ELEC3020 - Lecture 4

← Back to Home

ELEC3020 - Lecture 3
Memory - Slides

Types of Memory

ROM $-$ Read Only Memory

• Read only
• Keeps data when powered down
  RAM $-$ Random Access Memory
• Read/Write access
• Loses data when powered down

---

ROM

• Keeps data when powered down
• Contents cannot be changed
• Used for system boot

Types of ROM

• ROM $-$ Read Only Memory
  • Can never be changes
• PROM $-$ Programmable ROM
  • Comes empty
  • Can be ‘programmed’ but changes are permanent
• EPROM $-$ Erasable Programmable ROM
  • Original state can be restored through UV exposure
• EEPROM $-$ Electronically Erasable Programmable ROM
  • Can be electrically programmed/erased many times

ROM Implementation

The basic schematic of a ROM implementation consists of three primary components:

• Address Inputs ($a_0,a_1,a_2,\dots$): specify which byte in memory is selected
• Decoder: an $n$-to-2${}^n$ decoder takes $n$ input address lines and activates exactly one output row line corresponding to that address
• Data Bus / Column Lines ($d_0,d_1,\dots ,d_7$): the output lines that display the byte stored at the activated address line

The horizontal lines are connected to any number of vertical lines through either a diode or transistor connection. When a singular horizontal line is powered by the decoder, it activates the connected vertical output lines.
 
As we can see, the diode/transistor connections are persistent and will remain the same even if the ROM is powered off.

---

RAM

• Loses data when powered down
• Content can be changed at any time
• Can store programs or data

There are two types of RAM: static RAM and dynamic RAM. SRAM is more expensive and works as intended. DRAM is cheaper, but ‘leaks’ memory contents over short time period -> requires periodic ‘refresh’ every memory cell.

RAM Implementation

RAM implementation is similar to ROM, but we replace the connection element '$\bullet$' by an active switch.

DRAM has only one Field-Effect Transistor (FET) per bit but requires a ‘refresh’ operation every ~64ms for the whole memory.

DRAM Transistor Connection:
SRAM uses 6 FETs per bit using CMOS (Complimentary Metal Oxide Silicon) technology.

---

Memory Size

One Byte = Eight Bits

Memory Sizes

• Kilobyte (KB = $2^{10}$ Bytes = $1024$ Bytes)
• Megabyte (MB = $2^{10}$ KB = $2^{20}$ Bytes)
• Gigabyte (GB = $2^{10}$ MB = $2^{30}$ Bytes)
• Terabyte (TB = $2^{10}$ GB = $2^{40}$ Bytes

Note: upper-case symbols (KB, MB, etc.) follow traditional computing memory sizes (KB = 1024), but lower-case symbols (kB, mB, etc.) are rounded using decimal math (kB = 1000).

We call a set of 8 data lines a data bus. Instead of running two bytes of memory on two bus lines, we instead reuse the same line. Only one device is allowed to send data at a time. We ensure this through the use of tri-state buffers.

Given our address lines have a bus size of $n$, we can store $2^n$ bytes!

Memory Control

We use several control lines to control our memory chips:

• Chip-SELECT’ -> activate memory chip
• Output-ENABLE’ -> write data to bus
• Read/Write’ -> activate writing (RAM only)

These control lines turn ON when they are unpowered (logical ‘0’) which is indicated by the ${}^{\prime }$ in their name. This is the standard for most memory chips. You can set a control line to always be ON by routing it to ground.

---

Address Space

A memory cell’s width is called a word. We can calculate the total size of a memory module through the formula:
$\text{size = number of cells * cell size}$

Memory module sizes are always in powers of 2. We increase the number of address lines by one starting with one line for 2 cells.

The number of data lines required is equal to the number of bits in the memory cell.

---

ELEC3020 - Lecture 5