"hot" pixel on CCD: what is it and should I be concerned?

A friend bought an astro CCD camera recently and ended up sending it back
due to several "hot" pixels. Since I have heard this term before and am
debating whether or not to get into CCD imaging, I have a some questions:

1) What is a hot pixel?

2) Are hot pixels just a one time event, or can more appear on the CCD over
time? This is an important question for me as I don't want more to build up
simply from the age or much use of the camera.

3) How are hot pixels created and what is the best way to avoid getting
them? Of course, some may answer this in #1 above which is fine.

Thank you,
Mike

On Sun, 12 Jun 2005 21:05:51 GMT, "Mike Renner"
wrote:

A friend bought an astro CCD camera recently and ended up sending it back
due to several "hot" pixels. Since I have heard this term before and am
debating whether or not to get into CCD imaging, I have a some questions:

1) What is a hot pixel?

2) Are hot pixels just a one time event, or can more appear on the CCD over
time? This is an important question for me as I don't want more to build up
simply from the age or much use of the camera.

3) How are hot pixels created and what is the best way to avoid getting
them? Of course, some may answer this in #1 above which is fine.

Thank you,
Mike


Olympus has pixel-mapping which solves this. That, and sensor
cleaning. I hope other mfgs. adopt this.
-Rich

On Sun, 12 Jun 2005 21:05:51 GMT, "Mike Renner"
wrote:

A friend bought an astro CCD camera recently and ended up sending it back
due to several "hot" pixels. Since I have heard this term before and am
debating whether or not to get into CCD imaging, I have a some questions:

1) What is a hot pixel?

All pixels accumulate electrons from thermal effects in the silicon.
This is called dark current. In a perfect world, every pixel would
accumulate this signal at the same rate, and it could simply be
subtracted as a constant from the final image. However, each pixel
behaves a little differently, which is why long exposures usually
require that a dark frame be subtracted. A dark frame essentially allows
a specific value for each pixel to be subtracted. A hot pixel is one
which accumulates dark current very quickly. If it does this so fast
that it saturates before the exposure is finished, it needs to be
removed from the final data by replacing its value with the average of
the surrounding pixels. If it doesn't saturate, it will probably be
removed by the dark subtraction.

2) Are hot pixels just a one time event, or can more appear on the CCD over
time? This is an important question for me as I don't want more to build up
simply from the age or much use of the camera.

Hot pixels are relatively fixed- you will see them in the same place in
multiple images. The number you have depends on the quality of the
sensor- even very good cameras normally have at least a few.

Over time, the pattern of hot pixels may shift. Some sensors (such as
Sony HAD types) develop new hot pixels over time, possibly in response
to cosmic ray damage. (BTW, cosmic rays hit sensors all the time, and
produce one-shot saturated pixels, but these are different from hot
pixels).


3) How are hot pixels created and what is the best way to avoid getting
them? Of course, some may answer this in #1 above which is fine.

They are intrinsic to the semiconductor fab process. Normally, you
accept that you will have some, and don't worry about it. Simple
calibration (dark and flat frames) that you need to use anyway will
probably eliminate them. If you have a few very bad ones, using a defect
map to eliminate them is common (most astronomical imaging software
supports this). Hot pixels which have been calibrated out do not result
in any visible defects in your final image.

_________________________________________________

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

"Mike Renner" wrote in message
nk.net...
A friend bought an astro CCD camera recently and ended up sending it back
due to several "hot" pixels. Since I have heard this term before and am
debating whether or not to get into CCD imaging, I have a some
questions:

1) What is a hot pixel?

2) Are hot pixels just a one time event, or can more appear on the CCD
over
time? This is an important question for me as I don't want more to
build up
simply from the age or much use of the camera.

