I don't want to spend huge amounts of money on a reducer for that if it is of no benefit. 

Alternative is to increase aperture (while keeping sampling rate the same). Again some "processing" will be required here - like binning the data, but again, I don't see that being disruptive for photometric work in any way.

 

[Mr Spock]

So are we saying a 6" f4 Newt @ ISO 400 and 30s would go a lot fainter but not increase the background light pollution?

The software will stack but I'm trying to keep things as simple as possible. If things like image processing get too complicated I'll lose interest. The images are just a means to an end for me.

[vlaiv]

25 minutes ago, Mr Spock said:

So are we saying a 6" f4 Newt @ ISO 400 and 30s would go a lot fainter but not increase the background light pollution?

It will certainly increase level of light pollution recorded - more aperture gathers more light from both the source and from the sky.

What more aperture and longer integration times bring is higher SNR. Sky background signal is nothing more than DC offset. It gets removed in aperture photometry when surrounding annulus value is subtracted from the star.

What is problematic is shot noise associated with LP/Sky background signal. Shot noise grows as square root of sky background level - so it grows slower than signal itself.

If you increase aperture by x2 (by diameter) or integration time by x4 - you will increase signal by factor of x4 (either like time or like surface of aperture), but associated noise will grow by square root of that - or only by factor of x2.

Overall SNR improves by factor of x4 / x2 = x2.

Once you have good SNR - you can distinguish faint stars from the noise of the background and they become detectable.

Just using longer exposure with your current setup does the same - and in principle - you could just use longer single exposure - if there was not for the issue of limited full well capacity.  Pixels can only hold so much signal - and if you saturate them - you loose information about the value they were holding. This is where increase in aperture and stacking come in handy - they allow you to gather more signal without pixels being saturated.

36 minutes ago, Mr Spock said:

The software will stack but I'm trying to keep things as simple as possible. If things like image processing get too complicated I'll lose interest. The images are just a means to an end for me.

I understand how you feel about image processing, and yes, that is something that you should not waste your time with - for this purpose.

However, there is data preprocessing step - that is always performed in scientific work - like calibration and data deduction steps. These are well defined and not subject to ones opinions or preferences (like image processing is). They can be automated really well. With suitable software and definition of a few macros - whole process can be reduced to just pointing to a single folder containing data to process (ie - here is a bunch of recordings - please produce final calibrated and stacked data image that is equivalent to single very long exposure without fear of pixel saturation).

[Mr Spock]

I'm going to do more experiments with the kit I have to see how far I can push it. Starting with the FS-60CB - I reckon as that is f5.9, just more than a stop slower than I've been using, upping the ISO to 1000 should produce the same exposure at 30s. 355mm should enable me to get tight in on Mira stars at their minimum. Such as T Uma which is currently 12.8 and fading.

I should add all these variable star readings will be for my own use. I had a look at the BAA's photometry spreadsheet - sadly I won't be going down that route as I don't have a degree in gibberish  The visual programme is estimates to 0.1 magnitude with a huge scatter amongst observers. I'm not sure my current +-0.02 magnitude accuracy is appropriate. 

I'll be using their charts mostly, but with my own accurate magnitudes added (for my use). Most of the charts are accurate enough for visual - apart from T Dra. The magnitude sequence on there is wild! One star is 0.6 magnitude out with other also way out.

[Mr Spock]

I realise my equipment is ideal for eclipsing binaries and stars with very low amplitudes, but they aren't my area of interest. I'm purely into pulsating variables such as SRs and Miras.

[Mr Spock]

I should add I'm taking multiple exposures, such as 2s, 8s, 30s, so I can get the variable within a magnitude range free from saturation.

[vlaiv]

6 hours ago, Mr Spock said:

I'm going to do more experiments with the kit I have to see how far I can push it. Starting with the FS-60CB - I reckon as that is f5.9, just more than a stop slower than I've been using, upping the ISO to 1000 should produce the same exposure at 30s. 355mm should enable me to get tight in on Mira stars at their minimum. Such as T Uma which is currently 12.8 and fading.

I don't think that you should apply "common knowledge" from photography to astronomy as it can be very misleading.

F/ratio does not tell the whole story even for astrophotography, let alone something like photometry.

Stellar photometry is rather interesting in this regard. Here we have two very different phenomena that compete in SNR race.

Stars provide the signal, while noise mostly comes from light pollution. Former is point source while later is extended / surface source. These two behave differently with respect to "magnification" - up to a point.

Take for example 4" F/5 scope vs 4" F/10 scope and case where we are under sampling. This means that in both F/5 scope and in F/10 scope - stellar disk is not resolved (seeing disk rather than airy disk) - and both take up roughly the same number of pixels.

Aperture does not change - so amount of photons don't change and those photons from the same star lend on roughly the same number of pixels - signal per pixel stays the same.

On the other hand - sky is extended source and changing focal length will change sampling rate or how much sky is covered by any one pixel. Using longer focal length will reduce amount of background signal in this case.

F/10 scope will have only 1/4 of background sky level for same conditions and same exposure length.

Interestingly enough, in this particular case F/10 scope is x4 faster than F/5 scope - although common daytime photography knowledge would suggest it was the other way around.

For best photometric performance - one wants for stars to be as tight as possible (occupy as few pixels as possible) - while sky background to be stretched as much possible - or occupy as much pixels per arc second squared as possible.

This happens when the seeing is good and optics is sharp. It is usually achieved with larger apertures rather than smaller as aperture also plays a part in how larger stellar profile will be.

[andrew s]

4 hours ago, vlaiv said:

For best photometric performance - one wants for stars to be as tight as possible (occupy as few pixels as possible) - while sky background to be stretched as much possible - or occupy as much pixels per arc second squared as possible.

