🚀 go-pugleaf

On Mon, 13 Aug 2012 04:52:24 -0700, Mark Borgerson
<mborgerson@comcast.net> wrote:

>In article <34je285htmmfscg6643s4qqlqdfjbchhbs@4ax.com>, 
>jonk@infinitefactors.org says...
>> 
>> On Sat, 11 Aug 2012 23:22:35 -0700, I wrote:
>> 
>> >On Sat, 11 Aug 2012 22:56:20 -0700, Mark Borgerson
>> ><mborgerson@comcast.net> wrote:
>> >
>> >>In article <h6h528drc8h9e7772prit4j5r3drmrlc1p@4ax.com>, 
>> >>jonk@infinitefactors.org says...
>> >>> 
>> >>> On Wed, 8 Aug 2012 05:03:53 -0700 (PDT), George Herold
>> >>> <gherold@teachspin.com> wrote:
>> >>> 
>> >>> >><snip>
>> >>> >> >Re spectrometer calibration:  We had this "black body" source. (from
>> >>> >> >Ortec?)  That could be used for intensity calibration.  PITA though.
>> >>> >>
>> >>> >> The problem for students is that they cannot afford the $800
>> >>> >> it takes to get a $8 tungsten bulb calibrated. (It can cost a
>> >>> >> lot more than that, too.) Then they need to set up a careful
>> >>> >> station for making the calibration. It's just plain too
>> >>> >> expensive to do. And I don't know a way around it. Wavelength
>> >>> >> calibration is CHEAP!! $12 buys a merc-argon calibration
>> >>> >> lamp, which fires up with the Argon first (for about a minute
>> >>> >> or two where you get some very nice Argon lines in the longer
>> >>> >> wavelengths) and then the Mercury kicks in and dominates with
>> >>> >> it's lines. A couple of well-timed snapshots and you've got
>> >>> >> enough (with software) to completely calibrate your pixel
>> >>> >> positions in a cheap ($10) megapixel camera.
>> >>> >>
>> >>> >> I'd LOVE to figure out a cheap intensity calibration setup,
>> >>> >> though. So would a lot of other people, I suppose.
>> >>> >>
>> >>> >> >But maybe you could do something on the cheap with a tungsten light
>> >>> >> >bulb.
>> >>> >>
>> >>> >> But that's the problem. I can get a cheap tungsten lamp. They
>> >>> >> are VERY CHEAP. In fact, the EXACT SAME ONE I would buy,
>> >>> >> calibrated, is dirt cheap uncalibrated. It is the calibration
>> >>> >> data that comes with a newly calibrated lamp that costs so
>> >>> >> darned much. It needs to be traceable to standards. And you
>> >>> >> pay for that. Plus, you need an expensive power supply to
>> >>> >> operate it (0.1% current control is standard.) And it ages,
>> >>> >> too. Within 100 hours or so, it's drifted enough that it
>> >>> >> needs recalibration.
>> >>> >
>> >>> >OK first a disclaimer, I know squat about the details of spectrometer
>> >>> >calibration.
>> >>> >
>> >>> >It would seem to break down into two pieces.  Absolute sensitivity (at
>> >>> >one wavelength) and then relative wavelength sensitivity.
>> >>> 
>> >>> The two calibrations I need are:
>> >>> 
>> >>> (1) Wavelength calibration which assigns a specific center
>> >>> wavelength to each pixel in the array (or matrix in the case
>> >>> of a camera.) Wavelength span is another issue, but it can be
>> >>> finessed a bit with several different alignments of the
>> >>> grating, if details are needed.
>> >>> 
>> >>> (2) Intensity calibration which assigns sensitivities to each
>> >>> pixel (no two pixels have the same sensitivities and there
>> >>> are other losses elsewhere, as well.) The whole system as a
>> >>> unit needs to be calibrated for intensity on a per pixel
>> >>> basis so that you know how much you are getting when you get
>> >>> it. This is an absolute level calibration, not a relative
>> >>> one.
>> >>> 
>> >>> Between the above two, you get accurate wavelength and
>> >>> accurate intensity. Getting sources to help calibrate
>> >>> wavelength is cheap. Getting sources (and setups) to help
>> >>> calibrate intensity is not cheap.
>> >>> 
>> >>> >For the first part you could maybe do something with a laser diode
>> >>> >into photodiode and then the same laser (though a bunch of
>> >>> >attenuators) into the spectrometer.
>> >>> 
>> >>> I'd like to hear more about this and how I might make that
>> >>> work for a megapixel camera over the entire CCD matrix. I'm
>> >>> at a loss regarding absolute calibration here. And keep in
>> >>> mind that it needs to account for the entire system's losses.
>> >>> So I can't go tickering around with the optical path to move
>> >>> the laser beam around because it will never be the same
>> >>> again. (I don't think so, anyway.) Maybe that would work with
>> >>> a monochromator? I don't have experience with those, though
>> >>> in any case it's not what I'm caring about right now.
>> >>> 
>> >>> >For the second you have the students use the black body curve and the
>> >>> >tungsten lamp.  Operate the lamp at two different temperatures.
>> >>> >(Maybe build a nice stable DC supply for the lamp.)  It seems like you
>> >>> >should be able to extract an approximate sensitvity curve from two
>> >>> >black body curves, even if you don't know the exact temperatures.
>> >>> >(But I'll leave that as an excersise for the students :^)
>> >>> >
>> >>> >How good a calibration do you need?
>> >>> 
>> >>> Good enough to do real work that is traceable to standards.
>> >>> Otherwise, it's not repeatable across instruments.
>> >>> 
>> >>> For now, the students get something useable on wavelength
>> >>> alone. And there is a lot of "science" that can get done
>> >>> without more than that. Examining fluorescence wavelengths of
>> >>> plant materials, for example. Stuff like that. They don't
>> >>> have to have intensity to replicate some of the science
>> >>> results they see in books.
>> >>> 
>> >>> But it would be very interesting to come up with a scheme
>> >>> that is cheap for intensity calibration. But like temperature
>> >>> calibration (freeze points, etc) it's not so cheap to do.
>> >>> 
>> >>Would good ol' Sol be an adequate calibration source?  You can
>> >>probably look up the daily solar intensity somewhere on the net.
>> >>Then you correct for local atmospheric effects.   Of course,
>> >>if this project has to work in Oregon, it will probably have
>> >>to be used only during summer school!   ;-)
>> >>
>> >>Oh,  and the spectral output is pretty well known and even
>> >>includes a few good marker lines.
>> >
>> >yeah. See:
>> >  http://www.google.com/patents/US5835267
>> >
>> >My intent is to only use parts that a student can afford to
>> >buy and OWN at home. Not school equipment. Not access to
>> >expensive tools. Plain, simple, stuff. I have worked out a
>> >folding piece of cardstock paper that makes the box and has
>> >all the right angles set up for them, with baffles (also out
>> >of paper.) The DVD-RW is cheap and is a precision diffraction
>> >grating. The camera is cheap (blister-pack, $10 or less, at
>> >common grocery stores.) The merc-argon bulb is $8 to $12,
>> >depending. That's the ballpark I'm talking about. They build
>> >everything themselves, take it home, and own all of it.
>> >
>> >If you come up with a scheme that can align with the sun in a
>> >consistent fashion (without burning their poor eyes out),
>> >get's good data on the sky filtering and effects due to angle
>> >of incidence, uses a stop and baffles in the optical path
>> >that children make by hand using glue and scissors and a
>> >ruler, let me know.
>> >
>> >Think 12-year old and up.
>> 
>> What I'm looking for would be analogous to using an ice bath
>> for temperature calibration. In almost every way, temperature
>> calibration is difficult. That is, everywhere except for ONE
>> temperature. And there, it's easy and cheap and any idiot can
>> do a decent job no matter how clumsy they may be. And single
>> point temperature calibration is better than nothing at all.
>> 
>> Kind of like that. It needs to be cheap and idiot proof. I'm
>> not holding my breath. But that doesn't mean I've stopped
>> looking, either. I just don't think it is in the cards, yet.
>> 
>OK.  What is the final goal as a teaching experience and how will
>the difference between 0.1%,  1%, and 10% calibration  affect
>the learning experience?
>
>1%  solar calibration should not be too difficult.  Set up
>a couple of baffles so that direct sunlight falls on a circle
>of known diameter on a piece of white card stock visible to
>the spectrometer.  The spectrometer looks at this circle of
>light and reports a result.

