Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

From CMOS to sCMOS and back again!

_ Andrew Kirby, Technical Specialist, Atik Cameras

Introduction
Despite existing CCD being a very mature and well-developed imaging technology, there
are now both performance and supply limitation reasons to consider using new generation
CMOS devices. This will eventually also apply to applications such as Gel Documentation,
where the use of CCD is almost certainly still preferable. Here, the subject is static and the
main criterion is principally low light sensitivity, where CCD is clearly an excellent fit. In the
short term, CCD can continue to be the technology of choice for this sort of application.
However, due to the reprioritisation of R&D, and production scale back by the major
players in this market, the lack of widespread availability of CCD sensors after 2025 is likely
to cause issues. Consequently, even if you are currently happy with selecting CCD cameras
for your particular application, there will eventually be a time when it is necessary to
consider alternative technologies such as CMOS or sCMOS. But what exactly is sCMOS,
and how does it differ from CMOS?

Discussion – the necessity behind the evolution of CMOS


The technical motivation behind the development of existing CMOS imaging sensors,
lays with the fundamental limitations of CCD technology. The most significant of these, is
probably frame rate. As CCDs are inherently serial devices, each pixel has to effectively
‘wait its turn’ for its data value to be read out. This restriction places an upper limit, of only
a few frames per second, on the useful frame rate of such devices, and consequently
limits the number of applications that these sensors can be used for. There are countless
other areas, in both domestic and industrial markets, that would greatly benefit from
imaging sensors that have the fabulous sensitivity of CCD technology but with a much-
increased frame rate. CMOS devices do not have such a limitation, since their data stream
is inherently parallel in nature. It would seem desirable to develop a new type of imaging
sensor with the best qualities of both existing technologies.

Early CMOS devices went some way to broadening their market appeal by being a good
fit for all manner of domestic devices such as web cams, compact cameras, home video,
and mobile phones etc. With it, this wide adoption brought the benefits of economies of
scale - something that CCD could never really achieve, particularly at high pixel counts.
However, these first-generation CMOS sensors were, in some respects, relatively crude.
Whilst they worked reasonably well in domestic applications, where the light level was
almost always plentiful, their performance was lacking in low light environments, and
they had only a modest dynamic range. Low light photography requires the use of longer
exposures for a satisfactory image, particularly as these first-generation CMOS devices
had somewhat limited quantum efficiencies (QE). These original CMOS sensors have pixels
based around a three transistor architecture. In this arrangement, the areas responsible for
charge collection and readout are poorly separated which limits the noise performance.

_1_
In order to achieve a fast read out speed, each pixel in a CMOS sensor has its own
associated amplification and shutter circuitry. Additionally, once read, the image
information is manipulated ‘on chip’ by some relatively sophisticated image processing
hardware. Often without the end user needing to be aware of every step. Therefore,
these devices can be considered largely digital in nature – they are basically miniature
computers relying heavily on software instructions. This is a world apart from the very simple,
purely analogue CCD devices, where image information is largely untouched until it is
converted to a digital array ‘off-chip’.

Although, at the time, hybrid CCD/CMOS devices were being experimented with, these
were extremely expensive to fabricate. Therefore, the principal choice was between slow,
expensive, yet super sensitive traditional CCD, or the faster, low cost, but lower performing
CMOS devices, with a modest dynamic range. Clearly what was desired was a highly
evolved imaging device that could bridge the two technologies, and combine their
best qualities.

The Appearance of sCMOS


In mid-2009, a whitepaper was
published by Coates et al (1)
of the Andor/Fairchild/PCO
consortium, which detailled a
new imaging sensor technology
that was termed as sCMOS or
‘Scientific CMOS’. Both rolling
and global shutter modes were
available in the same device,
depending on the application.
Although many of the key
technologies behind creating
this new family of sensors lay
undisclosed, a brief description
of the read-out mechanism was
outlined. Here an innovative
amplification and ADC circuit with
two selectable gain factors was
employed (Fig 1). This effectively Fig 1. Analogue to digital converters
allowed the sensor to record two
passes at each (hi/lo) gain setting, and subsequently recombine them in order to generate a
final image with a much improved dynamic range. The result is similar to that as if the image
had actually been recorded with a sensor featuring double the number of bits of information.
Thus, circumventing one of the principal limitations of the original CMOS design. Featuring a
low read noise of less than 2e- even at high frame rates, these new sensors were attractive
for applications such as fluorescence microscopy and flow cytometry. Additionally, their high
speed was also generally beneficial in low light applications, where smooth manual focussing
could be considerably enhanced and the implementation of auto focussing became a
much simpler proposition. Delving a bit deeper into some if the design differences (2.) over
traditional CMOS, reveals that the pixels in sCMOS sensors utilise quite a radical four transistor
architecture. Along with the use of the correlated double sampling technique, familiar from
CCD designs, this leads to an arrangement with a particularly low readout noise.
From a manufacturer’s perspective, and a decade later, sCMOS imaging sensors can
now deliver an almost ideal set of specifications, making them relatively straightforward to
_2_
integrate into a diverse range of scientific instruments and applications. It is probably only the
extreme low light, and very long exposure applications, where traditional CCD sensors are
still preferred. Over time, the technology behind the original sCMOS sensors has subsequently
been adopted into a large number of the current ‘regular’ CMOS designs. Therefore, in many
cases, they are no longer distinguishable in a meaningful way, and the two terms are now
often used interchangeably.

Summary
In the medium term, CCD devices are still available for those low light applications that
benefit from this technology. Looking forward, manufacturers of scientific instruments will
need to evaluate which of the sCMOS sensors best suit their needs – either as a replacement
for an existing sensor or as a first venture into this exciting technology. Prospective users should
not be confused by the terms sCMOS and CMOS, as in the majority of cases they are now, in
fact, one in the same.

1. Coates C, Fowler B, and Holst G. sCMOS Scientific CMOS Technology - A High-Performance Imaging
Breakthrough (White Paper). 2009, published online
2. Holst G. Scientific CMOS camera technology: A breeding ground for new microscopy techniques. 2014,
Microscopy and Analysis

Should you need assistance in choosing a camera solution for your particular application,
please feel free to contact us via the web form:
https://www.atik-cameras.com/contact/

v1.1

Unit 8 Lodge Farm Barns, Norwich, Norfolk, NR9 3LZ, UK. www.atik-cameras.com

_3_

You might also like