This is not always best practice.

If you have bright stars they are often defocused to avoid saturation rather than shortening the exposure. 

In addition for optimally exposed stars fixed pattern noise dominates and can be exacerbated by switching between a few pixels if guiding is not perfect.  It can be reduced by having the star cover more pixels to average it out. 

Sky background noise can be reduced by by using a good sized annulus. 

Regards Andrew 

[Mr Spock]

A little teaser to show what I have from the D500 with Sigma 150mm f2.8, 30s @f4 ISO 400. Screenshot is from IRIS, with annotation added in PaintshopPro. This is a 1:1 crop from an image which is 5568 x 3712 pixels. Output is from the 10.01 star.

The faintest star you make out clearly is 13.6.

 

[glafnazur]

This is very interesting and something I've never thought of trying, what software are you using?

[Mr Spock]

4 hours ago, glafnazur said:

This is very interesting and something I've never thought of trying, what software are you using?

IRIS is the one I'm using - it's a bit basic but functional.

There are others out there.
APT - couldn't get it to work. It kicked up a Java error.
AIP4win - what everyone is using - asked for a serial number which I don't have.

[Mr Spock]

At the moment I've gotten my accuracy to +-0.02 mag.

[andrew s]

Astroimagej is very good and I use it for exoplanet photometry. 

Not the simplest to get into but very easy to use once set up.

Regards  Andrew 

[dph1nm]

14 hours ago, andrew s said:

Sky background noise can be reduced by by using a good sized annulus. 

Well yes and no - using a larger annulus for the sky can produce a better estimate of the mean sky under the star  (assuming the sky is flat, otherwise you are in trouble), which is important to avoid systematic photometry errors, but the photon noise from the sky, which is the dominant source of noise for faint photometry, is determined by the aperture used for the star, and so is unaffected by the size of the sky annulus. I seem to remember that an aperture of diameter ~1.4x full width half maximum is optimal in terms of s/n.

NigelM

[drjolo]

7 hours ago, Mr Spock said:

IRIS is the one I'm using - it's a bit basic but functional.

There are others out there.
APT - couldn't get it to work. It kicked up a Java error.
AIP4win - what everyone is using - asked for a serial number which I don't have.

You may also check C-Munipack https://c-munipack.sourceforge.net/

I wrote a short intro to it some time ago https://astrojolo.com/more-than-pictures/photometry-with-muniwin/

If you want to get fainter then you need to improve SNR - either by increasing aperture or by increasing exposure time. If you cannot go over 30 seconds, then you may stack images. Munipack itself allows stacking or averaging. 

 

[andrew s]

9 minutes ago, dph1nm said:

Well yes and no - using a larger annulus for the sky can produce a better estimate of the mean sky under the star  (assuming the sky is flat, otherwise you are in trouble), which is important to avoid systematic photometry errors, but the photon noise from the sky, which is the dominant source of noise for faint photometry, is determined by the aperture used for the star, and so is unaffected by the size of the sky annulus. I seem to remember that an aperture of diameter ~1.4x full width half maximum is optimal in terms of s/n.

NigelM

Yes it's a complex issue and depends on many variables and as with all things a balance is required. 

Which noise component dominates is an interesting question. Many of the "standard" answers were derived in the early CCD days and don't reflect the reality of modern CMOS detectors with their small pixels. They suffer significant telegraph noise which is not included in the standard noise equation used by, for example, astroimagej. 

I have never even seen this discussed recently other than on C Buils spectroscopy web site. It is, however mentioned in the pioneering work of Jansick but was a minor issue compared to others at the time.

Another area is scintillation noise in short exposures. 

Regards Andrew 

[Mr Spock]

Noise isn't the problem here, it's light pollution. Here's what happens with the same exposure going from ISO 200 to 1600...

The image at 1600 doesn't reveal any more stars as it brightens the background by the same amount. 1600 is noisy though so not usable.

No more stars become visible, I just get a brighter background. Stacking doesn't help as it brings the background up too. If you look at the ISO 400 image further up it has no noise - it's the camera's best SNR. What I'm trying to achieve is brightening the stars by at least two mags but not the background while retaining the same ISO.

[vlaiv]

3 hours ago, Mr Spock said:

No more stars become visible, I just get a brighter background.

Not sure if you can compare images like that.

If you for example post DSC_0690 and DSC_0693 and if their background is removed (DC offset subtraction) and we look at them at certain zoom level - one that nicely shows stars - you should be able to see the difference between the two.

I'd be happy to perform data reduction for this particular case if you post raw files.

[Mr Spock]

Test results are in:

Nikon 50mm @ f4 - mag 10.9

Nikon 90mm @ f4 - mag 11.7 - not quite perfect stars with this lens and difficult to focus manually

Sigma 150mm @ f4 - mag 13.6

FS60-CB 355mm @ f5.9 - mag 14.1 - horrible stars anywhere away from dead centre. I also spent ages trying to get this to point at the right field. Needs a finder! I also had to increase the ISO to compensate for it being f5.9 v f4 of the others.

So for now, I'll be using the 50mm and the 150mm. 13.6 isn't bad, it gets me most of the way there for Mira minimums. I'll just have the bottom of the curve missing. I'll probably need a much larger scope to reach mag15.

This was in the direction of UMa. Cas, low to the north east, is more embedded in light pollution. The 50mm could only reach 9.8, but, it's better than I expected. It looked really light and misty in that direction.

[Mr Spock]

Excerpt from IRIS with the 50mm showing (the absence of) RS and T Uma. I know T is currently 12.8 which is picked up in the 150mm; I haven't measured RS yet.