Magnitude from photon counts

wrote:
Chris L Peterson writes:
[CCDs vs PMTs]
They are much more stable with time.

On the contrary, what matters is the entire system, not just the
detector itself. When a new speck of dust falls on the filter or
the dewar window, you've just created another artifact in the flat
field. Even though the senstivity of the detector itself could be
rock steady, the effective sensitivity has changed in the sense
that it affects the image.

Does dust not fall on PMTs altering their effective sensitivity?


Tim
--
This is not my signature.

On Thu, 27 Jan 2005 21:39:37 GMT, wrote:

You are supposing a lot of problems that are routinely and trivially
dealt with.

Not as trivial as you think.

What can I say? I've been routinely engaged in CCD photometry for over
five years. I don't find the process of collecting calibration frames
particularly challenging, and I am accustomed to residuals of a few
millimag.


Twenty six years worth of photometry. Far easier to get millimagnitude
results using a PMT.

It has been 20 years since I used a PMT for photometry. Getting good
results was certainly a challenge, far more difficult than what I do
now. I never achieved millimag accuracy back them. Maybe things have
gotten better. Of course, most observatories don't use PMTs for
photometry anymore.


particularly where absolute magnitude is at issue.

Apparent magnitude.

Confusing choice of words. I meant "absolute" as opposed to "relative".
Getting an absolute magnitude in that sense (a real number) is much
easier working with a CCD field that contains multiple reference stars
than with a single channel of CCD data where any references are
separated in time.

They have a much better signal to noise ratio,

You're repeating yourself. As I noted previously, there are noise
sources that create a floor that is difficult to get below.

Which you haven't identified. I don't consider sources of instrumental
error to be noise, and to use the term "noise" in that way is grossly
inaccurate. Noise can never be corrected for, but instrumental errors
can be reduced to an arbitrarily low level by good design (of both the
hardware and the experiment). The two sources of error are handled
completely differently in analysis.


They are much more stable with time.

On the contrary, what matters is the entire system, not just the
detector itself. When a new speck of dust falls on the filter or
the dewar window, you've just created another artifact in the flat
field. Even though the senstivity of the detector itself could be
rock steady, the effective sensitivity has changed in the sense
that it affects the image.

I must say that such a thing has never happened to me, in years of
collecting data. It is rare enough for a new piece of dust to show up on
a window in the first place. It would be extraordinarily rare for it to
just happen to show up in a location that occluded the target object, or
a reference object. After all, the majority of the pixels are never
involved in photometric calculations.

In any case, if you are happy with your results performing photometry
with PMTs, that's great, more power to you. It is the results that
matter, after all, not the tools used. Speaking for myself, I'm happy
that I'm using a CCD now, and not a PMT. With an inexpensive system I'm
able to get photometric results that would have required very expensive
and difficult to operate equipment just a few years ago.

_________________________________________________

Chris L Peterson
Cloudbait Observatory
http://www.cloudbait.com

Thanks, fair enough, it was the "adding outside the camera" that confused
me - I couldn't see the advantage over a longer exposure. The answer is
saturation avoidance.

I must say that this has been one of the most interesting ng threads I have
read in a long time. Watching you two argue the relative merits of
photomultipliers and CCDs has taught me a lot about a subject I knew zero
about, and its really quite interesting. Thankyou both for making your
discussion public.

On Fri, 28 Jan 2005 13:25:12 +1100, "Peter Webb"
wrote:

I must say that this has been one of the most interesting ng threads I have
read in a long time. Watching you two argue the relative merits of
photomultipliers and CCDs has taught me a lot about a subject I knew zero
about, and its really quite interesting. Thankyou both for making your
discussion public.

Thanks. Regardless of the relative merits of PMTs and CCDs, I can
guarantee you that both beat the heck out of the way photometry was done
a couple of hundred years ago. I recall seeing an instrument that
consisted of two telescopes that could be pointed independently, but the
light brought together in the same field. One was set up on an unknown,
and the other moved to various reference stars until the two spots
looked the same. Nobody was thinking of millimag accuracy back then!

