Jump to content

SkySurveyBanner.jpg.21855908fce40597655603b6c9af720d.jpg

Couple of questions on spectral calibration


Recommended Posts

This one is not directly related to spectroscopy, but in some sense it is.

In this thread:

I've described problem in detail. There seems be issue with reported / measured QE graph of particular sensor, so I'm thinking of doing (crude) measurement myself. However, there are couple of issues that I need to resolve, and need help resolving them.

Idea is to use Star Analyzer to measure sensor response (low res - such as to provide QE graph) by comparing captured spectra to that of known source. I can either use known Star to do it, or known Source to do it - both present certain challenges.

First - known Star approach. Problem is of course that I'll be creating instrument response rather than sensor response with this approach. So here are questions related to that:

1. Is there any way to differentiate between atmospheric response, instrument response (in this case limited to telescope only) and sensor response. In regular instrument response all three are combined, and I don't really see any way to distinguish between them as they are multiplicative in nature.

2. Mirrors vs lens - in order to minimize impact of optical surfaces on curve, and knowing that mirror coatings have particular frequency response depending on type of coatings, would it be better to use refractive optics rather than mirror system. I don't want to rely on manufacturer provided data as this seems unreliable (at least in case of sensor). How lens fare with this - is there varying transmission curve and is wavelength dependency variation less or more compared to mirror coatings?

3. Do I really have to worry about atmospheric response and how much? I'm only interested in relative QE curve, and not absolute, and I don't care for small scale features (like isolated emission/absorption lines) - something like R100 would be plenty of resolution to characterize sensor response curve. Is there general model that can be applied for correction, and if so, what would be usual deviations in real life from this general model (based on dust / ozone / other gasses content in atmosphere on night of observation)?

If there is general model - that would be great, as in the end I might include it back into colorimetric calculations that I'm interested in (color calibration for RGB based astronomy imaging) if its contribution is significant.

Second - known source approach. This one is "simpler" in terms that I can use optic fiber (being close to point source and able to transmit light) and source of known spectra and do measurements "in doors" at relatively close range, so any atmospheric extinction is taken out of equation. Problem is of course that I don't want to fund proper broadband calibration source, and I'm thinking of DIY-ing one.

4. What would be best way to DIY light source of known spectra? I was thinking of simple incandescent / tungsten light bulb - but keeping temperature constant might be an issue. Is there way to calculate / measure temperature vs current relationship? Or measure filament temperature when it stabilizes? Or maybe do dynamic / feedback type of thing with simple controller and current (or voltage) regulator? How close is spectrum of tungsten light bulb to black body of the same temperature?

Maybe regular calibration sources don't cost fortune (I would need broadband one), does anyone know cheap (does not need to be uber precise) broadband source of known spectral distribution (bonus if it's small and operated on battery or 12V - like pocket size)?

5. Is just repeat of above question number 2. I'm thinking of using refractor for this "in doors" measurement, do I need to worry about glass transmission curve (for the level of precision I'm after)?

6. How would I do wavelength calibration indoors? Will close focusing change anything if I do wavelength calibration on a real star (hydrogen lines, two or multi point calibration) and use it indoors? I would be careful to place both sources at same position and orientation

Just few more "general" questions, related to both methods:

7. Flat fielding when working with Star Analyzer - I'm a bit confused with this one since I'm not sure how to do it - I suspect it's like regular flat fielding - I just use flat box, leave focus and everything as is and shoot flats - but I'm not 100% sure if that is the way to do it - grating would disperse complete spectra all over the chip and it would work fine?

8. Spectrum separation. Since I would like to cover 380nm - 780nm range for measurement (or even full sensitivity range in case someone else needs this sort of data) - there is issue of overlapping spectrum - even at this narrower range. 380 x 2 = 760, so 760-780 will be polluted. Actually spectrum overlap will probably come somewhere around 600nm as I believe sensor is sensitive down to about 300nm, so I'll need some sort of filtering to isolate regions of interest - and here in lies the problem. First - I'll probably need at least two filters? Second - filters don't have sharp edges and 100% transmission - so I'll add another modification to captured spectra and I would like to account for that somehow - and if at all possible without using supplied sensitivity graphs (again, I'm skeptical about accuracy of published data). If there is no other way - I'll use published data, but above mentioned case (see linked thread) - got me thinking - if I can measure things for myself - I'd rather do that then blindly rely on published specs.

Any help with the above would be appreciated!

 

Link to comment
Share on other sites

4 hours ago, wxsatuser said:

On a quick look it might well be the thing that solves most of the issues - it has a section on led light source and it's spectrum - if there is exact led manufacturer / model that is not expensive and its spectrum is known - that would enable me to do indoors experiment.

Link to comment
Share on other sites

Even if you use a standard light source with a known spectrum, Measuring QE of the sensor using  a spectrograph as you propose, separate from all other effects of the spectrograph, the grating response in particular  (and atmosphere if using a standard star as a standard source) is not practical with equipment available to the amateur. You can however compare cameras on the same system as Christian Buil did here for example (near bottom of the page)

http://www.astrosurf.com/buil/isis/noise/result.htm

Cheers

Robin

 

 

Link to comment
Share on other sites

2 minutes ago, robin_astro said:

Even if you use a standard light source with a known spectrum, Measuring QE of the sensor using  a spectrograph as you propose, separate from all other effects of the spectrograph, the grating response in particular  (and atmosphere if using a standard star as a standard source) is not practical with equipment available to the amateur. You can however compare cameras on the same system as Christian Buil did here for example (near bottom of the page)

http://www.astrosurf.com/buil/isis/noise/result.htm

Cheers

Robin

 

 

Even if I need relative QE curve, so I'm not interested in absolute QE curve, nor max QE value?

