by Sergey Samarin
01/11/2004 | 11:56 PM
Certain peripheral devices look like a “thing in itself” for the outsider, especially if the device includes both: electronics and mechanical parts. They are something that seems too hard to be grasped by a common person. Flatbed scanners are an example of such devices.
At first glance, construction of a scanner has nothing sophisticated about itself: a case with a few connectors and a couple of buttons, a lid and a glass where you place the original for scanning. That’s quite simple and you of course can use your scanner without bothering about technicalities. At second glance, when you try to find out how the stuff works and what those mysterious numbers in the specifications stand for, things become more complicated. So if you don’t have a scanner, but are planning to buy one, you should learn what real things stand behind the characteristics declared by the manufacturer. In this article I will try to take a scanner apart (literally) to show you what it is made of. Doing this exciting job, we will also learn certain facts about scanners in general.
I guess we should start out with the most important component of each scanner. Its eyes, which are a light-sensitive matrix.
It is the matrix that mostly determines what a scanner is. It transforms color and brightness of the incoming light flux into analog electrical signals that would make sense for its only mate – analog-to-digital converter (ADC). The ADC acts like both an interpreter and a guide, always coming with the matrix. Without interpretation, no processor or controller would understand the analog signal from the matrix. But they do understand digital signals produced by the ADC – a string of zeroes and ones. The digital electronics of a scanner are blind – you can shine at them with your flash-light and never get a single reaction. The matrix is quite another thing: a stream of light falls onto its surface literally beating electrons out of its sensitive cells. The higher the intensity of light, the more electrons the matrix releases. And as a result, the stronger will these electrons be when they all rush to the exit. However, the electrons current is too weak to be heard by even the most sensitive ADC. That’s why there is an amplifier at the matrix exit, which can be compared with a megaphone making a siren wail out of a mosquito hum. The amplified signal (analog so far) will be “weighed” by the converter so that each electron could receive a digital value, according to its current strength. That’s where the matrix quits the game: we’ve got digital information now to be processed by other devices. Image reconstruction doesn’t require the matrix any more.
This is the general picture, let us now delve somewhat deeper into technical details. Many contemporary scanners for home of office use (Small Office Home Office – SOHO) come with a matrix of one of the two types: Charge-Coupled Device (CCD) and Contact Image Sensor (CIS). This fact alone produces two natural questions: “What is the difference and which one is better?” You can catch the difference with a naked eye: a CIS scanner comes in a thinner case compared to a similar CCD machine (its height is usually about 40-50mm). The second question is rather harder to answer, since both matrix types have advantages and shortcomings. They are listed in the following table:
Support high resolutions (contemporary low-cost CCD-scanners support up to 2400dpi optical resolution);
Relatively high pricing (compared with CIS-scanners);
Limited optical resolution (up to 1200dpi);
A CCD scanner boasts higher depth of field than its CIS counterpart. This is achieved by using an objective lens and a system of mirrors:
For your convenience, there is only one mirror in the picture,
although a typical scanner has at least three or four of them
CCD scanners are more widespread than CIS ones. This is obviously due to the fact that you usually buy a scanner not only to digitize text documents, but also to scan photographs and color images. If this is the case, the user would want to have a scanner with an accurate and realistic color reproduction, and a CCD scanner is much more reliable especially for reproducing color tones, lights and halftones than a CIS scanner. For your reference: standard CCD scanners have a color dispersion error of around ±20%, while CIS devices have an error of ±40%.
Schematic representation of a CIS sensor
CIS array consists of a line of LEDs that illumine the original, the self-focusing micro-lenses and the sensors themselves. The whole construction is quite compact, so this is a general rule that a contact-sensor scanner is always a little thinner than its CCD counterpart. This type of scanners also consumes little power and is highly mechanically stable. However, their functionality is somewhat limited as they don’t generally support slide-adapters and automatic document feeders.
