Transcript M262 Putting Computers Systems to Work

Arab Open University – Lebanon – Spring 2011
1
T325: TECHNOLOGIES FOR
DIGITAL MEDIA
Block I part2: Information storage
1
Outline
2




Introduction
Rotating media
Optical media.
Solid-state memory
Arab Open University – Lebanon – Spring 2011
Introduction
3
 Rotating media (hard disk drives (HDDs), CDs and
DVDs), and semiconductor memory (flash
technology), has seen remarkable development
over recent years.
 In this part we will be looking in more detail at:
 how rotating media and semiconductor memories work?
 why it has been possible to make such remarkable
improvements?
Arab Open University – Lebanon – Spring 2011
Rotating media
4
 How did they survive competition with semiconductor memories?
 On the face of it, it is rather surprising that rotating media such as magnetic
hard disks, CDs and DVDs, with their high precision drive mechanisms, have
survived competition from semiconductor memories, which have the clear
advantages of simple construction with no moving parts to wear out or get
damaged by mechanical shock, together with fast access times.
 In fact, magnetic disks have made remarkable leaps in capacity and
reductions in physical size as a result of major innovations, and this has
enabled them to retain their place as large read/write data stores; while
optical media continue to offer a cheap and effective way of distributing data
offline and making backups.
Arab Open University – Lebanon – Spring 2011
Disc or Disk ?
5



Both spellings are common
Disk for magnetic media (Hard Disk)
Disc for optical media (Blue-ray disc)
Arab Open University – Lebanon – Spring 2011
Rotating media – Magnetic disks
6

Magnetic media have a ferromagnetic surface on which
areas can be magnetised (some metals, such as iron and its
alloys, may be magnetised).
 Information may be encoded as a pattern of magnetism
 This pattern can subsequently be detected and the information
retrieved
 Because the patterns persist until the material is demagnetised or remagnetised, the medium is ‘non-volatile’ and does not depend on a
continual supply of power, unlike many semiconductor memories
Arab Open University – Lebanon – Spring 2011
Rotating media – Magnetic Disks
7
 In a magnetic disk drive, a write head converts
electrical signals to magnetised areas on the disk
surface, or changes an already magnetised area.
 Conversely, a read head produces an electrical signal in
response to a magnetic field.
 The two heads are constructed as a single assembly
and mounted on an arm which can move radially across
the spinning disk, so that any point on the recording
surface may be reached.
 Data is recorded in concentric tracks, with each track
being written or read with the heads at a certain
nominal radius
Arab Open University – Lebanon – Spring 2011
Rotating media – Magnetic disks
8
Arab Open University – Lebanon – Spring 2011
Rotating media - Electromagnetism
9
 Close relationship between electricity and magnetism was not
discovered until the 19th century
 Any electrical conductor carrying a current has associated with
it a magnetic field.
 For example, the wiring in your house has a magnetic field around it
when current is flowing, though it is rather a weak one
 The live and neutral conductors carry equal currents in opposite directions, so
their magnetic fields tend to cancel each other out.
 If a wire is wound into a coil and a current is passed through it, then
the magnetic fields produced by each turn of the coil all add
together, making a much stronger field.
Arab Open University – Lebanon – Spring 2011
Rotating media - Electromagnetism
10
 Hans Christian Oersted (1777--1851) noticed that a compass needle was deflected by
a nearby wire carrying an electric current, demonstrating a connection between
magnetism and electricity.
 Investigating further, he concluded that a magnetic field encircled the current
carrying wire.
Arab Open University – Lebanon – Spring 2011
Rotating media - Electromagnetism
11
 André-Marie Ampère (1775-1836) showed that two current
carrying wires attracted or repelled each other (depending on
the direction of the currents) due to their magnetic fields. So,
magnetic forces could be obtained without any permanent
magnets being involved.
 Ampère formulated a mathematical law linking electric currents
and magnetic fields and showed that, by winding wire into a coil
or ‘solenoid’, a strong magnetic field like that of a bar magnet
could be obtained.
Arab Open University – Lebanon – Spring 2011
Rotating media - Electromagnetism
12


Wrapping wire around a bar magnet did not
appear to produce an electric current.
Michael Faraday (1791-1867) demonstrated that
electric currents could be produced by a magnetic
field as long as the field was varying with time,
an effect expressed mathematically in his law of
magnetic induction.
Arab Open University – Lebanon – Spring 2011
Rotating media
13
 Essentially, a write head is an electromagnet
constructed so as to concentrate the magnetic field
over a very small area of the surface of the disk.
 If there is sufficient current through the write head,
the magnetic field is intense enough to magnetise
this area permanently (or until the next write).
 By varying the current as the disk rotates under the
head, a pattern of magnetism is built up along the
track, corresponding to a stream of bits.
Arab Open University – Lebanon – Spring 2011
Rotating media
14
 Is reading procedure the converse of the operation ?
 It does, but there is a subtle and important difference.
 Faraday’s law of magnetic induction means that an unchanging magnetic
field does not generate a current; what is needed is a changing field.
 As the disk rotates under the head, the magnetic field it encounters does
indeed change frequently, with consecutive areas being magnetised in
either of two directions.
 So, a head of this type does produce a voltage signal, but this signal
represents the transitions between magnetic states rather than the states
themselves.
 This is allowed for in the way that data bits are encoded for writing and
subsequently decoded for reading.
Arab Open University – Lebanon – Spring 2011
15
Rotating media
Giant magneto-resistance (GMR)
 Discovered by Albert Fert in France and Peter Grϋnberg in
Germany, working independently.
 They jointly received the Nobel Prize in Physics in October 2007 for
their discovery.
 The effect of magnetoresistance is a change in the electrical
resistance of a conductor when it is placed in a magnetic field.
 Relates the voltage across a conductor to the current flowing through it in
accordance with Ohm’s law: V
=IxR
 So R can be measured by applying a known voltage to the conductor
and measuring the current through it
Arab Open University – Lebanon – Spring 2011
16
Rotating media
Giant Magnetoresistance
 Magnetoresistance for a read head mechanism:
 Changes in resistance are, in principle, rather easier to detect than the
tiny amounts of energy that the magnetic bits of a high density disk can
generate by induction.
 The power used for the measurement now comes from an external
voltage (or current) source.
Arab Open University – Lebanon – Spring 2011
17
Rotating media
Giant magneto resistance
 Although magnetoresistance was observed by William Thomson (Lord
Kelvin) 150 years ago, the change in resistance of below 1% was too small
to be practically useful.
 The real breakthrough came with the discovery of GMR, when huge
resistance changes (e.g. 50%) were reported.
 GMR read heads were one of the first products of nanotechnology.
 The term nanotechnology covers a wide variety of ideas and techniques,
but its defining characteristic is that it deals with objects with dimensions
in the order of a nanometre (10 - 9 m, or a millionth of a millimetre).
Arab Open University – Lebanon – Spring 2011
Rotating media
18
Perpendicular magnetic recording technology
 Another technique which has been used to increase a
real densities is perpendicular magnetic recording.
 In conventional recording, or longitudinal magnetic
recording, the magnetic bits may be thought of as tiny
bar magnets which point along the track.
 In perpendicular recording, the bar magnets are
aligned at right angles to the disk and point in and out
of the disk surface.
Arab Open University – Lebanon – Spring 2011
Rotating media
19
Arab Open University – Lebanon – Spring 2011
Rotating media
20
 Materials used for the recording surface should have a much
higher ‘coercively’.
 Coercivity is a measure of how difficult a material is to
magnetise or demagnetise.
Arab Open University – Lebanon – Spring 2011
Optical media
21
 While the capacity increases in optical storage
have not been as spectacular as those in magnetic
disks and semiconductor memory
 there have been steady improvements from the
audio CD standard agreed back in 1979 (the ‘Red
Book’ standard), with a capacity of less than a
gigabyte, to newer formats capable of storing high
definition movies of many gigabytes.
Arab Open University – Lebanon – Spring 2011
Optical media
22
 A laser beam is focused onto the track and reflected back.
 The reflected beam is analyzed to detect the pits. Feedback mechanisms
(servomechanisms, or ‘servos’ for short) involving optics, mechanics and
electronics ensure that the beam is kept focused on the track and does not
drift away from the track.
 They compensate effectively for minor warping or eccentricity and for some
vibration.
 The data surface is at the opposite side of a transparent polycarbonate
substrate from the laser reader.
 The laser light penetrates the transparent surface before it reaches a point
of focus, so it is spread over a relatively wide area and, therefore, is less
affected by minor scratches and dust particles.
Arab Open University – Lebanon – Spring 2011
Optical media
23





For read-only CDs, the pits are actual indentations in the data
surface
For rewritable media they are simply areas of different reflectivity
from the ‘land’ that surrounds them.
These discs use a material that can exist in one of two physical states
or phases: crystalline (where the molecules are arranged in regular
patterns) and amorphous (where there is a lack of order). One
phase reflects light better than the other.
The phase depends on how the material is heated and cooled, and
can be changed from one to the other and back again by heat from
the laser.
This is called phase-change recording.
Arab Open University – Lebanon – Spring 2011
Optical media
24
Arab Open University – Lebanon – Spring 2011
Optical media
25
 Blu-ray disc (BD) and HD DVD use blue lasers with a wavelength of
405 nm. As this is shorter than the wavelength of the red laser light
used with earlier formats, the laser beam can be focused to a
smaller spot on the data surface.
 The pits and track pitch (the spacing between tracks) could be
reduced in size, increasing the data capacity achievable.
 Data in Blu-ray is recorded in a similar way to previous rewritable
optical discs, although the details differ.
 The recording surface uses a phase-change material, and the pits
are areas of contrasting reflectivity.
Arab Open University – Lebanon – Spring 2011
Optical media
26

CD and DVD -- essential features The ubiquitous 12
cm diameter silvery discs we use, including DVD in
its various formats, are all descendants of the
original audio CD developed by Philips and Sony,
and the essential principles remain the same. Data
is written on a spiral track as a series of ‘pits’ of
various lengths
Arab Open University – Lebanon – Spring 2011
Optical media
27
 One important design decision was the thickness of the
transparent layer covering the data surface.
 In CDs the whole 1.2 mm thickness of the disc was used,
while the thickness in DVDs was 0.6 mm.
 However, there are optical advantages to be gained by
having a much thinner transparent cover layer.
 If the disc tilts at all, then a thick layer could deflect
and defocus the beam.
 Would having a thinner transparent layer make the disc
more or less resistant to the effects of fingerprints and
scratches? What measures might be taken to reduce
their effects?
Arab Open University – Lebanon – Spring 2011
Optical media
28
 Having a thinner transparent layer means that where the beam




crosses the surface it covers a smaller area.
So, any scratches, fingerprints or particles of dust would be more
likely to block out or distort the beam.
The disc would be less resistant to their effects.
At first it was thought that scratching would be such a problem that
the disc would need to be contained in a protective cartridge, like a
floppy disk. However, a cheaper and less bulky solution was
adopted by giving the disc a very hard coating that would resist
scratching.
Another possible measure is to accept a poorer raw error rate
performance but use better error detection and correction methods
to compensate.
Arab Open University – Lebanon – Spring 2011
Optical media
29
 Unlike the original CD, where the pits were laid in a spiral
pattern on an otherwise flat surface, Blu-ray discs have a
physical spiral track on which the pits are recorded.
 This track is a spiral groove.
 However, it would be no good just having a completely flat
recording surface, as the recording beam would have
nothing to follow.
 The preformed groove is there to guide the beam so that it
deposits pits in the correct spiral pattern.
Arab Open University – Lebanon – Spring 2011
Optical media
30
 Recall that a CD player, or indeed any optical disc
player, uses a servomechanism to stay on track.
 This is a form of closed-loop control where the position
of the beam on the track is constantly monitored.
 This signal is fed back to the motors that position the
beam in such a way as to correct the departure.
 The alternative, open-loop control, does not use
monitoring and feedback, but simply relies on very
precise mechanics and favourable operating conditions
to keep errors to a minimum.)
Arab Open University – Lebanon – Spring 2011
Optical media
31
Arab Open University – Lebanon – Spring 2011
Optical media
32
Arab Open University – Lebanon – Spring 2011
Optical media
33
 As the disc rotates, the servomechanism ensures that the laser follows the




groove. This means that the servomechanism will be moving the laser from side
to side to follow the wobble.
The rate at which it moves from side to side will be determined by the length
of the wobbles (the wobble period – the distance between peaks) and the
speed of rotation.
The length of the wobbles is predefined, so the equipment can determine the
speed of rotation from the rate at which the laser is moving from side to side.
A feedback loop is used to control the speed of rotation, based on measuring
the rate at which the servomechanism is having to move the laser from side to
side.
In Blu-ray, the ‘wobbling’ technique is taken further, and is used to address 64
Kbyte blocks of data on the disc. This is done by modulation of the wobble
signal.
Arab Open University – Lebanon – Spring 2011
Optical media
34

Important features of Blu-ray are as follows:





Blu-ray data is held as a pattern of pits along a track.
The laser beam reading the data is kept on the track and in focus
by a servomechanism.
The track also produces a wobble signal. The wobble information is
used in locating blocks of data.
When data is recorded it is encoded with an error-correcting code
(Reed--Solomon (or RS) code). This allows a large proportion of the
errors to be corrected.
Small dust particles may corrupt only a few bytes, but marks or
scratches might extend for some distance along the track and cause
errors in a long sequence of data.
Arab Open University – Lebanon – Spring 2011
Solid-state memory
35
 If you have been using portable digital equipment over
recent years you can’t fail to have noticed the
extraordinary developments in data storage based on
flash memory.
 Memory keys, memory cards (as used for digital
cameras) and the built-in memory of mobile phones are
all based on flash memory technology.
 Even by the standards of digital technology, the
development of flash memory has been impressive, in
terms of both the storage capacity available and the
cost.
Arab Open University – Lebanon – Spring 2011
Solid-state memory
36
 Flash memory is an example of memory that is described as ‘solid-
state’.
 This use of the term ‘solid-state’ has an interesting history, and
exactly what it implies is a little ambiguous.
 In science ‘the solid state’ means solid as distinct from liquid, gas or
plasma.
 This is the sense in which the term ‘solid-state electronics’ was used,
drawing the distinction between the transistor, which is entirely
made of material in the solid state, and the thermionic valve, which
is based on the behavior of electrons in a vacuum tube.
 The distinction now is with rotating media; the significance of the
term ‘solid’ seems in contrast with the moving mechanical parts of
the hard drive.
Arab Open University – Lebanon – Spring 2011
Solid-state memory
37
Memory basics
 Our main interest is in memory which stores digital media files --
audio, and still and moving images.
 Main memory is comprised of integrated circuits, and these
integrated circuits are of two sorts: ROM and RAM.
 The processor can both write to and read from the other type of
memory integrated circuit used in main memory, RAM.
 The abbreviation RAM stands for random-access memory. This name
arises from the fact that a data word in any location in the whole
memory can be accessed just as quickly as a data word in any other
location.
Arab Open University – Lebanon – Spring 2011
Solid-state memory
38
 An important point about RAM is that it is usually volatile, which
means that all the programs and associated data stored in it are
lost when the power supply to the RAM is switched off.
 This contrasts with ROM and with secondary storage media, both of
which are non-volatile.
 ROM is used for the programs needed on start-up.
 It is therefore necessary to ensure that all important programs and
associated data held in RAM are copied to a computer’s secondary
memory before the computer is switched off.
Arab Open University – Lebanon – Spring 2011
Solid-state memory
39
 Flash memory is a development from a type of memory known as EPROM.
 EPROM is ‘erasable’ PROM, and PROM is ‘programmable’ ROM.
 Non-programmable ROM has the data put into it when it is manufactured.
 Programmable ROM, by contrast, is manufactured with no data stored in it
and can be programmed by the user (the user in this case might be a
computer manufacturer).
 Special equipment is needed to program PROM, and it cannot be
changed once it is programmed (it cannot be erased).
 The term ‘programming’ is used, rather than ‘writing’, for putting data into
these devices, because it is all done at once ‘up front’, rather than writing
individual bits or bytes when needed, although for some devices the
distinction is not so clear-cut.
Arab Open University – Lebanon – Spring 2011
Solid-state memory
40
 There is also EEPROM, which is electrically erasable
PROM.
 As its name suggests, EEPROM is erased with an
electrical signal rather than needing UV light.
 Whereas erasing EPROM has to be done to the whole
of the device (UV light is shone onto the whole of the
‘chip’), EEPROM can be erased one bit or one byte at
a time.
 A disadvantage of EEPROM compared with EPROM is
that EEPROM is more complex and, therefore, costs
more and it is not possible to fit as much memory on a
single integrated circuit.
Arab Open University – Lebanon – Spring 2011
Solid-state memory
41
Flash memory is a variation on EEPROM.


The word ‘flash’ is used in reference to the way it is erased, which is
done either on the whole device at once or on a block of data in the
device and can be done quickly, ‘in a flash’.
Offering the electrical erase capability, traditionally featured by
the expensive EEPROM, at cost and density comparable to EPROM,
Flash memories not only have taken a big portion of their
progenitor’s markets, but in addition they have greatly expanded
the fields of application of non-volatile memories. (Source: Bez and
Cappelletti, 2005, p. 84)
Arab Open University – Lebanon – Spring 2011
Solid-state memory
42
The flash memory cell.
 The memory cell is the part of the memory that can store a single bit.
 To do this the cell must be able to exist in each of two states: one state that
represents a data 0 and one state that represents a data 1.
 Reading a cell is the process of detecting which of the two states it is in.
 Erasing a block of memory sets all the cells in a block to the same known
state, that representing a data 1 (see next Figure).
 Writing to the cells (programming the memory) is the process of setting
some of the cells to the other state, that which represents a data 0.
 Since the remaining cells are already in the state representing a data 1
anything to them there is no need to do anything to them.
Arab Open University – Lebanon – Spring 2011
Solid-state memory
43
The flash memory cell.
Arab Open University – Lebanon – Spring 2011
Solid-state memory
44
 Flash cells use semiconductor technology.
 They are constructed using the same or similar semiconductor materials and
manufacturing processes to those used in microelectronics such as
microprocessors.
 The basic building block of microelectronics is the transistor.
 In flash memory a memory cell uses a single transistor, whereas several
transistors are needed for a single memory cell in some other types of
semiconductor memory.
 The fact that flash only needs one transistor for a memory cell is one of
the reasons that it allows high storage densities.
 In the simplest terms, as used in digital electronics, a transistor can be
thought of as a controlled switch. Next Figure 2.9 illustrates this.
Arab Open University – Lebanon – Spring 2011
Solid-state memory
45
Arab Open University – Lebanon – Spring 2011
Solid-state memory
46
 In very broad terms, there are two categories of transistors: bipolar
junction transistors (BJTs) and field-effect transistors (FETs).
 The terms I used above, namely source, gate and drain, are the
terms used for FETs, which are the transistors that are most widely
used in digital circuits.
 (In BJTs, the roughly equivalent terms are the emitter, base and
collector.)
 The ‘field effect’ in a field-effect transistor is the effect of an
electric field that controls the switch.
 Next Figure illustrates the structure of a FET, specifically a type of
FET known as a metal-oxide-semiconductor field-effect transistor
(MOSFET).
Arab Open University – Lebanon – Spring 2011
Solid-state memory
47
Arab Open University – Lebanon – Spring 2011
Solid-state memory
48
 The substrate is made of a semiconductor material,
such as silicon, as are the source and the drain.
 The substrate differs from the source and the drain,
however, by having different doping.
 Doping of semiconductor material is the addition of
small amounts of ‘impurities’ to change subtly the
electrical characteristics of the semiconductor.
 The substrate will be made into a ‘p-type’
semiconductor and both the source and drain into
an ‘n-type’ semiconductor
Arab Open University – Lebanon – Spring 2011
Solid-state memory
49
 in the absence of any electric field across the
substrate, little or no current will flow between the
source and the drain.
 If there is a big enough electrical field across the
substrate in the right direction, then an electrical
channel is created between the source and the
drain through which electricity can flow
Arab Open University – Lebanon – Spring 2011
Solid-state memory
50
 The basic technique of building silicon integrated
circuits (a process referred to as ‘fabrication’) consists
of manufacturing a thin disc, referred to as a wafer,
out of a crystal of pure silicon, then modifying it or
adding material to the surface to create the transistors
or other components.
 For example, adding impurities to the silicon (‘doping’
the silicon) changes its electrical conductivity.
 Polycrystalline silicon is used where good electrical
conductivity is needed, and silicon oxide is used where
a good electrical insulator is needed.
Arab Open University – Lebanon – Spring 2011
Solid-state memory
51
Arab Open University – Lebanon – Spring 2011
Solid-state memory
52
 The flash cell differs from a basic MOSFET by having two gates.






One of these has no electrical connections to it (it is surrounded by insulators) and is
described as a floating gate.
The gate above it, which does have a connection, is called the control gate.
Though the floating gate has no electrical connections to it, if it does in some way
get electrical charge on it, then the charge will affect the electrical properties of
the transistor.
Specifically, the presence of charge on the floating gate alters the field between
the gate and the substrate, which in turn alters the resistance of the channel
between the source and the drain.
This, then, is the principle that is used to store a data bit. The cell has two states.
In one (representing a data 1) there is no charge on the floating gate and in the
other (representing a data 0) there is charge on the floating gate.
Arab Open University – Lebanon – Spring 2011
Solid-state memory
53
 What has to be done to erase a cell? What has to be
done to program a cell?
 From earlier, erasing involves setting all cells to the
state representing a data 1.
 This is the state when there is no charge on the floating
gate; so, erasing a cell requires the removal of charge
from the floating gate.
 In programming, setting a cell to the state representing
a data 0 requires getting charge on to the floating
gate.
Arab Open University – Lebanon – Spring 2011
Solid-state memory
54




What is needed, therefore, are mechanisms to add charge to the
floating gate and to remove charge from it.
In essence the ‘trick’ is simply to use high voltages and/or high
currents to force charge on or off!
Applying a high voltage between the control gate and the substrate
forces electrons on or off the floating gate through a process known
as Fowler-Nordheim (FN) tunnelling
Alternatively, there are methods which impart high energy to
electrons which cause them to ‘jump’ through the insulator. These
methods are described as ‘hot electron’ or ‘hot carrier’ methods.
Arab Open University – Lebanon – Spring 2011
Solid-state memory
55




To read data from a cell, a mechanism is needed to
detect whether or not there is charge (electrons) on the
floating gate.
To read data from a cell, you apply a voltage to the
control gate and see whether it succeeds in allowing
current to flow between the source and the drain.
If it does allow current to flow -- if current is detected
at the drain -- then the data on the cell is a 1;
if it does not allow current to flow -- if no current is
detected at the drain -- then the data on the cell is a 0.
Arab Open University – Lebanon – Spring 2011
Solid-state memory
56
Varieties of flash
 Such is the demand for flash memory that
manufacturers produce a large range of products,
customised to specific applications.
 Devices can be designed to be optimised for different
applications through, among other things, the details of
the cell design, the configuration of the cells on the chip,
and the electronic circuits that accompany the
memorycells.
 It is impossible -- and would be of limited value -- to
discuss all variations, but we will briefly look at some
products which will highlight a number of features.
Arab Open University – Lebanon – Spring 2011
Solid-state memory
57
NAND flash memory
 The word NAND comes from binary logic, and is an abbreviation of ‘NOT




AND’.
It is used in reference to the way in which the flash memory cells are
interconnected (which is similar to the way transistors are connected to make
a digital electronic NAND gate).
In my description of the flash cell earlier, I said that there would either be
charge on the floating gate or there would be no charge.
I did not say anything about how much charge there might be on the gate.
The simplest flash cells only distinguish between these two options: charge
or no charge.
Multi-level flash is more sophisticated, distinguishing between more than
two levels.
Arab Open University – Lebanon – Spring 2011
Solid-state memory
58
 This allows more than one bit to be stored in each cell, and hence a higher
storage density.
 For example, suppose in a binary cell a data 1 corresponds to no charge
on the gate and a data 0 corresponds to an amount of charge we will call
C.
 If the system is able to add charge more precisely and also measure how
much is on the gate accurately, it might be possible to work with four levels
of charge: 0, C/3, 2C/3, C. These four levels could be used to represent
two bits of binary data, defined as, for example:
 0 charge = 11
 C/3 = 01
 2C/3 = 00
 C = 10
 Memory using this system would be described as ‘2 bits per cell’ memory.
Arab Open University – Lebanon – Spring 2011
Solid-state memory
59
Arab Open University – Lebanon – Spring 2011
Solid-state memory
60

How does the random access time of this NAND
flash compare with that of a HDD that has a latency
time of 4.17 ms?
The average random access time (latency) for the
HDD was 4.17 ms, which is a lot longer than the 60
micro second for the flash.
This is a ratio of

Arab Open University – Lebanon – Spring 2011
Solid-state memory
61
NOR flash memory
 The word NOR comes from binary logic, and is an
abbreviation of ‘NOT OR’.
 As with NAND in NAND flash, NOR in NOR flash is
used in reference to the way in which the flash
memory cells are interconnected within a memory
chip (which is similar to the way transistors are
connected to make a digital electronic NOR gate),
but for our purposes it is simply a label.
Arab Open University – Lebanon – Spring 2011
Solid-state memory
62
 Whereas NAND flash is optimised for bulk storage, similar to the
function of rotating media, NOR flash is optimised for storage of
computer code.
 We are more interested in this course in memory for bulk storage,
so we will not consider NOR flash in as much detail as we did NAND
flash.
Arab Open University – Lebanon – Spring 2011
63
Solid-State memory
Activity 2.9


Figure 2.14 is a simplified display of the
memory array configuration (it ignores the
use of two planes).
Use information from the data sheet in
Figure 2.13 to find:



n, the number of pages per block
m, the total number of blocks.
From these numbers, calculate (giving
separate answers for main memory and
spare memory in each case):



the total number of bits in a page
the total number of bits in a block
the total number of bits in the whole
memory. (This should, of course, agree with
the total memory size of 16 Gbits of main
memory and 512 Mbits of spare memory.)
Arab Open University – Lebanon – Spring 2011
Solid-state memory
64
The layout of memory cells in NAND flash allows memory cells to be packed
more closely together than in NOR flash.
According to a paper by Bez and Cappelletti (2005), if the technology node is
F, then the space required by a NOR cell is 10F2 and the space required by
a NAND cell is 4.5F2.
If the technology node F is 65 nm, use these formulae to calculate:


The cell size of flash NOR and NAND memory for binary cells (1 bit per
cell).
The number of bits that can be stored on a chip that is 5mmx5mm.
Arab Open University – Lebanon – Spring 2011
Solid-state memory
65
(a) F = 65 nm = 65 x10-9 m.
NOR cell size is 10F2, which is 10 x(65 x10-9)2m2
= 4.2 x10-14 m2.
NAND cell size is 4.5F2, which is 4.5 x(65 ·10-9)2m2
= 1.9 x10-14 m2.
 (b) 5 mm square is (5 x10-3)2m2 = 2.5 x10-5m2.
The number of NOR cells that will fit in that area is
(2.5 x10-5)/(4.2 x10-14) = 5.9 x108. That is 560 Mbits (with M in the sense of
220 = 1048 576).
The number of NAND cells that will fit in that area is
(2.5 x10-5)/(1.9 x10-14) = 1.3 x109. That is 1.2 Gbits (with G in the sense of
230 = 1073 741824).

Arab Open University – Lebanon – Spring 2011
Solid-state memory
66






The NAND device has a much larger capacity than the NOR device (16
Gbits compared with 1 Gbit).
The random access time for the NOR flash is much shorter than for the
NAND flash (96 ns compared with 60 ms).
NOR can program individual words, whereas NAND needs to program a
page at a time.
NAND and NOR programming times are similar, being of the order of tens
of microseconds.
It takes much longer to erase a block in the NOR flash than in the NAND
flash (1 s compared with 2.5 ms).
The endurance of the NOR flash product is around 10 times that of the
NAND flash (100 000 cycles compared to 10 000 cycles).
Arab Open University – Lebanon – Spring 2011