No way to take multiple measurements and subtract something from something to get sensor response :D ?

 

Link to comment
Share on other sites

@robin_astro

On a separate note and related to Buil's page you linked:

These graphs look very much like published data. There might be slight variations, but overall shape (when scaled, since original graphs go to 1000nm, and Buil's measurement up to 800nm) is about the same.

If you take a look at thread I linked in the first post - ASI1600 published graph and Buil's measurement are quite a bit different, even after my attempt of correction. I wonder why that might be so?

Link to comment
Share on other sites

Maybe you could use the same continuous spectrum source (like halogen lamp) and use second camera with well known QE to do measurement. Take spectra with reference sensor, then with 1600 and you should be able to get proper data.

PS - ups, it was already said :) 

Link to comment
Share on other sites

Just now, drjolo said:

Maybe you could use the same continuous spectrum source (like halogen lamp) and use second camera with well known QE to do measurement. Take spectra with reference sensor, then with 1600 and you should be able to get proper data.

Yes that is what Buil did.  You need a camera with a known response to compare with though

Buil's measurements were done independently of the manufacturer's published sensor figures and do not always agree. For example the cameras have a protective glass which also has a filtering effect

Link to comment
Share on other sites

2 minutes ago, robin_astro said:

The masters thesis above also uses a similar technique but a monochromator is used in place of the spectrograph and a calibrated silicon diode in place of the reference camera.

What sensor do you want to measure? Is there not a published QE curve for it?

Robin

Just check the thread at the top of the page in my first post - I linked original thread in camera discussion forum - its ASI1600, so Panasonic MN34230 sensor.

There is published graph - one and the same, all over the internet. I can't find any such info on Panasonic website, and have no clue what is the source of that published graph.

Link to comment
Share on other sites

11 minutes ago, drjolo said:

Maybe you could use the same continuous spectrum source (like halogen lamp) and use second camera with well known QE to do measurement. Take spectra with reference sensor, then with 1600 and you should be able to get proper data.

PS - ups, it was already said :) 

This is excellent idea - I did not think of that in the way you are proposing it. I just need a stable source - led, or even incandescent light source if stable (both need to get to stable regime I guess, so 10 minutes or so from power up) and I can measure with two different cameras - ASI185 and ASI1600.

All the things in the setup would be the same except for sensors. From first measurement I can get combined light source / instrument (scope + SA200) response and use it on second measurement! Great!

Except of course if Sony published graphs for ASI185 sensor are flawed as well :D (see paranoia develop there? :D )

Link to comment
Share on other sites

22 minutes ago, vlaiv said:

@robin_astro

On a separate note and related to Buil's page you linked:

These graphs look very much like published data. There might be slight variations, but overall shape (when scaled, since original graphs go to 1000nm, and Buil's measurement up to 800nm) is about the same.

If you take a look at thread I linked in the first post - ASI1600 published graph and Buil's measurement are quite a bit different, even after my attempt of correction. I wonder why that might be so?

Have you asked Christian Buil ?

Robin

Link to comment
Share on other sites

The

Here is the full description of the technique Buil used, from his website. Note his comment  that the response of a complete camera  may not be the same as the sensor because of the protective glass

quantum efficiency of Sony ICX285 and ICX694 CCD devices is calculated from comparison with observed signal on a KAF-3200ME CCD camera (QSI-532) and a KAF-8300 CCD camera (QSI-583). The KAF-3200ME and KAF-8300 QE are assumed to be known and exact. The measurement is madewith a spectrograph LISA. The formula used for extract quantum efficiency is:

formula3.png

Note: the computed effciency is the "apparent QE" for a given camera (camera glass windows and detector coverglass optical transmission impact the result).

 

Link to comment
Share on other sites

17 minutes ago, robin_astro said:

The

Here is the full description of the technique Buil used, from his website. Note his comment  that the response of a complete camera  may not be the same as the sensor because of the protective glass

quantum efficiency of Sony ICX285 and ICX694 CCD devices is calculated from comparison with observed signal on a KAF-3200ME CCD camera (QSI-532) and a KAF-8300 CCD camera (QSI-583). The KAF-3200ME and KAF-8300 QE are assumed to be known and exact. The measurement is madewith a spectrograph LISA. The formula used for extract quantum efficiency is:

formula3.png

Note: the computed effciency is the "apparent QE" for a given camera (camera glass windows and detector coverglass optical transmission impact the result).

 

Useful info, especially about sensor cover / camera cover glass. This can be nicely seen if we compare his measurement with published sensor data:

image.png.1c736e567f9bf9f7bbfd77522a4a783e.png

Red - published, blue - measured (for KAF8300) - That is something I would expect from glass cover - very small percent of light loss.

Link to comment
Share on other sites

Yes the camera protective window/sensor cover glass response should be reasonably flat in the visible range (Though could affect the absolute value significantly if not AR coated, which some are not) The effect in the UV is more obvious eg Buil's measurements here

http://www.spectro-aras.com/forum/viewtopic.php?f=45&t=2085&start=30

 

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. By using this site, you agree to our Terms of Use.