Because of the technology peculiarities, CIS array provides a relatively low focal range. For example, CCD scanners have a focal range of ±30mm, while CIS ones of only ±3mm. In other words, if you place a thick book onto the bed of such a scanner, you will get a scan with a blurred band in the center, where the original doesn’t touch the glass. A CCD device would produce a sharp image, since it features a system of mirrors and a focusing lens. However, you have to “pay” for this better quality: the bulky mirror system doesn’t allow a CCD scanner to become as thin as its CIS analog. Note also that the optical system of a CCD scanner should meet strict requirements, so those rumors about some scanner models using “plastic mirrors” are quite groundless.
CIS scanners cannot compete with CCD ones as far as their optical resolution is concerned, either. Some models of CCD scanners for home and office use provide currently an optical resolution of 3200dpi, while CIS scanners offer no more than 1200dpi (if I am not mistaken). Still, it wouldn’t be prudent to discard CIS technology altogether: such scanners found their niche digitizing sheet originals rather than books. The fact that such scanners are fully power-supplied by the USB bus and ask for nothing more is very attractive for portable computers owners: they can digitize an original and transform it into a text file wherever they are, without having to seek for a wall outlet. This makes up for the drawbacks of the contact sensor. That’s why the question “Which scanner is better?” should be answered basing on your own particular needs.
A CCD matrix is the most important component of a scanner
A CCD matrix (array), like the one you see above, looks like a “big microcircuit” with a glass window – that’s where the light beam reflected from the original falls. The array works all the time until the scan head passes to the end of the glass. Note: the full distance for the head to travel along the Y-direction is referred to as sample rate or mechanical scanner resolution (we will talk about these two characteristics a little bit later). The array fully covers one horizontal line of the bed (raster line). After the line has been processed, the head moves on to the next line.
Side view of a CCD-matrix
Do you see the two ordinary screws in the snapshot above? When the scanner was assembled, these screws helped to align the matrix (note also the U-shaped slits in the PCB in the photo with a view from above) so that light reflected from the mirrors could be evenly distributed on the matrix surface. By the way, if one of the optical system elements is misaligned, you will get a “striped” image.
Parts of a CCD matrix magnified
(the macro-snapshot was taken with Canon EOS D60 digital camera)
You can see in the enlarged snapshot of the CCD matrix that it has its own RGB filter. The filter is the main component of the color separation system, which has always been the talk of the town, but which appears pretty complicated to understand. You usually get an explanation like “a standard flatbed scanner uses a source of light, a color separation system and a charge-coupled device (CCD) to collect optical information about the scanned object”. In fact, light can be divided into color constituents that are then focused on the matrix filters. The scanner objective lens is also very important for the proper work of the color separation system.
The scanner objective lens is not as big as it seems in the snapshot
A scanner case should be rigid enough to avoid any misalignments in construction. Of course, it would be best for a scanner to have a metal chassis in its base. However, many currently produced SOHO scanners come in plastic cases – to reduce their production cost. In this case, the thing is made stronger by adding special stiffening ribs.
This is where the main functional units of a scanner are located in the case
A transportation lock is another important element of the case: it protects the scan head from damage during transportation. Before turning on a scanner, you should release this lock. If you forget about this, you are running the risk of damaging the scanner. Manufacturers usually emphasize this nuance with bright labels and warning captions.
Some users think the case doesn’t influence the quality of scanning in any way. This is not exactly so. The optical system of the scanner cannot stand dust, so the case should be hermetic, without any slits or holes. I came across some models that didn’t meet this requirement, so pay attention to the case when shopping for a scanner.
A removable lid is a useful feature, too. It’s especially required when you are trying to scan thick books or magazines.
The edges of the bed should slope downwards – this way it is easier to take the original up from the glass. Besides that, there should be no clearance between the glass and the bed that would hinder your removing the original. Make sure the scanner has marking rules around the perimeter of the bed.
All scanners are controlled by the computer to which they are attached and the user can adjust scan settings via control software. It means that SOHO scanners don’t necessarily have their own control units, although some manufacturers help the beginner and insert a few “fast scan” buttons (generally, in the front panel of the case).
Fast scan buttons are something you can well do without
Each button is marked by an icon. Typical functions assigned to the buttons include: launching a standard scan operation with output to the printer, to your e-mail client, to the fax and so on. Of course, there are definite scan settings for each button. Anyway, when you push a button, an application responsible for it will start up. By the way, control units are available in few SOHO scanners, and are always missing in professional machines.