I think 1% will be too difficult that way. I already know how
hard that is to achieve, with good, accurate equipment and
careful procedures. If I get within 1% of NIST as it is, I
consider it very very accurate and very very lucky.

I can't say I know exactly what you are suggesting -- too
many specific possibilities arise in mind. You didn't address
any of my points. So I need to leave it there.

To do an intensity calibration, the students will have to do
a much better job on the entrance slit. I've held off of the
old 'pair of razor blades' idea. (Safety.) But I can
resurrect it if I get in mind something useful to do and it
seems practical.

>The alternative is to use your own light source and provide
>a sample of known reflectance and spectral characteristics. 
>After calibration with that standard, the student can examine
>leaves, fabrics, minerals, etc, and get a pretty good estimate
>of their spectral reflectivity.

The intensity calibration setup I have is rather large right
now and takes up a table. As I said, wavelength calibration
is NOT a problem.

We already use the sun as a source for transmission methods
in a small box. A fluorescent yellow highlighter dipped a few
times into distilled water, kept still transparent clear so
very dilute, produces a beautiful green, for example. It does
require a strong light source to be striking, so that's why
the sun.

DVD-RW is 1350 lines/mm or spaced by 0.74µm. The yellow
spectral doublet of mercury at 577nm and 579nm is very
visible to the eye and a lot of the solar Fraunhofer lines
are also clearly discernable. Which is way better than I'd
expected from cheap DVDs when I started doing this.

Anyway, I will make some thoughts first and then maybe some
attempts regarding the sun (in a few months time) and see
where it takes me. I'll have to consider safety, consistency
regardless of time of year, position of sun in sky, ... and
work out a procedure students can achieve.

Thanks,
Jon