_________________________________________________

Chris L Peterson
Cloudbait Observatory
http://www.cloudbait.com

I agree with Chris completely that the situation
is better now (with both PMTs and CCDs available
cheaply to fit any particular situation) than it
was in the old days. But there's just one little
nit-picking item I can't ignore :-)

In this exchange

particularly where absolute magnitude is at issue.

Apparent magnitude.

Confusing choice of words. I meant "absolute" as opposed to
"relative".
Getting an absolute magnitude in that sense (a real number) is much
easier working with a CCD field that contains multiple reference
stars
than with a single channel of CCD data where any references are
separated in time.

let me clarify things. Chris really is talking about
relative photometry, also called differential
photometry. Given a CCD image with two bright stars, it
is easy to measure the difference in accumulated
signal to a precision of less than one percent.
That is _not_ necessarily the ACTUAL difference
in the stars' brightness, because there can by
small, systematic errors in sensitivity from
one area of the CCD to another. Yes, I know
some people will claim that such differences can
be removed by flatfielding, but (having spent
several years recently studying this very problem),
let me assure that the simple, easy flatfielding
procedures most people use are not guaranteed
to get it right at this level. I would be
happy to go into further details if anyone wishes...

Anyway, what you CAN do very well indeed with
CCDs is to monitor this field over and over and
over again on a single night and measure
changes in the differential magnitudes of
one star relative to another (or relative to
several others). If you are very careful to
make sure that the stars don't move on the chip,
and that you spread the light out over a large
number of pixels, you can get down to the millimagnitude
level of precision in these changes. A good PMT
can do the same thing just as well (or even a bit
better), and without the lingering doubts about
the relative sensitivity, under the proper conditions.
But the CCD can measure changes in tens or hundreds of stars
all at once on a single image, which a PMT cannot do.

Sorry, I'm straying. What I wanted to say
is that "absolute photometry" is (in my neck
of the woods) a different animal entirely.
It refers to calculating the flux in physics
units (ergs per sq. cm. per second per Hertz
per steradian) based on measurements of stars
made with any sort of detector. It is really,
really, REALLY hard, and most people are content
to refer to some work by Oke and Gunn over
twenty years ago and leave it at that.

Okay, now that's out of my system :-)
Michael Richmond

Tim Auton writes:

Chris L Peterson writes:

[CCDs vs PMTs]

They are much more stable with time.

On the contrary, what matters is the entire system, not just the
detector itself. When a new speck of dust falls on the filter or
the dewar window, you've just created another artifact in the flat
field. Even though the senstivity of the detector itself could be
rock steady, the effective sensitivity has changed in the sense
that it affects the image.

Does dust not fall on PMTs altering their effective sensitivity?

Any change in effective sensitivity shows up with the next
standard star observation. CCD observers generally do not
interrupt the night to acquire new flats.

Chris L Peterson writes:

You are supposing a lot of problems that are routinely and trivially
dealt with.

Not as trivial as you think.

What can I say? I've been routinely engaged in CCD photometry for over
five years. I don't find the process of collecting calibration frames
particularly challenging, and I am accustomed to residuals of a few
millimag.

I've been doing it longer than that.

Twenty six years worth of photometry. Far easier to get millimagnitude
results using a PMT.

It has been 20 years since I used a PMT for photometry. Getting good
results was certainly a challenge, far more difficult than what I do
now. I never achieved millimag accuracy back them. Maybe things have
gotten better. Of course, most observatories don't use PMTs for
photometry anymore.

I don't find the process of collecting calibration data for a PMT
particularly challenging, and I am accustomed to residauls of a few
millimag.

particularly where absolute magnitude is at issue.

Apparent magnitude.

Confusing choice of words. I meant "absolute" as opposed to "relative".
Getting an absolute magnitude in that sense (a real number) is much
easier working with a CCD field that contains multiple reference stars
than with a single channel of CCD data where any references are
separated in time.

The time separation isn't a problem. If the transparency of the
sky is variable on that sort of time scale, then the sky isn't
photometric.

They have a much better signal to noise ratio,

You're repeating yourself. As I noted previously, there are noise
sources that create a floor that is difficult to get below.