Some manufacturers often remove certain options from the scanner driver, which most of the users don’t use. For example, SOHO scanners from Hewlett-Packard don’t allow changing the gamma, loading ICC profiles and so on. However, Hewlett-Packard is known for pleasing its customers with “fast scan” buttons in most products.
Every scanner has an internal lighter. It is a small and powerful unit responsible for turning the scanner lamp (or anything that serves as a lamp) on and off. A CIS scanner uses a LED line as a source of light, and this is why it consumes so little power.
CCD scanners generally illumine the original with a fluorescence lamp with a cold cathode. This lamp is incomparably brighter than light-emitting diodes, but high voltage is required to make the gas inside the lamp shine. A unit called inverter provides this voltage.
A high-voltage unit is required for supplying power to the lamp
The inverter pulls the voltage up from 5 volts to several thousand volts, and also transforms direct current into alternating current.
Generally, there are three types of lamps used in scanners:
However, for several reasons, SOHO scanners only use lamps of the last type (with a cold cathode).
Cold Cathode Fluorescent lamp
The scanner lamp rests on a plastic chassis of the scan head right above the lampshade. The lampshade is shaped like a reflector (a “collector” and “thrower-off” of light) or magnifying glass. The reflected light is more intense and shines at the object placed on the bed. Reflecting from the original, light goes through the chassis slit (its contour is outlined with blue in the snapshot) and falls onto the first, longest mirror of the optical system.
An evident advantage of a cold-cathode lamp is its long service time, about 5,000-10,000 hours. That also explains why some scanners don’t turn the lamp off after a scanning operation is over. Then, lamps of this type do not require any additional cooling and have low manufacturing cost. The disadvantage of these lamps is slow turning-on. It takes from 30 seconds to several minutes for a lamp like that to warm up.
The lamp matters a lot for the quality of scanning. Even a slight deviation in the characteristics of the light source may affect the light stream that is falling onto the reception matrix. This fact explains why it is necessary to warm up the lamp properly before scanning. Some drivers allow reducing the warm-up time if you don’t want to have excellent scanning quality (when scanning text documents, for example). The lamp’s characteristics degenerate with the time, which is a natural inevitable process, but scanners can automatically calibrate themselves according to a black-and-white pattern that is integrated into the case.
As you can see in the snapshot, the plastic of the case
and the calibration target get dimmer after being exposed to light
The scanner I took as an example complies with the rule. The photo above shows the color pattern according to which the scanner adjusts colors before scanning to compensate for “ageing” of the lamp. The plastic inside the case, which is permanently illumined by the lamp, as well as the calibration pattern become dimmer with time. This brings in more color distortion.
A cold-cathode lamp resembles a daylight lamp… only a small one
If you are inventive enough, you can design a desk lamp
from the inverter and a cold-cathode lamp
That’s another possible use for the scanner lamp. The inverter unit was connected to a standard computer PSU with wires and an adapter. Add a holder and you will have a bright desk lamp.
An analog-to-digital converter helps to establish communication between the processor of the scanner and its matrix. The ADC measures the input voltage and describes it in numbers that are then sent to the processor.
For example, we’ve got 4V at the input of the ADC, then this value changes to 9V. It means we get a string of 00000100, then 00001001 at the output. These are binaries for numbers 4 and 9. The number of zeroes and ones used by the ADC to represent the value is referred to as bit capacity (measured in bits). It is an important parameter of a scanner that indicates the precision of the input signal as measured by the ADC.
Today you can come across an inexpensive scanner that uses a converter with a bit capacity of 24 – 48bit. In theory, you should go for a scanner with a higher bit capacity, but there is one thing to remember about: sometimes the manufacturers write “48 bit” in big letters on the box, with a small print somewhere in the corner that reads “software 48bit, hardware 36bit”. In this case, the real bit capacity of the used ADC is 36bit and you should take this number into account. I should admit that it’s next to impossible to distinguish between scans taken on a 36-bit and 42-bit scanner (the human eye can distinguish between about 16.7 million colors or 24bit). In our case, the bit capacity of the converter and the color depth is the same thing as the converter just calculates the colors of dots the image is made of. The higher the bit capacity, the more true-to-life image you receive.
