Network of Small Telescopes Discovers Distant Planet Orbiting Another Star

http://www.lowell.edu/press_room/rel...rES-1_rls.html

For Immediate Release

Lowell Observatory
August 24, 2004

This is a joint announcement from the Astrophysical Institute of the
Canaries (IAC), National Center for Atmospheric Research (NCAR),
Harvard-Smithsonian Center for Astrophysics (CfA), Lowell Observatory,
and California Institute of Technology.

Note to Editors: High-resolution artwork and animation of the newly
discovered planet TrES-1 is posted online at
http://www.lowell.edu/press_room/TrES-1_images.html

Network of Small Telescopes Discovers Distant Planet

Flagstaff, AZ - Fifteen years ago, the largest telescopes in the world had
yet to locate a planet orbiting another star. Today telescopes no larger
than those available in department stores are proving capable of
spotting previously unknown worlds. A newfound planet detected by a
small, 4-inch-diameter telescope demonstrates that we are at the cusp of
a new age of planet discovery. Soon, new worlds may be located at an
accelerating pace, bringing the detection of the first Earth-sized world
one step closer.

"This discovery demonstrates that even humble telescopes can make huge
contributions to planet searches," says Guillermo Torres of the
Harvard-Smithsonian Center for Astrophysics (CfA), a co-author on the
study.

This is the first extrasolar planet discovery made by a dedicated survey
of many thousands of relatively bright stars in large regions of the
sky. It was made using the Trans-Atlantic Exoplanet Survey (TrES), a
network of small, relatively inexpensive telescopes designed to look
specifically for planets orbiting bright stars. A team of scientists
co-led by Edward Dunham of Lowell Observatory, Timothy Brown of NCAR,
and David Charbonneau (CfA), developed the TrES network. The network's
telescopes are located in Palomar Observatory (California, USA), Lowell
Observatory (Arizona, USA), and the Canary Islands (Spain).

"The advantage of working as a network is that we can 'stretch the
night' and monitor our fields for a longer time, increasing our chance
of discovering a planet," says Georgi Mandushev (Lowell Observatory), a
co-author of the paper.

This research study will be posted online at
http://arxiv.org/abs/astro-ph/0408421 and will appear in an upcoming
issue of The Astrophysical Journal Letters.

"It took several Ph.D. scientists working full-time to develop the data
analysis methods for this search program, but the equipment itself uses
simple, off-the-shelf components," says co-author David Charbonneau
(CfA/Caltech).

Although the small telescopes of the TrES network made the initial
discovery, follow-up observations at other facilities were required.
Observations at the W. M. Keck Observatory which operates the world's
two largest telescopes in Hawaii for the University of California,
Caltech, and NASA, were particularly crucial in confirming the planet's
existence.

Planet Shadows

The newfound planet is a Jupiter-sized gas giant orbiting a star located
about 500 light years from the Earth in the constellation Lyra. This
world circles its star every 3.03 days at a distance of only 4 million
miles (6 million kilometers), much closer and faster than the planet
Mercury in our solar system.

Although such planets are relatively common, astronomers used an
uncommon technique to discover it. This world was found by the "transit
method," which looks for a dip in a star's brightness when a planet
crosses directly in front of the star and casts a shadow. A
Jupiter-sized planet blocks only about 1/100th of the light from a
Sun-like star, but that is enough to make it detectable.

"This Jupiter-sized planet was observed doing the same thing that
happened in June when Venus moved across (or transited) the face of our
Sun," says Mandushev. "The difference is that this planet is outside our
solar system, roughly 500 light years away."

To be successful, transit searches must examine many stars because we
only see a transit if a planetary system is located nearly edge-on to
our line of sight. A number of different transit searches currently are
underway. Most examine limited areas of the sky and focus on fainter
stars because they are more common, thereby increasing the chances of
finding a transiting system. However the TrES network concentrates on
searching brighter stars in larger swaths of the sky because planets
orbiting bright stars are easier to study directly.

"All that we have to work with is the light that comes from the star,"
says Tim Brown (NCAR), a study co-author. "It's much harder to learn
anything when the stars are faint."

Most known extrasolar planets were found using the "Doppler method,"
which detects a planet's gravitational effect on its star by looking for
shifts in the star's spectrum, or rainbow of colors. However, the
information that can be gleaned about a planet using the Doppler method
is limited. For example, only a lower limit to the mass can be
determined because the angle at which we view the system is unknown. A
high-mass brown dwarf whose orbit is highly inclined to our line of
sight produces the same signal as a low-mass planet that is nearly edge-on.