Which you haven't identified. I don't consider sources of instrumental
error to be noise, and to use the term "noise" in that way is grossly
inaccurate. Noise can never be corrected for, but instrumental errors
can be reduced to an arbitrarily low level by good design (of both the
hardware and the experiment). The two sources of error are handled
completely differently in analysis.

On the contrary, I have identified some of the sources of noise,
and I don't care whether you choose to not call certain sources
of error "noise" or not. What matters to me is the plus or minus
you attach to a measurement. If you routinely leave out instrumental
error because it isn't "noise", then you are misleading the reader.

Furthermore, you don't have to identify all sources of noise to know
that they exist. I had a long conversation yesterday with someone
who has encountered a source of noise in CCD data, but still hasn't
identified what is causing it.

They are much more stable with time.

On the contrary, what matters is the entire system, not just the
detector itself. When a new speck of dust falls on the filter or
the dewar window, you've just created another artifact in the flat
field. Even though the senstivity of the detector itself could be
rock steady, the effective sensitivity has changed in the sense
that it affects the image.

I must say that such a thing has never happened to me, in years of
collecting data. It is rare enough for a new piece of dust to show up on
a window in the first place. It would be extraordinarily rare for it to
just happen to show up in a location that occluded the target object, or
a reference object. After all, the majority of the pixels are never
involved in photometric calculations.

Meanwhile, I've seen it happen multiple times. When the characteristic
dust doughnut shows up in processed images taken later in the night,
you know that you acquired a new dust speck since the evening flats
were taken.

In any case, if you are happy with your results performing photometry
with PMTs, that's great, more power to you. It is the results that
matter, after all, not the tools used. Speaking for myself, I'm happy
that I'm using a CCD now, and not a PMT. With an inexpensive system I'm
able to get photometric results that would have required very expensive
and difficult to operate equipment just a few years ago.

It's not just a matter of personal preference. The rest of the
community interested in the interpretation of scientific observations
should also be interested in the quality of the data on which the
interpretation is based. If one person publishes a model based on
PMT data good to a few millimag and another person publishes a model
based on CCD data good to a centimag, all other things being equal,
whose model are you going to prefer?

Stupendous Man writes:

I agree with Chris completely that the situation
is better now (with both PMTs and CCDs available
cheaply to fit any particular situation) than it
was in the old days. But there's just one little
nit-picking item I can't ignore :-)

In this exchange

particularly where absolute magnitude is at issue.

Apparent magnitude.

Confusing choice of words. I meant "absolute" as opposed to
"relative". Getting an absolute magnitude in that sense
(a real number) is much easier working with a CCD field
that contains multiple reference stars than with a single
channel of CCD data where any references are separated in time.

let me clarify things. Chris really is talking about
relative photometry, also called differential
photometry. Given a CCD image with two bright stars, it
is easy to measure the difference in accumulated
signal to a precision of less than one percent.
That is _not_ necessarily the ACTUAL difference
in the stars' brightness, because there can by
small, systematic errors in sensitivity from
one area of the CCD to another. Yes, I know
some people will claim that such differences can
be removed by flatfielding, but (having spent
several years recently studying this very problem),
let me assure that the simple, easy flatfielding
procedures most people use are not guaranteed
to get it right at this level. I would be
happy to go into further details if anyone wishes...

Well, Chris wants to see certain sources of noise identified, so it's
worth mentioning, for example, the problems that certain mechanical
shutters can create. One might be able to create the perfect flat
field calibration frame, in the sense that it accurately reflects
the varying sensitivity of the system over the entire detector.
But what if the shutter design prevent the entire focal plane from
being uniformly illuminated? If the shutter completely opens in,
for example, 0.1 sec, and your exposure is at least 100 sec long,
then illumination variations are guaranteed to be less than 0.001 mag
over the entire field, but one problem I've encounted with CCDs is
that the available standard stars in a particular photometric system
are too bright, forcing short exposures to avoid saturation, and
then you start running into questions about how the shutter
illuminates the field. It is true that certain shutter designs are
better than others at uniformly illuminating a focal plane, but some
of those designs introduce a different source of noise, namely in
the temporal domain.

