Doug Kerr
Well-known member
I was interested to read, a few weeks ago, about the award of the Nobel Prize in physics to Willard Boyle and George Smith for their invention, at Bell Telephone Laboratories, of the charge-coupled device (CCD). This was truly inspired work, and resulted in an elegant solution to a very important need of the time - a solution from which we here especially continue to benefit.
But I was dismayed to read that this was described as the invention of the first practical digital image sensor. It was not.
The charge-coupled device, as developed by those inventors, and as described in their definitive patent, did not involve imaging at all. What they invented was a shift register, one capable of manufacture in large sizes by the relatively-primitive integrated circuit semiconductor technology of the time. Its originally-intended use was for data storage.
Shift registers as implemented with discrete components involved at least four transistors per cell - possibly eight. Implemented via that era's integrated circuit technology, a shift register of any substantial size would have been prohibitively bulky and costly. (Today it is of course hardly a big deal).
Boyle and Smith developed a shift register made essentially of only two capacitors per "cell", with the capacitors linked with those on either side by a special kind of gate.
By properly placing a two-phase clock on some buses leading to all these gates, the charges in the "main capacitors" of the cells (which represented individual data numbers - not necessarily digital, and binary - perhaps analog values) were made to "flow" to the adjacent "auxiliary" capacitors, and thence to the next "main" capacitors. Thus, the charges were handed from cell to cell in "bucket-brigade" fashion.
By observing the voltage representing the charge in the "last" cell, we receive a serial readout of all the data numbers.
Data could be written in essentially the same way. The numbers to be stored were sequentially applied to the "first" cell, and the sequence was shifted until the entire register was filled.
During this same era, other workers were trying to develop a practical image sensor (largely for use in a proposed videotelephone, the "Picturephone"). The workers realized that an array of photodetectors could readily be fabricated using reachable integrated circuit technique, but were frustrated by the question of how to read out several thousand cells. (Again, even a shift register approach looked impractical owing to the number of transistors needed per cell.)
Then, voilà! The Boyle-Smith charge-coupled device provided an elegant solution. In fact, if the individual photodetectors were of the "photo-capacitor" type, they could just be made the "main" capacitors of CCD cells.
The result, as we all know, was history.
Carla called to my attention at this morning's intelligence briefing (known, to the gastronomically-oriented, as breakfast) an article describing how other former Bell Laboratories employees, familiar with the history, have raised objections.
Their point was not to in anyway demean the gigantic accomplishment of Boyle and Smith. Rather, they were concerned that Boyle and Smith were being given credit for inventing the first practical digital camera sensor. (They had indeed invented a critical ingredient that made it possible.) They point out that the Boyle and Smith patent makes no mention of the use of the CCD in support of an imaging array - nor any mention of imaging at all.
Of course, the general public could much more appreciate the development of the first practical digital image sensor than the invention of the first practical integrated-circuit large-scale shift register! So this mischaracterization is perhaps - understandable.
We all share in this misdirection when we believe that "CCD" describes a type of camera imager. It describes one ingredient - a pivotal, and perhaps even defining, one, of course.
Among other things, the objectors (including Eugene Gordon and Mike Tompsett), feel that this has unfairly diverted recognition from Tompsett, who they claim was actually the developer of the CCD-using imager. In fact, there has been some suggestion that Tompsett (and others) should have perhaps been treated as co-inventors of the CCD on the patent. (I do not opine on co-inventor catfights off-the-clock.)
Again, my point here is also not to demean at all the gigantic contribution of Boyle and Smith, to which we are all indebted, especially in the area of digital sensors. And in fact, so far as I know, it is only in that context that the CCD has found extensive commercial use.
In fact, one thing that was perhaps earlier recognized at Bell Telephone Laboratories than anywhere else is that when we seem to solve one problem, the solution may have its biggest use in solving an instance of that problem not even recognized at the outset, or another problem altogether.
But, as you well know from suffering from my writings, another hallmark of "The Bell Laboratories Way" was to insist on properly characterizing things. The CCD deserves that. And let us all grant belated kudos to Messrs Boyle, Smith, and Tompsett.
