6.3.1
Modern computers usually use magnetic disk drives for long term storage, as it is considered to be relatively non-volatile, as in it does not lose its data when turned off. RAM, on the other hand, requires constant power to maintain its memory. Before we start, an understanding of the insides of a hard drive is important.
Hard drives are very rarely composed of a single disk. They are usually made of between two and four platters. The platters are typically made of either aluminum or glass, and are given a magnetic surface which is polished to a mirror finish. The normal desktop hard drive spins at 5400 RPMs (although they are available at speeds up to 15000 RPMs). RPM speeds are constant during normal operation; other than when spinning up at powerup time or spinning down when being turned off, a hard drive cannot adjust its rotational speed.
A head is a device used to read and write bits, and is held on with an actor arm. A magnetic sensitive head scans over the platter to read and write magnetic bits in one direction, but because the platters spin, the heads can access bits at any part of the hard drive. There is one head per platter. Most modern platters have a high bit density allowing 40 gigabytes of data to be stored on each. However, higher RPM drives usually have lower bit density, allowing the head to accurately seek to data while the platters are moving so fast.
Each platter has its data organized in concentric circles, called tracks. There are many tracks, not a single spiral-shaped track like you would see on a CD, DVD, or FMD. When you have multiple platters stacked up in a drive, you can refer to tracks neighboring each other vertically as cylinders. They are called cylinders because if you picture the equally sized tracks stacked up, they would look like a cylinder.
A sector is a section of a track. Tracks are divided into sectors so that data on a specific part of a platter can be referred to. Imagine a circular disk first being divided into concentric circles (tracks) then being cut up into equal sized sections like a pie. The sections cut out of each track are called sectors.
Blocks are groups of bytes on a hard drive which are handled, stored, and accessed together as a single logical unit. All data is organized in blocks at a time; if only a portion of one block is used, the entire block is effectively used up.
Here are some pictures of my dissection of a Seagate hard drive.
Hard drive, meet your maker
“Product Warranty Will Be Void if This Label is Removed”
Here’s me peeling that sticker off
Labels removed, Torx screws removed
IDE Interface Controller Board Removed
Off comes another warranty-voiding sticker
Cover removed, platters, servos, actuator arm, and heads visible
Close up of the servo/actuator arm assembly
Another of the servo/actuator arm assembly
Close up of the central platter hub
There is a Neodymium Boride magnet (rare earth) inside the hard drive
used to control the servo. Here I’m using a quarter to hold the hard drive off
the table.
Here you can see the two platters inside the drive. This is an old hard
drive that uses horizontal mapping and contains only 50 megabytes per platter.
The blue wires on the arms lead to the tiny heads above and below each platter.
The actuator arm and head assembly is removed. Each “head” is actually
a combination of two heads.
Hard drives contain rare earth magnets, which are very strong. Here it
is supporting a half-full recycle bin.
6.3.2
Access time is a big issue in hard drives. Delays before data is processed in a hard drive are referred to as latency. There are many causes for drive latency.
Seek time is the time it takes for the hard drive to locate data by means of head movement. This could occur because the requested data operation is on a different track than the one the heads are currently focused over. Seek time is minimized when the track that needs to be found is adjacent to the current track. A worst case scenario would be where the heads need to seek from the outermost track to the innermost track or vice versa. Physically smaller disks have smaller average seek times because of this. Most modern drives use the smaller 3 ½” form factor rather than 5 ¼”.
Rotational latency is the delay created when the heads are waiting for the correct sector to arrive under the actuator arm. If the data needed has just passed under the head, then it needs to spin almost one full rotation again before it can access that data. Because of this, drives with faster rotational speeds generally have lower rotational latencies and higher throughput. Most desktop drives spin at 5400 RPMs, but 7200 RPM drives are becoming more popular. Seagate is known for its Cheetah series drives, which spin at 10 and 15 thousand rotations per minute.
Transfer time is the amount of time it takes for a hard drive to read or write a given amount of data.
6.5.8
Serial ports are some of the oldest connection methods used for external devices in computers. While they were once used for practically everything from mice, to modems, to printers, no one manufactures devices that make use of the old-fashioned 9 pin serial ports anymore. In fact, it’s gone the way of the ISA slot and is no longer available on many newer computer motherboards, such as ABIT’s MAX series. However, the old 9 pin style serial ports are not the only serial interfaces available; newer developments, such as USB, USB 2.0, and IEEE 1394 (also known as FireWire and iLink) also use serial interfaces to communicate. Serial devices send data one byte at a time. Picture a line of sherpas marching in a row; they’re all going to the same destination, but only one will meet its destination at a time. Speed often suffers in serial applications, but costs are kept down because only one conductor is needed to carry data in one direction. Prior to sending any data down a serial interface, a serial port sends a start bit to signify the beginning of a data stream. It is a single bit with a value of 0. Some connections also send parity bits. Serial ports are often referred to as COM (communication) ports, and are full duplex. To provide full duplex, multiple conductors are needed, one for each direction.
Serial ports also require a computer’s parallel data to be converted to serial data. This is done by a controller chip called a UART (Universal Asynchronous Receiver/Transmitter). It accepts parallel data into a buffer and outputs it one bit at a time.
Parallel ports
are traditionally used for printers. However, they have also been used in the
past for scanners, CD burners, external storage devices, and for data security
dongles. When a parallel interface is used, data is sent side by side,
typically 8 bits at a time. Picture the same line of sherpas as previously
mentioned, but marching in a row side by side instead of one in front of the
other. Parallel interfaces are typically faster than serial interfaces. Speeds
are generally between 50 and 100 kilobytes per second on a traditional parallel
port, but later developments, such as the
6.5.9
Different speeds of different devices can vary quite a bit. Here is a list of some common computer devices with their speeds:
Enhanced
Extended
IEEE 1284
Super Enhanced
ATA33 Hard drive – 33MB/s (theoretical)
ATA66 Hard drive – 66MB/s (theoretical)
ATA100 Hard drive – 100MB/s (theoretical)
ATA133 Hard drive – 133MB/s (theoretical)
Ultra160 LVD SCSI Hard drive – 160MB/s
USB 1.0/1.1/1.5 Cable/ADSL/xDSL/RADSL/DSL/SDSL/etc modem – 1.54MB/s (theoretical)
Firewire/IEEE 1394 External hard drive – 50MB/s
USB 2.0 External hard drive – 60MB/s
T1 Internet connection – 1.54MB/s
T3 Internet connection – 5.625MB/s
v.92 Dialup modem – 57.6KB/s
32 bit 33MHz PCI bus - 133MB/s
6.5.10
Interconversion between analog and digital formats is essential to computer operation. The real world works in analog, facts are not discrete digital numbers. The images you see on your screen, input you put into your mouse, and the sound that comes out of your speakers are all analog in the real world where it interfaces with you, but digital inside the computer.
When you move your computer’s mouse, an analog distance is being measured in digital increments by either a set of wheels attached to a ball, or by a camera comparing images taken at set digital intervals. As best as we know, the precision in a physical mouse’s position is infinite. However, the computer must choose a precision at which to measure it. In a mechanical mouse, the mouse’s changes in position are measured by rolling a ball on the bottom as it moves. A set of wheels determine its motion in the vertical and horizontal axis, and their velocities are in turn used to determine the mouse’s velocity. These numbers are measured to a set precision; the position of the wheels used to calculate velocities are sampled at a set frequency. A typical serial bus mouse will refresh at 60Hz, a PS/2 mouse will be 80Hz (although it may be manually adjusted up to 200Hz) and a USB 1.5 mouse will be 120Hz.
Audio is another great example of how Interconversion between analog and digital formats takes place. Suppose you plugged a microphone into your computer’s sound card and spoke into it. A microphone has a piezoelectric crystal inside of it that produces a current which is proportional to any vibrations it is subjected to. The current that is carried down the microphone’s wire is an analog signal. The sound card’s D/A (Digital/Analog converter) would have to process the analog signal into numbers. Consider the sound waveform below:
The analog sound wave is a continuous shape; it cannot be reproduced 100% accurately in numbers. To replicate it digitally, samples are taken, in that the data is chopped up into parts of a defined size, and the waveform’s position is measured at those intervals, as seen below.
The system can now be represented with numbers using Cartesian coordinates. Usually the samples are taken a set distance apart, so it is not necessary to keep track of x coordinates, only the y coordinates. The y coordinates can be measured with varying precision. CDs use 16 bit audio, as do most computer applications. Professional sound mastering, such as DigiDesign’s Pro Tools, uses 24 bit audio. The dots can now be connected to simulate the original waveform.
As you can see, the digital copy is not perfectly true to the original; vertices are present, and there are no true curves. These artifacts can be reduced using two different methods. One is to increase the sampling rate, which would reduce the size of any artifacts, and another is to increase the precision at which the waveform “heights” are sampled. Both methods have to be used together for optimal results; CD audio is recorded at 16 bits, 44.1 kHz. Professional audio recording is done at 24 bits, 96 kHz.
Sound can also be moved via an optical cable, which has only two states; on and off. Optical cables are fibre optics, plastics which are made to hold light inside of them. When sound is transmitted over an optical cable, digital interfaces are needed to send and receive the data, and this data must be processed into analog data before it can be passed onto an amplifier, then speakers, and heard.
Without these analog-digital conversions, it wouldn’t be possible to store audio data on a computer, or even to process the audio data.
6.5.11
Advantages and Disadvantages of Digital Data
|
Advantages |
Disadvantages |
|
Data integrity – Digital data uses discrete numbers to store data, so it does not “wear out” over time. Theoretically, digital data lasts forever. |
Quality – Digital data can not perfectly replicate the analog data that people interface with, so the quality is rarely |
|
Clarity – Since digital data is held in discrete numbers, transfer of the data is simple. The numbers can be copied with perfect precision without any further loss of quality. |
|
Advantages and Disadvantages of Analog Data
|
Advantages |
Disadvantages |
|
Quality – A true sine wave cannot be reproduced digitally; old vinyl records will sound superior to their digital counterparts given that the parts are all in perfect working order. |
Data integrity – Analog data wears away over time and usage. Photographs, for example, fade with time, become tattered and worn out. Records get their grooves worn out and lose the data they once held. |
|
|
Replication – Analog data cannot easily be copied. To recreate a piece of analog data, some quality is lost. When an old analog photocopier is used to make a copy of a picture, it loses some quality. When a copy is made from the first copy, even more quality is lost. |