3) How are hot pixels created and what is the best way to avoid getting
them? Of course, some may answer this in #1 above which is fine.
It is worth saying, that just as with the quality of 'finish' on a normal
piece of equipment, you can 'hand select' CCD's for lower levels of hot
pixels, and other faults. Most manufacturers do various 'grades' of CCD,
which are warranted to have different levels of defects. The defects
involved, are 'hot pixels', 'dark pixels', 'column faults' etc.. In the
case of (for example), the Kodak sensors used in many cameras, the
categories used a
'Point defects'. These are pixels that differ by more than a fixed
percentage from their neighbours. They can be 'hot' or 'cold' defects, and
the percentage allowed, varies between the chips.
'Cluster defects'. These are groups of up to five point defects.
'Column defects'. A 'line' of more than five point defects along a single
column.
Now the numbers allowed change for different chips, but for a typical
example, a defect is deemed to be a variation by more than 10% from the
rest of the chip. A 'class 1' chip, must have no cluster or column
defects, and a maximum of five point defects, of which no more than two
will fall in the central region of the chip. A 'class 2' chip, can have up
to 10 point defects, of which up to five may fall in the central region,
and may have up to four cluster defects, with up to two of these in the
central region, but still no 'column' defects. A 'class 3' chip, then has
double the number of defects allowed again, and up to four column defects.
Generally, most astronomical cameras, use 'Class 2' chips, unless you
specifically pay extra for a 'class 1' sensor. The price difference for
the higher class sensor can be _huge_. Typically at least $1000 extra...
Now point defects are not that 'terrible', if you use dark-frames, and
flat-fields, they will together compensate pretty much for most defects.
Also, even on a 'perfect' chip, there will be other faults, such as cosmic
ray hits, that will routinely appear. It is well worth looking at some of
the 'raw' Hubble images, to see just how many defects these cameras have.
Typically, they will have positively huge values for particle impacts, but
even ignoring these, the number of hot, and cold pixels is quite
suprising.
Extra point defects do sometimes appear, but this is a rare event, and
even if they do, normal processing will remove most of the effects.
Generally, larger chips are allowed to have more defects, because the
difficulty in making them with fewer defects makes the pricing
impractical. Normal CCD's/CMOS sensors used in consumer cameras, often
have a defect map stored in the ROM of the processing chip, which
automatically applies a correction to conceal faults, because otherwise
the cost of making the chips would be far beyond the price that the camera
sells for.

Best Wishes

Thanks Rich, Chris, and Roger for your responses and discussion. Now I at
least know what they are, what to look for on the CCD, and how to remove
them if needbe.

I didn't realize CCDs were sensitive to cosmic radiation. What about
radiation sources closer to home such as gamma rays? If a camera was set up
for long exposure and placed under an x-ray source for a second, what would
the resulting image look like? I wonder if such cameras are used as
radiation detectors, the way Geiger/ scintillation instruments are?

Thanks,
Mike



"Roger Hamlett" wrote in message
...

"Mike Renner" wrote in message
nk.net...
A friend bought an astro CCD camera recently and ended up sending it back
due to several "hot" pixels. Since I have heard this term before and am
debating whether or not to get into CCD imaging, I have a some
questions:

1) What is a hot pixel?

2) Are hot pixels just a one time event, or can more appear on the CCD
over
time? This is an important question for me as I don't want more to
build up
simply from the age or much use of the camera.

3) How are hot pixels created and what is the best way to avoid getting
them? Of course, some may answer this in #1 above which is fine.
It is worth saying, that just as with the quality of 'finish' on a normal
piece of equipment, you can 'hand select' CCD's for lower levels of hot
pixels, and other faults. Most manufacturers do various 'grades' of CCD,
which are warranted to have different levels of defects. The defects
involved, are 'hot pixels', 'dark pixels', 'column faults' etc.. In the
case of (for example), the Kodak sensors used in many cameras, the
categories used a
'Point defects'. These are pixels that differ by more than a fixed
percentage from their neighbours. They can be 'hot' or 'cold' defects, and
the percentage allowed, varies between the chips.
'Cluster defects'. These are groups of up to five point defects.
'Column defects'. A 'line' of more than five point defects along a single
column.
Now the numbers allowed change for different chips, but for a typical
example, a defect is deemed to be a variation by more than 10% from the
rest of the chip. A 'class 1' chip, must have no cluster or column
defects, and a maximum of five point defects, of which no more than two
will fall in the central region of the chip. A 'class 2' chip, can have up
to 10 point defects, of which up to five may fall in the central region,
and may have up to four cluster defects, with up to two of these in the
central region, but still no 'column' defects. A 'class 3' chip, then has
double the number of defects allowed again, and up to four column defects.
Generally, most astronomical cameras, use 'Class 2' chips, unless you
specifically pay extra for a 'class 1' sensor. The price difference for
the higher class sensor can be _huge_. Typically at least $1000 extra...
Now point defects are not that 'terrible', if you use dark-frames, and
flat-fields, they will together compensate pretty much for most defects.
Also, even on a 'perfect' chip, there will be other faults, such as cosmic
ray hits, that will routinely appear. It is well worth looking at some of
the 'raw' Hubble images, to see just how many defects these cameras have.
Typically, they will have positively huge values for particle impacts, but
even ignoring these, the number of hot, and cold pixels is quite
suprising.
Extra point defects do sometimes appear, but this is a rare event, and
even if they do, normal processing will remove most of the effects.
Generally, larger chips are allowed to have more defects, because the
difficulty in making them with fewer defects makes the pricing
impractical. Normal CCD's/CMOS sensors used in consumer cameras, often
have a defect map stored in the ROM of the processing chip, which
automatically applies a correction to conceal faults, because otherwise
the cost of making the chips would be far beyond the price that the camera
sells for.