"When astronomers find a transiting planet, we know that its orbit is
essentially edge-on, so we can calculate its exact mass. From the amount
of light it blocks, we learn its physical size. In one instance, we've
even been able to detect and study a giant planet's atmosphere," says
Charbonneau.

Sorting Suspects

The TrES survey examined approximately 12,000 stars in 36 square degrees
of the sky (about half of the size of the bowl of the Big Dipper) in the
constellation of Lyra. Roi Alonso (IAC), a graduate student of Brown's,
identified 16 possible candidates for planet transits. "The TrES survey
gave us our initial line-up of suspects. Then, we had to make a lot of
follow-up observations to eliminate the imposters," says co-author
Alessandro Sozzetti (University of Pittsburgh/CfA).

After compiling the list of candidates in late April, the researchers
used telescopes at CfA's Whipple Observatory in Arizona, Oak Ridge
Observatory in Massachusetts, and Lowell Observatory in Arizona to
obtain additional photometric (brightness) observations, as well as
spectroscopic observations that eliminated eclipsing binary stars.

In a matter of two month's time, the team had zeroed in on the most
promising candidate. High-resolution spectroscopic observations by
Torres and Sozzetti using time provided by NASA on the 10-meter-diameter
Keck I telescope in Hawaii clinched the case.

"Without this follow-up work the photometric surveys can't tell which of
their candidates are actually planets. The proof of the pudding is a
spectroscopic orbit for the parent star. That's why the Keck
observations of this star were so important in proving that we had found
a true planetary system," says co-author David Latham (CfA).

Remarkably Normal

The planet, called TrES-1, is much like Jupiter in mass and size. It is
likely to be a gas giant composed primarily of hydrogen and helium, the
most common elements in the Universe. But unlike Jupiter, it orbits very
close to its star, giving it a temperature of around 1500 degrees F.

Astronomers are particularly interested in TrES-1 because its structure
agrees so well with theory, in contrast to the first discovered
transiting planet, HD 209458b. The latter world contains about the same
mass as TrES-1, yet is around 30% larger in size. Even its proximity to
its star and the accompanying heat don't explain such a large size.

"Finding TrES-1 and seeing how normal it is makes us suspect that HD
209458b is an 'oddball' planet," says Charbonneau.

TrES-1 orbits its star every 72 hours, placing it among a group of
similar planets known as "hot Jupiters." Such worlds likely formed much
further away from their stars and then migrated inward, sweeping away
any other planets in the process. The many planetary systems found to
contain hot Jupiters indicate that our solar system may be unusual for
its relatively quiet history.

Both the close orbit of TrES-1 and its migration history make it
unlikely to possess any moons or rings. Nevertheless, astronomers will
continue to examine this system closely because precise photometric
observations may detect moons or rings if they exist. In addition,
detailed spectroscopic observations may give clues to the presence and
composition of the planet's atmosphere.

The paper, "TrES-1: The Transiting Planet of a Bright K0V Star,"
descibing these results is authored by: Roi Alonso (IAC); Timothy M.
Brown (NCAR); Guillermo Torres and David W. Latham (CfA); Alessandro
Sozzetti (University of Pittsburgh/CfA); Georgi Mandushev (Lowell
Observatory), Juan A. Belmonte (IAC); David Charbonneau (CfA/Caltech);
Hans J. Deeg (IAC); Edward W. Dunham (Lowell Observatory); Francis T.
O'Donovan (Caltech); and Robert Stefanik (CfA).

The W.M. Keck Observatory is operated by the California Association for
Research in Astronomy, a scientific partnership of the California
Institute of Technology, the University of California, and the National
Aeronautics and Space Administration (NASA).

Funding for the research that led to this planet's discovery was
provided by NASA's Origins of Solar Systems Program.

Founded in 1894, Lowell Observatory pursues the study of astronomy,
conducts pure research in astronomical phenomena, and maintains quality
public education and outreach programs.

