hot pixels: worn out topic, new obervation

Soon after I bought my second 20d body last December, I shot a series of images at all ISO's (except 3200) with NR off, at 1, 3, and 5 minute exposures and then ran the hot pixel test program on them. I wrote down the results, also noting how many hot pixels were visible at normal magnification to the naked eye. The results were acceptable, and comparable to the same tests ran with 20d #1 a month earlier.

I take many, many long exposure shots, in the 2 to 60 second range, with NR off to save time. I recently noticed in a batch of ISO 200 images, taken at 13 to 20 second exposure times, two prominent new hot pixels. One's blue and the other one is red. They're there in every shot, at the same locations. I checked my original test shots from December, and even at the 3 and 5 minute shots at ISO 200 they were not there. That was about 1500 images ago, by the way.

A new hot pixel test reveals that the dead pixel count has remained constant (1), but the hot pixel levels have increased by about 75%.

It's still a tolerable situation, since a quick retouch can get rid of them, assuming the part of the image they're in is dark enough to reveal them in the first place. I realize hot pixels increase over time, and that temperature is another factor. But, this seems like a quick increase. Then again, it might very well be normal. I'm wondering if repeated long exposures is having any negative effect on the sensor.

Just my observation.

GS

Was the camera at the same temperature each time you ran the test?

For CCDs the dark count doubles about every 6 to 8°C, so temerature has a huge influence on hot pixels. I'm pretty sure it's about the same for CMOS chips. And the chip does warm up as you use it, you'll see a difference in hot pixels if you run the test when you first turn the camera on, and then use it for a bunch of shots and do the test again.

Over time the chip will change, this is usually slow and doesn't really depend on how long the exposures are, more on operating conditions. The chip really isn't doing any work when exposing, it's working when being charged up before and during readout after. Cosmic rays will also cause hot pixels, if you take your camera on a lot of airplane trips you have a slightly higher chance of getting more of them than staying on the ground.

--
Later,
Marty

http://science.widener.edu/~schultz/digipicts.html

Marty Schultz said:

Was the camera at the same temperature each time you ran the test?

Click to expand...

Yes, approximately 75 degrees.

Marty Schultz said:

For CCDs the dark count doubles about every 6 to 8°C, so temerature
has a huge influence on hot pixels. I'm pretty sure it's about the
same for CMOS chips. And the chip does warm up as you use it,
you'll see a difference in hot pixels if you run the test when you
first turn the camera on, and then use it for a bunch of shots and
do the test again.

Click to expand...

For the December tests, the camera had been on for about 10 minutes when I took the first of the three (the one minute shot) ISO 200 shots, having just taken the three ISO 100 shots totaling 9 minutes of open shutter.

Last night, I only did three ISO 200 shots. The first shot, with the camera freshly turned on for the first time of the day, was the one minute shot at ISO 200. It had about 75% more hot pixels than the one minute ISO 200 shot taken in December after the camera had already been on 10 minutes.

So, same temp, less expected "sensor heating" due to less operating time, but more hot pixels. The ratio is going in the opposite direction.

Marty Schultz said:

Over time the chip will change, this is usually slow and doesn't
really depend on how long the exposures are, more on operating
conditions. The chip really isn't doing any work when exposing,
it's working when being charged up before and during readout after.
Cosmic rays will also cause hot pixels, if you take your camera on
a lot of airplane trips you have a slightly higher chance of
getting more of them than staying on the ground.

Click to expand...

No plane trips, yet.

Marty Schultz said:

--
Later,
Marty

http://science.widener.edu/~schultz/digipicts.html

Click to expand...

Hmmm, I guess you'll have to monitor it over time. You may have had a bunch of pixels that were just under the threshold of being called "hot", and with normal use they have gone over the threshold. With the number of chips and cameras they are producing I doubt they get much 'burn in' time, the consumer is the 'burn-in' tester. Unless you got a bad chip future tests should only show a small increase.

Another point: running the test at ISO 200 will introduce more random electronic noise than at ISO 100. To truely test for hot pixels you should run the test at ISO 100, then compare the locations of the hot pixels to see if they are coming up in the same spot each time.

Greensteak said:

Marty Schultz said:

Was the camera at the same temperature each time you ran the test?

Click to expand...

Yes, approximately 75 degrees.

Greensteak said:

For CCDs the dark count doubles about every 6 to 8°C, so temerature
has a huge influence on hot pixels. I'm pretty sure it's about the
same for CMOS chips. And the chip does warm up as you use it,
you'll see a difference in hot pixels if you run the test when you
first turn the camera on, and then use it for a bunch of shots and
do the test again.

Click to expand...

For the December tests, the camera had been on for about 10 minutes
when I took the first of the three (the one minute shot) ISO 200
shots, having just taken the three ISO 100 shots totaling 9 minutes
of open shutter.

Last night, I only did three ISO 200 shots. The first shot, with
the camera freshly turned on for the first time of the day, was the
one minute shot at ISO 200. It had about 75% more hot pixels than
the one minute ISO 200 shot taken in December after the camera had
already been on 10 minutes.

So, same temp, less expected "sensor heating" due to less operating
time, but more hot pixels. The ratio is going in the opposite
direction.

Greensteak said:

Over time the chip will change, this is usually slow and doesn't
really depend on how long the exposures are, more on operating
conditions. The chip really isn't doing any work when exposing,
it's working when being charged up before and during readout after.
Cosmic rays will also cause hot pixels, if you take your camera on
a lot of airplane trips you have a slightly higher chance of
getting more of them than staying on the ground.

Click to expand...

No plane trips, yet.

Greensteak said:

--
Later,
Marty

http://science.widener.edu/~schultz/digipicts.html

Click to expand...

Click to expand...

--
Later,
Marty

http://science.widener.edu/~schultz/digipicts.html

Thanks for the information.

Marty Schultz said:

Hmmm, I guess you'll have to monitor it over time. You may have
had a bunch of pixels that were just under the threshold of being
called "hot", and with normal use they have gone over the
threshold. With the number of chips and cameras they are producing
I doubt they get much 'burn in' time, the consumer is the 'burn-in'
tester. Unless you got a bad chip future tests should only show a
small increase.

Another point: running the test at ISO 200 will introduce more
random electronic noise than at ISO 100. To truely test for hot
pixels you should run the test at ISO 100, then compare the
locations of the hot pixels to see if they are coming up in the
same spot each time.

Click to expand...