Dr. William R. Huber, PE (Retired)

Courses By:
Bio:
Print Brochure

Presents a SunCam continuing education series on the topic of:

Memories in Computers

Contact author at:
w.huber@ieee.org
108 Lewallen Court
P.O. Box 320
West End, NC 27376
(910) 673-7383

Dr. Huber obtained a Bachelor of Science degree in Electrical Engineering from the University of Pittsburgh in 1962, and graduated summa cum laude. He obtained a Master of Science degree in Electrical Engineering from The Ohio State University in 1963 and a Doctor of Science degree in Electrical Engineering from the University of Pittsburgh in 1969.

Dr. Huber worked from 1962 to 1982 for Bell Telephone Laboratories. At Bell Labs, he managed and guided design, development and introduction to manufacture of large scale, high density integrated circuits. In particular, he developed and applied the concept of redundancy to semiconductor memory chips, thus reducing the impact of manufacturing defects on yields and costs. For his work, he received the Best Paper Award at the prestigious 1979 International Solid State Circuits Conference (ISSCC). Virtually all modern integrated circuit memories now use redundancy for yield improvement and cost reduction. He also worked at the GE Microelectronics Center and at Harris Corporation. Since 1994 he has been a technical consultant and expert witness in over 45 patent litigation cases involving semiconductor memories.

Dr. Huber has presented over 20 papers and invited panel presentations in major industry publications and conferences. He holds 3 patents, and is a registered Professional Engineer (Retired) in Florida and North Carolina.

Course Title Rating Hours Price

102-Memories in Computers — Part 1: Overview and DRAM Introduction

(4.2) 15 Reviews

4 $90.00

Add 3-Course Set & SAVE 20% Add Course to Cart

Course Objectives: This continuing education course is written specifically for professional engineers with the objective of relating to and enhancing the practice of engineering.

Course Description:

Modern computers include many types of memories, including magnetic hard drives, Flash memories, Static Random Access Memories (SRAM) and Dynamic Random Access Memories (DRAM). The characteristics, advantages and limitations of each of these memory implementations are described and compared. The most important of these memory types is DRAM, and this course takes you inside the DRAM so you can understand how it operates and why it is the largest selling integrated circuit ever invented.

DRAMs were a $38B (USD) business in 2010, and appear in every computer (PC, Mac, tablet, laptop, desktop, server or mainframe) and home gaming console (PlayStation, Xbox or Wii). DRAMs first appeared on the scene in 1970, introduced by a small start-up company called Intel. Today they are a commodity, produced by companies all over the world in essentially interchangeable form.

Both the original DRAM architecture, called the asynchronous DRAM, and the higher speed synchronous version, the SDRAM, are discussed in detail. You will be introduced to the concepts of refreshing, access time, multiplexed addresses, fill frequency, CAS latency and burst length. No math beyond simple arithmetic is required.

Outline

Memories in Computers
DRAM History
DRAM Architecture and Basic Operation
Accelerated Access Modes
Fill Frequency
SDRAM Architecture and Operation
SDRAM Operation
Access Time in SDRAMs
Data Burst Operation
Multiple Data Pins
Multi-Bank Architecture
Mode Register
Memory Standardization

111-Memories in Computers — Part 2: DDR SDRAMs

(4.5) 4 Reviews

4 $90.00

Add 3-Course Set & SAVE 20% Add Course to Cart

Course Objectives: This continuing education course is written specifically for professional engineers with the objective of relating to and enhancing the practice of engineering.

Course Description:

As the largest dollar volume portion of the semiconductor industry, memories appear in every computer, printer, home game console and car; as well as a growing number of home appliances. Total memory sales were $124B in 2017, out of $412B worldwide semiconductor sales. DRAM sales alone grew by 77% in 2017, to a total of $72B. But memory demand is cyclical. 2018 saw DRAM sales of almost $100B, the highest ever. 2019 sales dropped to $62B; 2020 sales were $67B; and 2021 revenue soared to $98B.

Since the late 1990s, four increasingly capable Double Data Rate (DDR) Synchronous DRAM architectures have evolved. The first generation, DDR, doubled the rate of information flow to and from the memory device as compared to single-data-rate synchronous DRAMs. To meet the resulting very tight timing requirements, a new circuit called the delay locked loop was introduced.

The second generation, DDR2, added several features to improve device usability. In particular, on-die termination (ODT) improved signal integrity and reduced external component count.

The third generation, DDR3, provided a refined ODT capability and further defined external device timing in terms of clock cycles instead of internal device parameters.

The fourth generation, DDR4, continued increasing storage capacity and improving the rate of data transmission. But it also emphasized reducing power by many architecture modifications and feature additions.

The fifth generation, DDR5, built on DDR4 with more storage capacity, faster data rates, and reduced power. DDR5 added a new capability, on-chip error correction (ECC).

This course discusses the characteristics and advantages of each of those architectures in a clear and concise manner that any technically trained person can understand. You will be introduced to and become familiar with concepts such as CAS latency, burst length, delay locked loops, on-die termination, prefetch, mode registers and redundancy. Memory packaging in dual in-line memory modules (DIMMs) is also discussed.

This course builds on, but is independent of, Memories in Computers—Part 1. No math beyond high school algebra is required.

112-Memories in Computers — Part 3: Flash Memory

(4.8) 4 Reviews

4 $90.00

Add 3-Course Set & SAVE 20% Add Course to Cart

Course Objectives: This continuing education course is written specifically for professional engineers with the objective of relating to and enhancing the practice of engineering.

Course Description:

Flash memories will account for $26B in sales in 2010. They are used in all modern digital cameras, camcorders, cell phones, PDAs, music players, home video game machines and computers.

Why is flash so popular? Flash excels in three areas: technical, physical and financial. The technical reasons are (1) nonvolatility–flash memory retains its information for more than 10 years even with no power applied and (2) speed–flash memory is 100 to 1000 times faster than magnetic hard drives. The physical reasons are (1) density–flash memory has 8x as many bits per chip as DRAM, and (2) power–flash memory consumes far less power than magnetic hard drives. In the financial area, flash memory is about 40% of the cost of DRAM on a per bit basis.

Why isn't flash memory used in place of DRAM and magnetic hard drives? Again, there are three areas that hold back flash: reliability, speed and cost. Regarding reliability, flash memories can endure from 5000 to over 100,000 erase/program cycles. That is adequate for many but not all applications. Flash speed is indeed much faster than magnetic hard drives, but programming speed is far slower that DRAMs; much too slow for main memory applications. And flash cost, although lower than DRAMs, is still far above that of magnetic hard drives.

This course discusses all of the above issues, as well as the historical background, physical basis, cell structure, and chip architecture and operation of these omnipresent devices. You will learn and understand hot electron injection, Fowler-Nordheim tunneling, NOR and NAND array structures and operation, multi-level storage and error correction coding. The flash memory market, including applications and major producers, are explored; and two major portions of that market, flash memory cards and solid state drives, are examined in detail.

This course builds on, but is independent of Memories in Computers–Part 1. No high-level math is required.