#END#

contact: Steele Wotkyns
Public Relations Manager
(928) 233-3232

www.lowell.edu

For additional information:
This research study, "TrES-1: The Transiting Planet of a Bright K0V
Star," will be posted online at http://arxiv.org/abs/astro-ph/0408421
and will appear in an
upcoming issue of The Astrophysical Journal Letters.
High-resolution artwork and animation of the newly discovered planet
TrES-1 is online at
http://www.lowell.edu/press_room/TrES-1_images.html

Harvard-Smithsonian Center for Astrophysics press release
http://cfa-www.harvard.edu/ep/pressrel.html

National Center for Atmospheric Research press release
http://www.ucar.edu/news/releases/

Astrophysical Institute of the Canaries press release
http://www.iac.es/gabinete/noticias/noticias.htm

Thanks!
Very interesting information.
---Mac
**************************
On 24 Aug 2004 09:57:19 -0700, (Ron) wrote:
http://www.lowell.edu/press_room/rel...rES-1_rls.html
For Immediate Release
Lowell Observatory
August 24, 2004

This is a joint announcement from the Astrophysical Institute of the
Canaries (IAC), National Center for Atmospheric Research (NCAR),
Harvard-Smithsonian Center for Astrophysics (CfA), Lowell Observatory,
and California Institute of Technology.

Note to Editors: High-resolution artwork and animation of the newly
discovered planet TrES-1 is posted online at
http://www.lowell.edu/press_room/TrES-1_images.html

Network of Small Telescopes Discovers Distant Planet

Flagstaff, AZ - Fifteen years ago, the largest telescopes in the world had
yet to locate a planet orbiting another star. Today telescopes no larger
than those available in department stores are proving capable of
spotting previously unknown worlds. A newfound planet detected by a
small, 4-inch-diameter telescope demonstrates that we are at the cusp of
a new age of planet discovery. Soon, new worlds may be located at an
accelerating pace, bringing the detection of the first Earth-sized world
one step closer.
SNIP SNIP
******************************

It is certainly inspiring to learn that one can discover extrasolar
planets using a telescope with a roughly 4 inch aperture. I think I
understand why three different telescopes at three different sites were
needed to discover it, since they had to sift through so many candidates
and this helped with the process of weeding out false alarms. However,
now that it has been discovered and its discovery confirmed, what are
the difficulties one would face in using a beginner's telescope, say
one of the $200 computer controlled models from Mead, to look at the
star in question and confirm the observations oneself? That seems like
a more tractable project than discovering it or proving beyond a shadow
of a doubt that it is correct.

I looked at the article of Torres et al and didn't find as much detail
as I hoped for about the light gathering equipment and analytical techniques.
I think the basic reference for the equipment was Latham 1992. Is there
some kind of standard attachment one can add to the, say, Mead mentioned
above that is adequate to collect the light and send the information to one's
laptop for analysis?

It is nice to know it was done with a small telescope, but it would be nicer
to know that all the equipment one needs to duplicate the observation and
analysis could be equally humble.
--
Ignorantly,
Allan Adler
* Disclaimer: I am a guest and *not* a member of the MIT CSAIL. My actions and
* comments do not reflect in any way on MIT. Also, I am nowhere near Boston.

Allan Adler wrote in message ...
It is certainly inspiring to learn that one can discover extrasolar
planets using a telescope with a roughly 4 inch aperture. I think I
understand why three different telescopes at three different sites were
needed to discover it, since they had to sift through so many candidates
and this helped with the process of weeding out false alarms. However,
now that it has been discovered and its discovery confirmed, what are
the difficulties one would face in using a beginner's telescope, say
one of the $200 computer controlled models from Mead, to look at the
star in question and confirm the observations oneself? That seems like
a more tractable project than discovering it or proving beyond a shadow
of a doubt that it is correct.

I looked at the article of Torres et al and didn't find as much detail
as I hoped for about the light gathering equipment and analytical techniques.
I think the basic reference for the equipment was Latham 1992. Is there
some kind of standard attachment one can add to the, say, Mead mentioned
above that is adequate to collect the light and send the information to one's
laptop for analysis?

It is nice to know it was done with a small telescope, but it would be nicer
to know that all the equipment one needs to duplicate the observation and
analysis could be equally humble.

Hi, Allan,

I'm not much more of an expert on this subject than you are, but what
the heck. Sci.astro desperately needs an increase in its signal:noise
ratio.

There are amateurs observing known extrasolar planetary occulations.
You can find out more about them and their work at the American
Association of Variable Star Observers (http://www.aavso.org). If you
want to look for a *known* exoplanet, you stand a decent chance of
finding it.

I have a friend who is a member of this organization. He owns a Meade
8" Schmidt-Cassegrain reflector, and a hand-made CCD camera which saw
its first light a few months ago. I work with microscopes more than
telescopes. Still, many of the issues surrounding getting a good
quantitative image are the same.

You won't see these exoplanet transits by eye. Only a few extrasolar
planets have been observed by occultation so far. When these planets
pass in front of their parent stars, the light loss is pretty small,
peaking at around 2%. So you need to make really accurate
measurements of the intensity. Twinkling and other atmospheric
variations are a problem. The pixels on a CCD are not perfectly
uniform, either. How sharp is your focus? Is the light of your star
falling exactly on one pixel, or on several? What if the voltage that
you supply to the CCD varies a bit from time to time? Then,
successive images of the star would not be directly comparable. Have
you saturated any pixels? Is your CCD response linear? Is your
analog to digital conversion 8-bit or 12-bit?

To compensate for all of these possible problems, you would probably
want to image a star field that includes at least a few reference
stars that you do not expect to vary. You would want to take many
images, at a few different exposure times. Then you would need to do
a fair amount of math to tease out the variations as a function of
time.

I suspect that the use of three observing sites in the TReS study
improved the observations in at least three ways. First, one site
would often be able to observe when another was clouded out. Second,
the Canary Islands site and the Western U.S. sites were several time
zones apart, allowing almost 24-hour observations. Third, there would
be times of overlap, when light curves from multiple observing sites
could be compared.

So, can you go hunting for NEW expolanets yourself? Maybe. But
having a friend on another continent or two would help. And the
software to analyze the images is critical.

(Proposal for an amateur exoplanet hunting network: observers in
California, Chile, Canary Islands or Spain, South Africa, Japan, and
Australia.)

--
Rainforest laid low.
"Wake up and smell the ozone,"
Says man with chainsaw.
John J. Ladasky Jr., Ph.D.

(John Ladasky) writes:

There are amateurs observing known extrasolar planetary occulations.
You can find out more about them and their work at the American
Association of Variable Star Observers (http://www.aavso.org). If you
want to look for a *known* exoplanet, you stand a decent chance of
finding it.

Thanks for the pointer. It looks very interesting.

I have a friend who is a member of this organization. He owns a Meade
8" Schmidt-Cassegrain reflector, and a hand-made CCD camera which saw
its first light a few months ago. I work with microscopes more than
telescopes. Still, many of the issues surrounding getting a good
quantitative image are the same.

I didn't know one could make one's own CCD camera. Is that more expensive
than buying one?

When these planets
pass in front of their parent stars, the light loss is pretty small,
peaking at around 2%. So you need to make really accurate
measurements of the intensity. Twinkling and other atmospheric
variations are a problem. The pixels on a CCD are not perfectly
uniform, either. How sharp is your focus? Is the light of your star
falling exactly on one pixel, or on several? What if the voltage that
you supply to the CCD varies a bit from time to time? Then,
successive images of the star would not be directly comparable. Have
you saturated any pixels? Is your CCD response linear? Is your
analog to digital conversion 8-bit or 12-bit?

Presumably one also uses suitable software to analyze the light
falling on the CCD. Apart from spectral analysis of the light, it
seems that the software would be designed to deal with these issues.
At any rate, Torres et al used CfA Digital Speedometers (whatever they
are) and compared their "observed spectra with synthetic spectra calculated
by J. Morse using Kurucz models (Morse & Kurucz, private communication)"
(whatever that means). I'm just referring to stuff done with the little
scope. Their photometric and radial velocity data (on a big scope?)
are supposed to be at:
http://www.hao.ucar.edu/public/resea...data/TrES1.asc
They didn't say anything about pixels or CCD cameras.

I just did a google search for CfA Digital Speedometers. CfA apparently
stands for "Center for Astrophysics". Then I went to
http://adsabs.harvard.edu
and searched for digital speedometer in the abstracts. The
earlilests reference so far involving the CfA is in the Bulletin of the
American Astronomical Society, vol.14, p.82, and I'm now downloading it.
Since it is so specialized to the CfA, I gather that one can't simply order
the equivalent from a catalogue.

I suspect that the use of three observing sites in the TReS study
improved the observations in at least three ways. First, one site
would often be able to observe when another was clouded out. Second,
the Canary Islands site and the Western U.S. sites were several time
zones apart, allowing almost 24-hour observations. Third, there would
be times of overlap, when light curves from multiple observing sites
could be compared.

One of the special features of this observation, according go the article,
is the fact that the exosolar planet takes 3.03 days to go around the star.
Apparently, the fact this is so close to an integral number of days
placed severe constraints on the places where one could observe the transits.

