Sep 27, 2013

Analog to Digital Convergence? Not So Fast!


In 1996, Dan Sheingold, the editor of Analog Devices’ magazine, Analog Dialogue, wrote an introduction in the “Editor's Notes” entitled: Analog Dialectic.
A little more than 30 years ago, the undersigned held a responsible technical marketing position with George A. Philbrick Researches, Inc., and was editor of a journal called "The Lightning Empiricist." For the edification of a sizeable portion of our readers who weren't around at the time, GAP/R, a company with annual sales of $6M, virtually owned the operational amplifier market-such as it was. The only significant competitors were then a lot smaller-Tom Brown's Burr-Brown Research Corporation, and Al Pearlman and the late Roger Noble's Nexus Research Laboratory, Inc.Aye, what's in a name? Do you find it interesting that "Research" was part of all three names? The founders of a new company, Ray Stata and Matt Lorber, did. They believed that the existing companies were getting prices* that would interest only researchers (who had much fatter budgets in those days), and that a significant industrial volume market for operational amplifier modules with a wide variety of performance was waiting to be tapped (ICs with decent performance were still pretty far over the horizon, so you didn't need venture capital to buy an expensive fab; you could still build modules in your own garage, or in this case, a loft-or was it a basement?-in Cambridge, MA).So they gave their company the highly descriptive and serviceable name, Analog Devices, Inc., and offered high-performance op amps at reasonable prices, along with strong application support. They were right, of course! The company's sales took off almost immediately. And ADI was aided by an incredible stroke of luck-both Philbrick and Nexus were acquired by the giant Teledyne, Inc., and merged. As often happens in such circumstances, the effect was like to putting two resistors in parallel; their joint effectiveness actually decreased. Not only that, but ADI was able to acquire the services of a number of talented, and perhaps disgruntled, former employees of Teledyne Philbrick Nexus, a move that proved synergistic. The rest is history! Should the next sentence read: "And with its primacy in analog devices, ADI lived happily ever after?"*For example, the P2 solid-state parametric electrometer was priced at $227 (in mid-1960s US dollars!)
Sheingold was a brilliant editor and I love the historical context.  But that wasn’t his point for this editorial.  Sheingold went on to discuss DSP as a complement to analog circuitry.  Dialectic is a form of argument (e.g., the Socratic method) and the argument started well before 1996 and continues on today.  Digital will make Analog obsolete, says one side.  Digital increases the demand for Analog, says the other side.  From my vantage point, the argument gets more and more irrelevant as time goes on.  The analog circuit design techniques involved in today’s “digital” chips is often the differentiating factor due to the incredibly fast switching speeds, ultra-low voltages and dynamic power management requirements.  The digital circuitry inside the most “analog” of analog chips is growing such that Verilog is no longer a foreign language.  No, from my vantage point the chasm between analog and digital peaked in 1978 with the AD574 12-bit monolithic Analog-to-Digital Converter (ADC) – thank you, Paul Brokaw.  The 10-bit predecessor, AD571 was introduced in the IEEE Journal of Solid-State Circuits, (Vol. SC-13, pp. 736-745, December 1978) as the first complete monolithic ADC.  The AD574 was much more popular and widely copied.  It had a co-bonded reference and required an external sample-and-hold amplifier (SHA) but it represents the turning point in my mind for analog and digital integration.  Prior to that, everything was separate.  Since then, the floodgates were open and the two continue to merge and blur the distinction.  It has been 35 years, so perhaps "floodgates" is not the right image.  This is analog, after all.

I welcome your opinion, so please add your comments below.  Also, as always, if you feel something is incorrect or incomplete, please let me know and I make the necessary corrections!

NOTES:
“Analog Dialectic” Analog Dialogue, Volume 30, Number 3, 1996 (editor Dan Sheingold)
“The Data Conversion Handbook”, Analog Devices, edited by Walt Kester

Sep 26, 2013

Quick Transition Through Transitron

I often think about how many students at schools like Stanford and M.I.T. walk up the same steps as some of the analog legends. To the schools, it’s an honor to have so many people share those footsteps. Fairchild Semiconductor also has a rich history of footsteps leaving to start great companies. But another company was a virtual revolving door of brilliant engineers on their way to make history elsewhere. Few people remember Transitron.

Boston-based Transitron Electronic Corporation started in 1952 and found their early success in making diodes for the military market. Having secured the Western Electric transistor license they started into the semiconductor market sometime in 1953. By late 1954 Transitron had secured cross-licensing approval with several larger manufacturers. By 1955 Texas Instruments and Transitron were the leaders of the transistor industry with over 35% of the market. Its stock went from 36 to 60 in six months, and its sales from $7.4 million to $47 million between 1956 and 1960. According to TIME magazine in a Dec. 21, 1959 article, “only 30 minutes after being placed on the market, the first public offering of 1,000,000 shares of Transitron Electronic Corp. at $36 each was snapped up by investors. Not since the first public sale of 10.2 million Ford Motor Co. shares in 1956 has a stock issue attracted such broad public demand.”

As obscure as they now seem, their success was just as unorthodox. Transitron was founded in a former bakery by an eccentric pair of brothers named David and Leo Bakalar. One of the pair ran a shoe factory; the other was a Ph.D. who had worked at Bell Labs on transistor research. They recruited engineers to their firm by traveling through Europe and conducting interviews in major cities. They hired the brightest immigrants they could find, brought them to the United States, and worked them silly in the early days of discrete electronics components.

In alphabetical order, here are some of the more famous men who quickly passed through Transitron:

Hans Camenzind, from Switzerland, moved to the U.S. to attend Northeastern University. He worked at Transitron, Tyco Semiconductor (acquired by P. R. Mallory), then joined Signetics. He founded Interdesign, which he later sold to Ferranti (later GEC Plessey) and then went to Array Design Inc. He is famous for designing the popular 555 timer for Signetics and authoring several books on circuit and system design.

Wilf Corrigan, born and raised in Liverpool, England, emigrated to the U.S. to work as a production engineer at Transitron. Wilf remembers someone from Transitron coming into a room, where he was reading about semiconductors, and telling him that the production line was a mess: "They told me to get out there and get it running right, so I went, and I figured out semiconductors later." Transitron was that kind of place. He then went to Motorola and later to Fairchild Semiconductor with Lester Hogan after the departure of Robert Noyce in 1968. He co-founded LSI Logic Corporation.

Nick DeWolf, “a distant relative” of Ben Franklin, graduated from MIT, worked as an engineer for General Electric, became chief engineer at Transitron and then co-founded Teradyne with Alex d'Arbeloff.

Dave Fullagar, after earning a MSEE from Cambridge University and working for Ferranti in Edinburg, Scotland, emigrated to the U.S. to join Transitron. He then moved to Fairchild Semiconductor and eventually designed the µA741. Together with Jack Gifford and Fred Beck, he co-founded Maxim Integrated Products.

Pierre Lamond, from France, saw an employment ad in the New York Times for Transitron. They put him on the production line. In his third week, he was promoted to the head of production to replace his departing boss, and in a few more months, he was promoted to device development engineer. He briefly returned to France but, by 1959, he was back at Transitron as the head of development. In 1961, Lamond joined Fairchild Semiconductor, then in 1967, he joined National Semiconductor.

Tom Longo, founder of Performance Semiconductor, worked General Telephone who acquired Sylvania Electric, worked at Fairchild Semiconductor and Transitron.

Bob Swanson, was a founder of Linear Technology Corporation, VP and GM at National Semiconductor; he worked at Fairchild Semiconductor and Transitron. "I worked at Transitron for almost four years from 1960. I was in my fourth year at Transitron working on a PNP mesa structure transistor - this was the hot stuff then. I was trying to second source a Fairchild product. Fairchild, of course, was now a bright star on the West Coast and a friend of mine gave me an article that Fairchild had announced this thing called the planar process. It was like - ‘we're dead’”.

Les Vadasz started working at Transitron, then led the digital MOS technology sector at Fairchild where the man who had hired him into Transitron, Pierre Lamond, hired him again. Finally, he was part of the founding management team at Intel. At Transitron, Vadasz found himself building whatever he needed. Need vacuum chambers to distribute evaporated dopants? Build them yourself, out of a piece of tube. Need some silicon wafers? Build your own furnace; dump in chunks of silicon lumps; start it spinning; dip in a piece of seed crystal, rotating it in opposite directions; pull it out slowly; and hope for the best. Everything else pretty much required working under a microscope with instruments held in your own shaking hands. Everybody who passed through the Boston sweatshop seems to have come out the better for having been there, and for having left.

George Wells, from Scotland, worked at Transitron, then Fairchild Semiconductor, General Electric, eventually LSI Logic, and then he became CEO of Exar.


NOTES:

1. From the Computer History Museum profiles

2. “They Would Be Gods”, Upside, October 2001

3. Computer History Museum, abstracted from Rob Walker's 2006 interview with founder and CEO of Linear technology Bob Swanson, http://silicongenesis.stanford.edu/complete_listing.html

Sep 24, 2013

Analog Circuit Design, an Updated View

My previous post was based on a book written late in the previous millenium, so you would be right to expect that the attitudes and process of designing analog circuits has changed dramatically. Certainly, the knowledge base and the tools have changed in the past 30-odd years.

I was fortunate to find an article about analog design tools in Electronic Design from 2011. After a round-up of the tools, Don Tuite interviewed a design engineer at Linear Technology to see how the tools are used in circuit design today. Don asked Leonard Shtargot about his personal experience with analog design tools. He eventually answered the question, but first gave us some insight into the whole process of analog circuit design:

“LTC (Linear) is a really interesting place because a lot of the folks that work here were either part of the original wave of innovation that created the first analog ICs or the next generation. So seeing how their minds work about designing integrated circuits and solving problems is really valuable,” he said. 
“As an undergraduate engineering student, I learned analysis techniques, and how existing circuits work, but there wasn’t an emphasis on creating new circuits. Working at Linear, I learned the missing 70% or 80% about creating something new,” he said. 
“Analysis is certainly important, but doing something new requires the human mind to build abstractions in terms of the various building blocks, abstractions that are simple enough so you can synthesize them to do what you want. That’s a tool that you can only learn from people who’ve been doing this for decades. The people who did these things originally looked at circuits from their physical embodiment.”
Shtargot didn’t hold much hope for a future generation of analog EDA tools.
“Another thing you can actually learn from more experienced designers is what I’d call intuition. Current EDA tools don’t contain enough ‘intelligence’ to make an analog circuit work right the first time, regardless of the number of simulations you run,” he said. 
“Spice works well if your models are good, but characteristics change with how the device is drawn, or with relative temperatures. That’s hard to model. Essentially, the problem is that if you tried to build a generalized CAD system that could capture any variant of an analog circuit that you’d want to design, it would take much longer than to go and talk to experienced people who can pass their knowledge along and then just build the circuits. Maybe one day programming techniques will get better.” 
“With analog design, a big part of the challenge is defining the problem, defining what you really want,” Shtargot said. 
“One of the things that we learn from folks here who have a long history is that, when you visit a customer, you have to ignore what the customer said he wants, and listen to his problem. It’s often better to apply the fundamentals to solve the problem, rather than try to force an existing part into a role it wasn’t intended for.”  
While not quoting Gilder, he seemed to echo the ideas that they cannot rely on some prefabricated system of knowledge and that electrical engineering will not teach them everything. But Shtargot stopped short of saying that intuition is something that must be learned mostly by themselves. Instead he emphasized the continuity of learning from those who came before, following in their footsteps (figuratively, and possibly literally too).

NOTES:

1. "Contrasts Mark Analog Design Tool Use", Don Tuite, Electronic Design, Oct. 24, 2011

Sep 23, 2013

George Gilder on Analog People

Decades ago I picked up a book in the discount bin, "Microcosm" by George Gilder.  I started to read it but put it down and didn't pick it up again for several years.  In the meantime, I had changed from just working in high-tech, to working for an analog company.  Who knows why I started reading it again, but it was interesting this time.  Then I came upon a chapter called "Analog People".
"The first difference is their intense individuality.  Analog designers must learn their trade mostly on their own.  In reconciling macrocosm with microcosm, they cannot rely on some prefabricated system of knowledge.  Neither digital logic nor electrical engineering will teach them how to bridge the gap -- to create an amplifier that operates partly in the solid state and partly in the state of California, that is ruled on one level by the microcosm and on the other by Governor Deukmejian.  It is lore and intuition that the creators of new circuits must learn, mostly by themselves.  There is no mold by which to multiply their numbers; they threw away the mold when they made Bob Widlar."

Gilder credits Widlar as saying, "You can't force linear designs, each one is a separate invention of its own."  By this point in my career, I understood profitability and the value that analog circuits could command.  Expecting to read about Widlar's antics, which I did, Gilder also quoted Widlar discussing an economic aspect of circuit design that is rather insightful,
"There are far better approaches available than directly adapting discrete component designs to microcircuits.  Many of the restrictions imposed by monolithic construction can be overcome on a circuit design level.  This is of particular practical significance because circuit design is a nonrecurring cost in a particular microcircuit while restrictive component tolerances or extra processing steps represent a continuing expense in manufacture."
Invoking another legend,
As Barrie Gilbert puts it: "He understood the medium."

NOTES:

"MICROCOSM, The Quantum Revolution in Economics and Technology" George Gilder, 1989.

Photo credit: Fran Hoffart, http://www.computerhistory.org/revolution/analog-computers/3/156/402

Sep 22, 2013

IC Op Amps to the Left, Vacuum Tube Op Amps to the Right

History is obsessed with "the first" of whatever.  The first European to discover America. The first integrated circuit. The first operational amplifier (op amp).  New waves of historians then challenge "the first".  As school kids we were taught that in fourteen hundred and ninety-two Columbus sailed the ocean blue.  And now we know that Columbus probably didn't discover America.  The battle of who invented the IC was essentially settled as a draw between Robert Noyce and Jack Kilby; a topic for another day. As far as the first op amp, my college textbook [1] used the 741 (credit to Dave Fullagar) but claimed "the first" op amp was the uA709.  Bob Widlar designed the uA709 while at Fairchild Semiconductor and also receives credit for the uA702 as the first commercially successful op amp integrated circuit (IC).  I guess if the uA702 was commercially successful then the uA709 was wildly successful, an adjective that suited Bob Widlar. The uA701 is a footnote in history that we will discuss later. But they were integrated circuits.  Before integrated circuit op amps, there were vacuum tube op amps.

The landmark op amp ICs were developed in California, in what we call Silicon Valley, with notable exceptions.  The vacuum tube op amps were mostly developed on the East Coast.  In keeping with my fascination with historical footsteps, this bi-coastal aspect interests me.  But first, the question to be answered is who made the first vacuum tube op amp?

Credit generally goes to Lee De Forest for the first vacuum tube amplifier.  Historians now debate that.

De Forest was born in Council Bluffs, Iowa, and raised in Alabama.  He attended Yale University and worked generally on the East Coast.  However, in 1911 he moved west to become the research director of Federal Telegraph in Palo Alto, California [2].  There is that bi-coastal thing again - 60 years earlier.  Does anyone have a map that can show how far Federal Telegraph was from the site of the Wagon Wheel?

NOTES:

1. "Microelectronic Circuits", Sedra and Smith, 2nd Edition, 1987
2. "Fred Terman at Stanford", C. Stewart Gillmor, 2004

Sep 21, 2013

The Day I Looked Back

On June 18, 2011, Bob Pease died.  The San Jose Mercury News reported, "An elderly San Francisco man was killed after his car hit a tree...  The driver's 1969 Volkswagen Beetle veered to the right off of the roadway... The 70-year-old was not wearing his seatbelt..."  I knew Bob Pease, not well, but I spent a few years working at National Semiconductor while he was there.  The shocking part of the news was that he had just been attending a private memorial service for Jim Williams who had died the previous week.  I knew Jim Williams, too.  The thought crossed my mind that we need to put bubble-wrap around those remaining famous analog characters before the next one dies!

And then I thought that someone needs to write a book about those famous analog characters.  For months, I thought seriously about writing that book.  I started doing more focused research instead of the casual reading I had done in the past.  But I'm not an author.  At least not in the way people of my age think of an author.   Today, however, is the day when everyone is an author and everyone can publish their epic masterpiece.

To the outside world, Bob Pease was an elderly San Francisco man.  To those of us connected through the analog segment of the semiconductor industry, Bob Pease was a larger-than-life character.  His persona was a bit eccentric, perhaps.  Jim Williams was another larger-than-life character but more approachable and subdued.  Both were widely known because they authored a great deal of material on analog circuits - authored in the way people of my age think of an author.  Many young engineers learned the practical aspects of circuit design by reading their works, complementing their education.

Bob Pease joined National Semiconductor in Santa Clara, California, in 1976.  Longtime readers of his know that he spent many formative years at Philbrick (George A. Philbrick Researches) which he joined after studying at the Massachusetts Institute of Technology (MIT).  In 1979, Jim Williams joined National Semiconductor.  After leaving Philbrick.  After being at MIT.  It struck me that they walked down the same sidewalks, up the same steps and down the same hallways.  Literally in the same footsteps.  Of course, as we'll share in later posts, they did not follow the same footsteps, figuratively.

Bob Pease visited Linear Technology several weeks before he died.  He walked down the hallway to see Bob Dobkin; he probably walked right past Jim Williams' office.  Literally in the same footsteps as Jim did that day.

A few weeks after Jim died, I attended a tribute hosted at the Computer History Museum.  It was a panel discussion with several key people from Jim's life.  It's strange listening to people reflect on times and events that I shared.  That's the day I looked back and saw the footsteps.


NOTES:

1. Bill Schweber, "Analog expert Bob Pease dies in tragic accident," EE Times, June 20, 2011, "http://www.eetimes.com/document.asp?doc_id=1259751".