Sky & Telescope - May 2020 - 33

p Left: Problems with CMOS sensors only become apparent during long exposures. This dark frame is stretched to display the amplifier glow at
bottom right. Right: Another problem with CMOS detectors used for scientific imaging is that they produce non-repeating fixed-pattern noise. Note
how the dark pattern noise has shifted between frames. High-quality and newer CMOS cameras are doing much better at reducing this issue to nearimperceptible levels.

designs, but many newer sensors exhibit amp glow that is
quite low, and some camera vendors are doing "tricks" with
the electronics to keep it to a minimum. One scientist who
is searching for ultra-faint targets tells me there is currently
no CMOS sensor on the market that performs adequately for
his very exacting work. I'll have to check in with him again in
ﬁve years and see if this is still the case.
The last signiﬁcant issue that CMOS has yet to overcome
is the non-linear sensitivity to light. When we talk about
linearity of a sensor, we are talking about the ratio of the
signal recorded to the signal that was received. If you dump in
twice the amount of light, you should get twice the amount
of signal. CCDs that are intended for scientiﬁc use are 100%
linear (non-antiblooming), so that when the signal is so high
that they saturate, they spill the excess charge into adjacent
pixels, called blooming spikes. Many CCD detectors used for
astrophotography include an anti-blooming gate to redirect
overﬂow charge and lose their linearity only at the upper end
of their exposure range. CMOS sensors, on the other hand,
are non-linear throughout much of their range. This varies
from sensor to sensor. How this affects astronomy is that ﬂatﬁeld calibration, which corrects for pixel-to-pixel sensitivity
differences, among other things, does not always work well
with CMOS sensors; the math just doesn't work. Also, if your
intended use is photometry (accurately measuring the brightness of a target), your calculated values may be suspect. They
are likely close, but not as rigorous as if determined from a
truly linear sensor.
If your goal is simply aesthetic astrophotography, sensor
non-linearity is often negligible or can be mitigated in postprocessing. Often, it's very slight (CMOS is still far superior
to ﬁlm). Of course, it requires more work than having a properly calibrated image in the ﬁrst place, but it's not a showstopper. I've seen some academic work recently in which this
is the subject of intense research, and I wouldn't be surprised
if technological advancements solve this problem within a
generation or two.

Finally, another commonly discussed issue with CMOS
sensors is ﬁxed-pattern noise. Both CCD and CMOS sensors
have ﬁ xed-pattern noise, often noticeable during longer exposure shots when particular pixels are susceptible to giving
brighter intensities above the general background noise. This
is easily removed by calibration with a bias or dark frame on
a CCD sensor. With some CMOS cameras, the pattern noise
often varies from frame to frame, and since it's not repeatable, it can't be simply subtracted. However, a lot of this pattern noise has more to do with the surrounding electronics
implementation than it does with the sensor itself, and camera vendors are gaining more experience in taming this beast
for our market. For example, the issue is already a nonfactor
for most commercial DSLR cameras, as well as the latest generation of cell phones capable of low-light imaging.

Onward into the Future
Technological progress is often fraught with growing pains,
false starts, and compromises. There are still and possibly
always will be some applications where CCD-imaging technology works best, such as in space where radiation hazards
are very harsh on delicate electronics. To be sure, there will be
manufacturers willing to meet that need for a price. For those
of us on Earth without government-sized budgets, the juggernaut of CMOS is well beyond critical mass now. Clearly, the
Sun is setting on CCDs for amateur astronomers and most
ground-based imaging applications.
But I wouldn't be in a hurry to toss out your old CCD
cameras. The current generation of CCD cameras is going
to be available for a few more years yet, and it will offer
some important advantages for the discerning imager. While
CMOS sensors are still not quite up to the challenge of many
kinds of scientiﬁc imaging, we can say conﬁdently that they
are getting very large in the rear-view mirror.

¢ RICHARD S. WRIGHT, JR. is a software developer by day
and an avid astrophotographer by night.
s k y a n d t e l e s c o p e . o r g * M AY 2 0 2 0

http://www.skyandtelescope.org

Sky & Telescope - May 2020

Table of Contents for the Digital Edition of Sky & Telescope - May 2020