So, can you go hunting for NEW expolanets yourself? Maybe. But
having a friend on another continent or two would help. And the
software to analyze the images is critical.

I have no budget for astronomy and don't even own a scope. I have
an old pair of 10x50 binoculars and no mount for them. I rely on
friends who have telescopes to do any observing, by looking through
their scopes when they have them set up. However, I try to inform myself
about what things cost and at what point they become feasible, just
so that if I ever have any kind of budget for astronomy, I'll know what
is and what is not within that budget.

The CfA digital speedometers sound like they wouldn't be. So it's
good to know about the viability of CCD cameras for planet hunting.
--
Ignorantly,
Allan Adler
* Disclaimer: I am a guest and *not* a member of the MIT CSAIL. My actions and
* comments do not reflect in any way on MIT. Also, I am nowhere near Boston.

"John Ladasky" wrote in message
om...

... When these planets
pass in front of their parent stars, the light loss is pretty small,
peaking at around 2%. So you need to make really accurate
measurements of the intensity. Twinkling and other atmospheric
variations are a problem. The pixels on a CCD are not perfectly
uniform, either. How sharp is your focus? Is the light of your star
falling exactly on one pixel, or on several?

For this work, wouldn't it be better to use something
with a larger active area, a photodiode for example?
The focus shouldn't matter as long as all the light
falls on the detector.

I guess twinkling is more difficult since it can draw
in light from a wider effective aperture so the post-
processing would need to take care of this.

What if the voltage that
you supply to the CCD varies a bit from time to time? Then,
successive images of the star would not be directly comparable. Have
you saturated any pixels? Is your CCD response linear? Is your
analog to digital conversion 8-bit or 12-bit?

Are the rates low enough to do photon counting from
a photodiode? Counting events should be fairly
resistant to bias voltage variation and the linearity
and quantisation problems would be reduced unless
your bandwidth was low enough to get multiple photons
seen as individual events at a significant rate.

... Then you would need to do
a fair amount of math to tease out the variations as a function of
time.

Presumably the prime part of any processing would be
a Fourier transform and a major problem is the limited
and irregular observing times.

What am I missing?

George

Allan Adler wrote in message ...
(John Ladasky) writes:

There are amateurs observing known extrasolar planetary occulations.
You can find out more about them and their work at the American
Association of Variable Star Observers (http://www.aavso.org). If you
want to look for a *known* exoplanet, you stand a decent chance of
finding it.

Thanks for the pointer. It looks very interesting.

I have a friend who is a member of this organization. He owns a Meade
8" Schmidt-Cassegrain reflector, and a hand-made CCD camera which saw
its first light a few months ago. I work with microscopes more than
telescopes. Still, many of the issues surrounding getting a good
quantitative image are the same.

I didn't know one could make one's own CCD camera. Is that more expensive
than buying one?

Perhaps, but you won't be able to do much stellar photometry with an
off-the-shelf digital camera. The OTS digicams use decent CCD chips,
but there are others out there that are larger, and can gather more
light, if you are willing to pay. Also, the digicam CCD chips have
patterned RGB color masks in front of the pixels. What this means is
that in any one color range, only 1/3 of the chip is actually
receiving light. For some photometry work, you want to capture every
photon. The RGB chips throw 2/3 of them away. Finally, there's the
issue of thermal noise. A cold camera generates less background
signal. Consumer digicams aren't actively cooled.

My friend's custom rig uses a high-sensitivity CCD chip from Kodak,
one that doesn't have the color masks. He added an external color
filter wheel, for those rare times when he actually might want to
exclude certain colors, and a Peltier cooling device. Can you buy a
camera like this? It's similar in many ways to the cameras we use for
microscopes. We certainly buy those. But they'll cost a lot more
than your 10 X 50 binocs.

When these planets
pass in front of their parent stars, the light loss is pretty small,
peaking at around 2%. So you need to make really accurate
measurements of the intensity. Twinkling and other atmospheric
variations are a problem. The pixels on a CCD are not perfectly
uniform, either. How sharp is your focus? Is the light of your star
falling exactly on one pixel, or on several? What if the voltage that
you supply to the CCD varies a bit from time to time? Then,
successive images of the star would not be directly comparable. Have
you saturated any pixels? Is your CCD response linear? Is your
analog to digital conversion 8-bit or 12-bit?

Presumably one also uses suitable software to analyze the light
falling on the CCD. Apart from spectral analysis of the light, it
seems that the software would be designed to deal with these issues.
At any rate, Torres et al used CfA Digital Speedometers (whatever they
are) and compared their "observed spectra with synthetic spectra calculated
by J. Morse using Kurucz models (Morse & Kurucz, private communication)"
(whatever that means). I'm just referring to stuff done with the little
scope. Their photometric and radial velocity data (on a big scope?)
are supposed to be at:
http://www.hao.ucar.edu/public/resea...data/TrES1.asc
They didn't say anything about pixels or CCD cameras.

O.K., you're jumping to the second part of the TrES project -- looking
at Doppler velocity changes. Once you see a periodic, small change in
a star's light curve, you can't be SURE that it's due to a planet.
Suppose that you have two stars of almost equal intensity eclipsing
each other? Or a periodic, variable star? How can you distinguish
these possibilities from a planet?

This is what the radial velocity study will tell you. You can tell
whether a star is moving towards you or away from you by looking at
the blue-shifting and red-shifting of the star's light. A solitary,
variable star is not expected to move back and forth. Two stars
orbiting each other will fling each other back and forth hard -- the
velocity can change by tens of km/sec over the orbital period. A
planet will tug on its parent star fairly gently, resulting in
velocity changes which generally won't exceed 1 km/sec.

Velocity measurements are taken with spectrographs, rather than
imaging cameras. That's why you aren't seeing references to CCD's and
pixels in that part of the report.

There is a VERY dedicated group of amateurs trying to do Doppler
velocimetry:

http://www.spectrashift.com/

But take a look at their work... thirty years ago, this project would
have been worthy of an NSF grant!

I just did a google search for CfA Digital Speedometers. CfA apparently
stands for "Center for Astrophysics". Then I went to
http://adsabs.harvard.edu
and searched for digital speedometer in the abstracts. The
earlilests reference so far involving the CfA is in the Bulletin of the
American Astronomical Society, vol.14, p.82, and I'm now downloading it.
Since it is so specialized to the CfA, I gather that one can't simply order
the equivalent from a catalogue.

I haven't followed your link, but I'm guessing that the "digital
speedometer" is probably the spectrograph that they use to reference
atomic absorption lines in the star's spectrum against a laboratory
spectrum reference (like an arc lamp).

I suspect that the use of three observing sites in the TReS study
improved the observations in at least three ways. First, one site
would often be able to observe when another was clouded out. Second,
the Canary Islands site and the Western U.S. sites were several time
zones apart, allowing almost 24-hour observations. Third, there would
be times of overlap, when light curves from multiple observing sites
could be compared.

One of the special features of this observation, according go the article,
is the fact that the exosolar planet takes 3.03 days to go around the star.
Apparently, the fact this is so close to an integral number of days
placed severe constraints on the places where one could observe the transits.

Some other planets will eventually be found that have more
accomodating periods, and thus can be seen more readily from all the
sites.

So, can you go hunting for NEW exoplanets yourself? Maybe. But
having a friend on another continent or two would help. And the
software to analyze the images is critical.

I have no budget for astronomy and don't even own a scope. I have
an old pair of 10x50 binoculars and no mount for them. I rely on
friends who have telescopes to do any observing, by looking through
their scopes when they have them set up. However, I try to inform myself
about what things cost and at what point they become feasible, just
so that if I ever have any kind of budget for astronomy, I'll know what
is and what is not within that budget.

The CfA digital speedometers sound like they wouldn't be. So it's
good to know about the viability of CCD cameras for planet hunting.

And now you also know that variation in the light intensity of a star
isn't enough, by itself, to be sure that you have seen a planetary
transit.

Have fun. Astronomy is addictive!

--
Rainforest laid low.
"Wake up and smell the ozone,"
Says man with chainsaw.
John J. Ladasky Jr., Ph.D.

Hi, George,

As I mentioned in my message to Allan Adler, I work with microscopes
rather than telescopes. So I apologize in advance if I get in over my
head...

"George Dishman" wrote in message ...
"John Ladasky" wrote in message
om...

... When these planets
pass in front of their parent stars, the light loss is pretty small,
peaking at around 2%. So you need to make really accurate
measurements of the intensity. Twinkling and other atmospheric
variations are a problem. The pixels on a CCD are not perfectly
uniform, either. How sharp is your focus? Is the light of your star
falling exactly on one pixel, or on several?