Modern scanners come equipped with special-purpose processors. This processor makes all the circuits and units work in tune. It also comes up with the image data to be sent to the CPU. In some scanners, the processor also functions as an interface controller.
The list of software instructions for the processor is stored in a ROM chip; it is written into the chip at the production stage. The contents of the chip are called firmware. Some professional scanners allow updating their firmware, but it is not necessary for SOHO scanners.
Besides the ROM chip, scanners use 1MB or 2MB of RAM as a buffer. This is where the scanned information is sent to. After the contents of the buffer are sent to the computer, the scanner processor zeroes the buffer for a new scan. Instructions to be executed by the scanner processor are also stored in local RAM (the processor has a few kilobytes of its own memory). The processor memory is organized like a pipeline: the processor executes instructions one by one, so that the second instruction comes to the place of the first one, and the new instruction comes instead of the last one, as it shifts ahead.
The amount of the local memory used to be indicated in the technical specifications. Today it is generally omitted, as it doesn’t practically affect the scanner speed. It is also omitted when a particular scanner uses some portion of the computer main RAM, which is implemented in the driver.
It is the interface controller that ensures communication between the scanner and the computer. As I have mentioned above, a scanner may come without a controller chip, if the controller is integrated into the scanner processor. Throughout their history, scanners used to connect to the computer across SCSI, IEEE1284 (parallel or printer port) or even RS-232 (serial interface). Currently available models of SOHO scanners use USB, FireWire or SCSI. Once, Bluetooth scanners were rumored to appear, but they never did. Of course, scanners with different interfaces use different, incompatible, controller chips.
In our case, we have SCSI and USB ports as well as
two connectors for attaching additional units
Scanners with the SCSI interface were very popular just a few years ago, but their time has come to an end. Newer high-speed interfaces (USB and FireWire) that don’t require additional adapters or delicate and careful connection have become the killers of SCSI in the world of SOHO scanners. The advantages of the SCSI interface include high bandwidth and the opportunity to connect up to seven various devices to one SCSI bus (i.e. to one SCSI host-adapter). Its disadvantages are high cost and the necessity to use an additional controller.
The USB interface is the most popular for scanners produced today as this interface is integrated into all modern mainboards as the main means to connect peripheral devices to the computer. CIS scanners are power-supplied through the USB port, which is a nice feature for portable computer owners. This energetic freedom couldn’t be obtained with SCSI.
This is my favorite. FireWire is a serial high-speed I/O interface differing from USB by the fact that it doesn’t require a controller for establishing communication. FireWire works according the peer-to-peer principle that is why FireWire is less hungry for processor resources than USB.
A new version of this interface (FireWire 800 or IEEE1394b) is going to be used in peripheral devices soon. It is going to be the fastest peripheral interface ever invented.
There is one thing in a scanner that moves – the scan head. It includes an optical unit with a system of lenses and mirrors, a light-sensitive matrix, a cold-cathode lamp (if it is a CCD scanner) and an inverter board. The belt gear transfers the motion of the stepping motor to the scan head.
That’s where the belt is fastened to the scan head
Components of the belt mechanism
A special spring put onto the belt itself provides tight contact between the belt and the wheels. The carriage with the scan head moves along the guides that lead along the scanner case (see the picture).
The stepping motor rotates the spindle into either direction in small steps, so that the scan head moves at a certain distance. This motor resides in any flatbed scanner. It rotates the reducer (those cog-wheels you see in the snapshot) and moves the carriage in which the optical unit, lamp and matrix are. A special controller chip is responsible for changing the direction and rotation speed of the motor. The precision with which the carriage moves is referred to as mechanical resolution on Y-direction.