As a matter of time scale, all this happened during the latter part of the 1960s (just about the time I left Bell Laboratories, in late 1968). I was little aware of this specific matter at the time.
But I was dismayed to read that this was described as the invention of the first practical digital image sensor. It was not.
The charge-coupled device, as developed by those inventors, and as described in their definitive patent, did not involve imaging at all. What they invented was a shift register, one capable of manufacture in large sizes by the relatively-primitive integrated circuit semiconductor technology of the time. Its originally-intended use was for data storage.
Shift registers as implemented with discrete components involved at least four transistors per cell - possibly eight. Implemented via that era's integrated circuit technology, a shift register of any substantial size would have been prohibitively bulky and costly. (Today it is of course hardly a big deal).
Boyle and Smith developed a shift register made essentially of only two capacitors per "cell", with the capacitors linked with those on either side by a special kind of gate.
By properly placing a two-phase clock on some buses leading to all these gates, the charges in the "main capacitors" of the cells (which represented individual data numbers - not necessarily digital, and binary - perhaps analog values) were made to "flow" to the adjacent "auxiliary" capacitors, and thence to the next "main" capacitors. Thus, the charges were handed from cell to cell in "bucket-brigade" fashion.
By observing the voltage representing the charge in the "last" cell, we receive a serial readout of all the data numbers.
Data could be written in essentially the same way. The numbers to be stored were sequentially applied to the "first" cell, and the sequence was shifted until the entire register was filled.
During this same era, other workers were trying to develop a practical image sensor (largely for use in a proposed videotelephone, the "Picturephone"). The workers realized that an array of photodetectors could readily be fabricated using reachable integrated circuit technique, but were frustrated by the question of how to read out several thousand cells. (Again, even a shift register approach looked impractical owing to the number of transistors needed per cell.)
Then, voilà! The Boyle-Smith charge-coupled device provided an elegant solution. In fact, if the individual photodetectors were of the "photo-capacitor" type, they could just be made the "main" capacitors of CCD cells.
The result, as we all know, was history.
Carla called to my attention at this morning's intelligence briefing (known, to the gastronomically-oriented, as breakfast) an article describing how other former Bell Laboratories employees, familiar with the history, have raised objections.
Their point was not to in anyway demean the gigantic accomplishment of Boyle and Smith. Rather, they were concerned that Boyle and Smith were being given credit for inventing the first practical digital camera sensor. (They had indeed invented a critical ingredient that made it possible.) They point out that the Boyle and Smith patent makes no mention of the use of the CCD in support of an imaging array - nor any mention of imaging at all.
Of course, the general public could much more appreciate the development of the first practical digital image sensor than the invention of the first practical integrated-circuit large-scale shift register! So this mischaracterization is perhaps - understandable.
We all share in this misdirection when we believe that "CCD" describes a type of camera imager. It describes one ingredient - a pivotal, and perhaps even defining, one, of course.
Among other things, the objectors (including Eugene Gordon and Mike Tompsett), feel that this has unfairly diverted recognition from Tompsett, who they claim was actually the developer of the CCD-using imager. In fact, there has been some suggestion that Tompsett (and others) should have perhaps been treated as co-inventors of the CCD on the patent. (I do not opine on co-inventor catfights off-the-clock.)
Again, my point here is also not to demean at all the gigantic contribution of Boyle and Smith, to which we are all indebted, especially in the area of digital sensors. And in fact, so far as I know, it is only in that context that the CCD has found extensive commercial use.
In fact, one thing that was perhaps earlier recognized at Bell Telephone Laboratories than anywhere else is that when we seem to solve one problem, the solution may have its biggest use in solving an instance of that problem not even recognized at the outset, or another problem altogether.
But, as you well know from suffering from my writings, another hallmark of "The Bell Laboratories Way" was to insist on properly characterizing things. The CCD deserves that. And let us all grant belated kudos to Messrs Boyle, Smith, and Tompsett.
As a matter of time scale, all this happened during the latter part of the 1960s (just about the time I left Bell Laboratories, in late 1968). I was little aware of this specific matter at the time.