Consider the Megaprime camera on CFHT. The instrument designers
apparently built the instrument with the ability to do photometry
in mind. Therefore they know how long the shutter was open, but
the shutter opening time is the better part of a full second, and
they don't know exactly when the shutter was opened. For someone
doing astrometry of a fast-moving NEO, it doesn't matter if you
can generate an astrometric solution good to 0.1 arcsec using that
image if you don't know the time associated with the exposure. And
to complicate matters, the actual midtime of the exposure depends
on where you are in the focal plane.

On Fri, 28 Jan 2005 22:00:05 GMT, wrote:

On the contrary, I have identified some of the sources of noise,
and I don't care whether you choose to not call certain sources
of error "noise" or not. What matters to me is the plus or minus
you attach to a measurement. If you routinely leave out instrumental
error because it isn't "noise", then you are misleading the reader.

Personally, I find that an unacceptable response. If you have a problem
with instrumental error, you fix it (as in the case of a badly designed
shutter). If you don't know what is noise and what is instrumental
error, you aren't in a position to analyze your results.


It's not just a matter of personal preference. The rest of the
community interested in the interpretation of scientific observations
should also be interested in the quality of the data on which the
interpretation is based. If one person publishes a model based on
PMT data good to a few millimag and another person publishes a model
based on CCD data good to a centimag, all other things being equal,
whose model are you going to prefer?

In most cases, the millimag results, of course. My point was simply that
getting millimag results with a CCD is not terribly difficult.

One point of confusion here may be the groups this is being crossposted
to. I'm talking primarily to sci.astro.amateur, but perhaps you are
aiming your comments mainly at sci.astro. I would never suggest an
amateur photometrist try to use a PMT for measurements unless he had a
very good idea what he was getting into, and a very good reason for
doing so. Photometry with a CCD is so much simpler and so much cheaper
that I wouldn't ordinarily recommend any other method.

_________________________________________________

Chris L Peterson
Cloudbait Observatory
http://www.cloudbait.com

Chris L Peterson writes:

On the contrary, I have identified some of the sources of noise,
and I don't care whether you choose to not call certain sources
of error "noise" or not. What matters to me is the plus or minus
you attach to a measurement. If you routinely leave out instrumental
error because it isn't "noise", then you are misleading the reader.

Personally, I find that an unacceptable response.

Rather ironic, considering your unacceptable responses.

If you have a problem
with instrumental error, you fix it (as in the case of a badly designed
shutter).

Exactly how do you propose that someone fix something when they don't
have the authority to do so?

If you don't know what is noise and what is instrumental
error, you aren't in a position to analyze your results.

If you don't know what to include in an error bar computation when
publishing that error bar, you aren't in a position to publish
results. Anything that contributes to the uncertainty of a result
is noise, whether it be random noise or systematic noise.

It's not just a matter of personal preference. The rest of the
community interested in the interpretation of scientific observations
should also be interested in the quality of the data on which the
interpretation is based. If one person publishes a model based on
PMT data good to a few millimag and another person publishes a model
based on CCD data good to a centimag, all other things being equal,
whose model are you going to prefer?

In most cases, the millimag results, of course. My point was simply that
getting millimag results with a CCD is not terribly difficult.

Then why isn't the available CCD photometry of Pluto-Charon mutual
events better than the available PMT photometry?

One point of confusion here may be the groups this is being crossposted
to. I'm talking primarily to sci.astro.amateur, but perhaps you are
aiming your comments mainly at sci.astro.

Why should that make a difference? I'm talking about the accuracy of
photometry done with CCDs and PMTs, which is quite independent of
whether it's done by an amateur or a professional.

I would never suggest an
amateur photometrist try to use a PMT for measurements unless he had a
very good idea what he was getting into, and a very good reason for
doing so.

The same could be said for photometry with a CCD.

Photometry with a CCD is so much simpler and so much cheaper
that I wouldn't ordinarily recommend any other method.

Why do you say it's simpler? As for expense, that's entirely driven
by market forces. The Space Shuttle would be a lot less expensive
if Boeing was making hundreds of them for transportation companies
worldwide.