Best Wishes

"Mike Renner" wrote in message
ink.net...
Thanks Rich, Chris, and Roger for your responses and discussion. Now I
at
least know what they are, what to look for on the CCD, and how to remove
them if needbe.

I didn't realize CCDs were sensitive to cosmic radiation. What about
radiation sources closer to home such as gamma rays? If a camera was
set up
for long exposure and placed under an x-ray source for a second, what
would
the resulting image look like? I wonder if such cameras are used as
radiation detectors, the way Geiger/ scintillation instruments are?

Thanks,
Mike
Many X-ray detectors use CCD's, but these are normally 'viewing' a
phosphor, that converts the X-rays to light first. Another poster alluded
to the damage that can be done to CCD's by cosmic rays, and the same
applies to X-rays. This was a problem with the Chandra spacecraft, where
X-rays were taking an unexpected route to reach the CCD directly, and
potentially damaging the chip:
http://ledas-cxc.star.le.ac.uk/newsl...09/node12.html
At high energies, the detection rate falls, with most of the particles
going straight through the die, without being detected. There are special
CCD structures to improve the detection rate of higher energy particles,
and using semiconductors based on Gallium Arsenide, rather than Silicon,
is one of the common ones.
So conventional CCD's are not normally used for anything much beyond the
UV section of the spectrum.

Best Wishes

On Mon, 13 Jun 2005 13:59:53 GMT, "Mike Renner"
wrote:

I didn't realize CCDs were sensitive to cosmic radiation. What about
radiation sources closer to home such as gamma rays? If a camera was set up
for long exposure and placed under an x-ray source for a second, what would
the resulting image look like? I wonder if such cameras are used as
radiation detectors, the way Geiger/ scintillation instruments are?

Gamma rays can produce electrons in CCDs. They aren't very efficient at
it, however. CCDs used in x-ray imaging devices normally have very
special structures, or more often, phosphor or scintillation crystals on
the surface of the array.

Cosmic rays are extremely energetic. They typically strike the sensor
(or something close to the sensor) and produce a cascade of secondary
particles, which can result in one or more streaks in the image,
involving many pixels

_________________________________________________

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

Mike Renner wrote:
Thanks Rich, Chris, and Roger for your responses and discussion. Now I at
least know what they are, what to look for on the CCD, and how to remove
them if needbe.

I didn't realize CCDs were sensitive to cosmic radiation. What about
radiation sources closer to home such as gamma rays? If a camera was set up
for long exposure and placed under an x-ray source for a second, what would
the resulting image look like? I wonder if such cameras are used as
radiation detectors, the way Geiger/ scintillation instruments are?

The major sensitivity (for chips I've used) is to fairly energetic
particles, which can dump their energy over many pixels and wipe
out an annoying area of the image. Photons (up to X-rays) will
usually trigger a smaller region (unless they're energetic enough
for secondary pair production...). One especially unfortunate
materials choice in the late 1980s was a chip (RCA? IIRC)
which was built on a glass substrate that had, for some reason, been
doped with uranium hexafluoride. Not much use for long exposures,
that one. I've used one CCD fr spectroscopy that was physically
quite thick and extraordinarily good at not just registering
cosmic-ray events (well, ion events), but their trails,
sometimes with changes in direction as interactions occurred.
I have been told that this chip began as a particle detector,
being used for astronomy only because its extraordinarily
high quantum efficiency in the near-IR out to a bit past 1000 nm
made it worth the effort to deal with these problems in each exposure.

A nice example of this can be seen in some of the SOHO coronograph
movies, in which you see the coronla mass ejection, and some hours
later the field of view is speckled with particle impacts on the
detector.


Bill Keel