For this work, wouldn't it be better to use something
with a larger active area, a photodiode for example?
The focus shouldn't matter as long as all the light
falls on the detector.

I guess twinkling is more difficult since it can draw
in light from a wider effective aperture so the post-
processing would need to take care of this.

What if the voltage that
you supply to the CCD varies a bit from time to time? Then,
successive images of the star would not be directly comparable. Have
you saturated any pixels? Is your CCD response linear? Is your
analog to digital conversion 8-bit or 12-bit?

Are the rates low enough to do photon counting from
a photodiode? Counting events should be fairly
resistant to bias voltage variation and the linearity
and quantisation problems would be reduced unless
your bandwidth was low enough to get multiple photons
seen as individual events at a significant rate.

It has been ages since I've thought about operating a light-collection
system in single-photon mode, using either a photodiode or a
photo-multiplier tube. But you're right, this sounds like a way to
get around some of the problems a CCD creates. Why isn't it done this
way? Good question. I don't have a definitive answer. I'll guess.

My variable-star observing friend initially built a PMT housing for
the back of his telescope. A major hassle in the system was that, in
order to make calibrated measurements, you had to steer the scope back
and forth repeatedly between two stars. You say that a photodiode
isn't sensitive to position? A PMT certainly is. Lining up a star
the same way twice, so that you can compare successive PMT
measurements, will make you turn gray before your time. (Or would
have, anyway -- I come from the age before GOTO telescopes...)

Another tradeoff is in discriminating your star from its neighbors.
CCD pixels are small, so you can draw a really tight perimeter around
your star. A large photodiode might also gather light from dim,
unwanted sources adjacent to your star.

Another practical concern, at least for people who aren't doing this
for a living, is that CCD's can be used for other types of astronomy
when you aren't doing photometry. A photodiode will be used for just
one purpose, and it requires special hardware, too (a high-voltage
power supply, and a counter, if you're operating in single-photon mode
as you suggested).

Finally, the CCD allows you to perform photometry on many stars at
once, rather than just one at a time. In the case of TReS, the
specific short-period planetary transits that they wanted to find are
pretty rare, occurring perhaps in only one in several thousand stars.

... Then you would need to do
a fair amount of math to tease out the variations as a function of
time.

Presumably the prime part of any processing would be
a Fourier transform and a major problem is the limited
and irregular observing times.

What am I missing?

You've jumped to the end. You assume that you have an accurate light
curve for your star, and you can look for variations over time with a
Fourier transform.

Before you can do that, you would need to know that you accurately
gathered all the light from your star, even if that light falls on
several adjacent pixels. You need to subtract your background
accurately. You need to know how the pixels you used in that image
compare to others on the CCD, because there's no way that you'll image
your star onto the same pixels tomorrow might. (Or maybe even your
next image, if you bump into your scope.) You need to correct for
atmospheric conditions in successive images, which means comparing
your star to nearby reference stars. If the references differ much in
intensity from your test star (8 bit A/D will not even cover one
magnitude of brightness, if you want to get 2% accuracy for the
intensities of all of your stars), you'll need to take exposures of
varying lengths. You need to know whether and how the CCD output
deviates from linearity, if you want to compare values from two images
with different exposure times.

Whew! Was that enough?

--
Rainforest laid low.
"Wake up and smell the ozone,"
Says man with chainsaw.
John J. Ladasky Jr., Ph.D.

Thanks for answering my questions. I looked up the website of the
amateur spectroscopers and I'll read more of it later. Regarding the
home made CCD camera, where would one read detailed instructions on
how to do that? I like to read detailed instructions on how to do things,
even if I lack the skill or resources to actually do them.
--
Ignorantly,
Allan Adler
* Disclaimer: I am a guest and *not* a member of the MIT CSAIL. My actions and
* comments do not reflect in any way on MIT. Also, I am nowhere near Boston.

"John Ladasky" wrote in message
om...
Hi, George,

As I mentioned in my message to Allan Adler, I work with microscopes
rather than telescopes. So I apologize in advance if I get in over my
head...

That's ok, I'm designing a control system to drive a
train so I certainly know less than you. Regardless,
it's definitely raising the SNR!

Are the rates low enough to do photon counting from
a photodiode? Counting events should be fairly
resistant to bias voltage variation and the linearity
and quantisation problems would be reduced unless
your bandwidth was low enough to get multiple photons
seen as individual events at a significant rate.

It has been ages since I've thought about operating a light-collection
system in single-photon mode, using either a photodiode or a
photo-multiplier tube. But you're right, this sounds like a way to
get around some of the problems a CCD creates. Why isn't it done this
way? Good question. I don't have a definitive answer. I'll guess.

My variable-star observing friend initially built a PMT housing for
the back of his telescope. A major hassle in the system was that, in
order to make calibrated measurements, you had to steer the scope back
and forth repeatedly between two stars. You say that a photodiode
isn't sensitive to position? A PMT certainly is. Lining up a star
the same way twice, so that you can compare successive PMT
measurements, will make you turn gray before your time. (Or would
have, anyway -- I come from the age before GOTO telescopes...)

I can understand that, the efficiency of the coating
probably varies with many aspects. I would expect a
PIN diode with 80% QE to be more uniform but that is
just an expectation, I have no experience of using them.

Another tradeoff is in discriminating your star from its neighbors.
CCD pixels are small, so you can draw a really tight perimeter around
your star. A large photodiode might also gather light from dim,
unwanted sources adjacent to your star.

Inaddition a large area device would have a higher
junction capacitance and probably higher dark current
too. OK, I have to revise my ideas and suggest the best
size would be slightly larger than the size of a star
at the focal plane, which I understand to be theoretically
the PSF but in reality larger due to seeing conditions.

Another practical concern, at least for people who aren't doing this
for a living, is that CCD's can be used for other types of astronomy
when you aren't doing photometry. A photodiode will be used for just
one purpose, and it requires special hardware, too (a high-voltage
power supply, and a counter, if you're operating in single-photon mode
as you suggested).

I had a browse on the web and generally 3V to 25V seems
to be the range. Probably a 9V battery would be adequate
and the current is negligible. There would need to be
power for the analogue circuitry but probably peltier
cooling would be the biggest drain.

Finally, the CCD allows you to perform photometry on many stars at
once, rather than just one at a time. In the case of TReS, the
specific short-period planetary transits that they wanted to find are
pretty rare, occurring perhaps in only one in several thousand stars.

Ah, now there's the rub. Yes, that's a key point when
surveying large numbers. I had in mind an examination
of a single star.

... Then you would need to do
a fair amount of math to tease out the variations as a function of
time.

Presumably the prime part of any processing would be
a Fourier transform and a major problem is the limited
and irregular observing times.

What am I missing?

You've jumped to the end. You assume that you have an accurate light
curve for your star, and you can look for variations over time with a
Fourier transform.

Not quite, I assumed that any variations from battery
voltage, variable sensitivity of the device and so on
would be random in time and therefore produce a uniform
noise background in the frequency domain. Obviously
there would be harmonics of the Earth's rotation and
beats with the orbit through non-linearity (e.g. due
to the variation of the air mass) which would take
careful analysis. The bit I'm not sure about is how
regular periods without data (below the horizon) might
create false indications.

Before you can do that, you would need to know that you accurately
gathered all the light from your star, even if that light falls on
several adjacent pixels. You need to subtract your background
accurately. You need to know how the pixels you used in that image
compare to others on the CCD, because there's no way that you'll image
your star onto the same pixels tomorrow might. (Or maybe even your
next image, if you bump into your scope.) You need to correct for
atmospheric conditions in successive images, which means comparing
your star to nearby reference stars. If the references differ much in
intensity from your test star (8 bit A/D will not even cover one
magnitude of brightness, if you want to get 2% accuracy for the
intensities of all of your stars), you'll need to take exposures of
varying lengths. You need to know whether and how the CCD output
deviates from linearity, if you want to compare values from two images
with different exposure times.

Exactly why I would consider a PIN diode approach ;-)

Whew! Was that enough?

Excellent, thank you. It's given me a lot more to consider
in particular I now realise I don't know how to convert a
bolometric magnitude into a mean photon rate so I'm off to
do a bit of study. To get a significant signal from a small
number of photons means using a high resistance load, but
that with the device capacitance will limit the bandwidth
and the dark current could even overload the amplifier. In
fact there is a tradeoff between capacitance and dark current
so running at lower voltages (5v bias) may be best. I am now
thinking along the lines of short period integrate-and-dump
strategies, perhaps in the millisecond region, but that
depends on the bandwidth that could be achieved. I'll probably
never do this but just thinking it through is informative.

Another thought is that the lateral position sensor diodes
would even allow star tracking without using a guide star
though the dark currents seem higher. I still have to work
out the limiting magnitude equivalent to a 5nA dark current.

Thanks John.
George