Optical resolution of a scanner is X-direction,
and mechanical resolution is Y-direction
Overall, optical resolution is determined by the number of elements in a matrix line divided by the width of the operational area. Mechanical resolution is the number of steps the scanning carriage can make in Y-direction. You can often come across a notation like “600x1200” in a scanner’s specification. The first number stands for optical and the second – for mechanical resolution of the scanner. There is also the so-called interpolated resolution, which may be a few times higher than optical resolution, but doesn’t depend on the physical capabilities of the device. I would call it “scaling resolution”. Interpolation (enlargement of the original image) is performed by the scanner software. You can disregard this parameter altogether as it’s quite possible to do the same interpolation in an image-editing software like Photoshop.
Insides of the motor
The external side of the motor core is connected with a gearing which is a simplest reducer. Its large cog-wheel pulls in the belt with the connected scan head.
Power supply unit
SOHO scanners don’t consume too much power, so you can’t find powerful elements in their power supply units. The internal power unit of the device we examine today outputs 24V/0.69A, 12V/0.15A and 5V/1A. The light source – the cold-cathode lamp – needs high voltage of several thousand volts and it is power-supplied by another unit I have already discussed earlier.
There are numerous additional devices you can use with your flatbed scanner (they are usually purchased separately) like automatic document feeders and slide-adapters.
A scanner with an automatic document feeder is quite a bulky construction
A document feeder may be of invaluable help when you have to scan a ream of standard-size papers. To make sure your scanner supports feeders, take a look at its back panel and look for a connector marked as ADF (Automatic Document Feeder). An important note: auto-feeders are always “linked” with a specific scanner model or a series of models. There is no such thing as a “universal document feeder”! The reason for that lies in the fact that the feeder is controlled by the interface card of the scanner. Of course, it’s impossible for the feeder to work without a connection to the scanner, so be careful when shopping and make sure your scanner supports the feeder you want to buy.
This is the transparent window of the document feeder
from the other side of the glass
The feeder works like this: having completed the auto-calibration and status check, the scanner positions the head in front of the feeder’s transparent window. Then, original sheets are taken from the input tray one by one to be digitized as they pass through the transparent window.
The slide-adapter is an extra device for digitizing transparent originals (film, slides, negatives). There are two types of these adapters: passive ones use the scanner lamp and active ones shine through the original with their own lamp.
Some active slide-adapters are equipped with their own light source shining through a transparent original. Some models feature a moving carriage with a light source, which is moved with a motor and a belt mechanism. The light source is moving along with the scanner head. The scanner’s own lamp is turned off in this case.
Today, SOHO scanners without moving parts in the slide-adapter unit are the most popular. We have recently tested one device like that, EPSON Perfection 3200 Photo. Its light source is built into the scanner lid and occupies its entire useful surface. For the adapter and the scanner to work together, a cable comes from the lid and gets attached to a special connector in the scanner back panel (marked as “XPA”). The light of the adapter is automatically activated when you change the type of the original in the control program; an indicator in the scanner lid informs you about that. Transparent originals are installed into stencils included with the device: 35mm film of 12 stills, four 35mm slides in frames, 120/220 (6x9cm)/4x5” film. Then you lay the stencil onto the glass.
During the scan process, light stream goes through the transparent original, hits the input of the scanner’s optical system and is handled just like a non-transparent original. Of course, such features of the scanner like optical resolution or color depth do not change when you use a slide-adapter, but the range of optical density does. This last parameter directly depends on the brightness of the light source and the time of exposure. It can be described as follows:the darker is the original, the less light can go through it, and the more time it takes the CCD matrix capacitors to collect a necessary charge. The darkest of all transparent materials is X-ray film (up to 3.6D). In order to get a quality scan of such a film, you need a bright light source. However, the range of reproduced optical densities doesn’t depend on the lamp brightness alone. It is also determined by the precision (bit capacity) of the analog-to-digital converter, by the abilities of the optical system and of the light-sensitive matrix.
A passive adapter is a much simpler device. It uses the scanner’s own lamp and you get poorer results than with an active adapter (intensity of the light stream is lower just like the quality). However, you can use passive slide-adapters for scanning images for Web. They are cheaper.
The scanner is a sophisticated electronic device and I could discuss it for ages, getting deeper into specifics and nuances. So far, we have learned the following: