[telecom] What Hath Bell Labs Wrought? The Future

THE IDEA FACTORY Bell Labs and the Great Age of American Innovation

By Jon Gertner

422 pages. The Penguin Press. $29.95

What Hath Bell Labs Wrought? The Future

By MICHIKO KAKUTANI March 19, 2012

In today's world of Apple, Google and Facebook, the name may not ring any bells for most readers, but for decades - from the 1920s through the 1980s - Bell Labs, the research and development wing of AT&T, was the most innovative scientific organization in the world. As Jon Gertner argues in his riveting new book, "The Idea Factory," it was where the future was invented.

Indeed, Bell Labs was behind many of the innovations that have come to define modern life, including the transistor (the building block of all digital products), the laser, the silicon solar cell and the computer operating system called Unix (which would serve as the basis for a host of other computer languages). Bell Labs developed the first communications satellites, the first cellular telephone systems and the first fiber-optic cable systems.

The Bell Labs scientist Claude Elwood Shannon effectively founded the field of information theory, which would revolutionize thinking about communications; other Bell Labs researchers helped push the boundaries of physics, chemistry and mathematics, while defining new industrial processes like quality control.

In "The Idea Factory," Mr. Gertner - an editor at Fast Company magazine and a writer for The New York Times Magazine - not only gives us spirited portraits of the scientists behind Bell Labs' phenomenal success, but he also looks at the reasons that research organization became such a fount of innovation, laying the groundwork for the networked world we now live in.

...

formatting link
formatting link

***** Moderator's Note *****

I'm always wary of lionizing organizations or people that I don't know personnaly. Bell Labs made mistakes as well as breakthroughs, and it was always cursed by a gap between its blackboards and its breadboards. The feeling of the scientists and engineers at the labs was that manufacturing was Western Electric's business, and since the labs were actually a subsidiary of WE, there was a potent driver for a "We just design 'em" culture that limited many of the Labs' projects to theoretical advances.

The gap between the Labs and the field caused some very severe errors: most notably, the choice of an 8,000 Hz sample rate for the T-1 carrier system. Although it was Shannon's work that predicted the need for a sample rate of at least twice the highest frequency to be reproduced, the T-1 designers didn't do the field trials which would have revealed that sampling ambiguities at even sub-multiples of the "chop" rate would make 1,000 Hz test tones unusable. This is the reason that "Milliwatt" test tones are now slightly offset from 1,000 Hz - but I can only imaging the cost of rework and redesign that far-reaching error caused.

Bill Horne Moderator

Reply to
Monty Solomon
Loading thread data ...
***** Moderator's Note *****

Could you elaborate on this in layman's terms--particularly the statement-- "sampling ambiguities at even sub-multiples of the "chop" rate would make 1,000 Hz test tones unusable"?

It would seem to me that 8,000 Hz sample rate would make sense because it was double the voice band of 4,000 Hz.

Thanks.

***** Moderator's Note *****

I didn't write that clearly: it's not "even sub-multiples", it's "rational sub-multiples", i.e., sub-multiples that evenly divide into

8,000 Hz.

Bill Horne Moderator

Reply to
HAncock4

The problem with these frequencies is that they do not toggle all the bits in the encoding, thus testing at those frequencies is incomplete. Accordingly, the test sets for digital circuits test at 1015 Hz, wihich causes all the bits to toggle in the encoding.

There is nothing "magic" about 1 kHz. It does sort of correspond to the peak energy in voice (even though every voice is a bit different). Note that any sampling rate would have the same problem with frequencies that are, as Bill put it, "even sub-multiples" of the sampling frequency.

I doubt that most engineers would consider this a big screw up by Bell Labs, and certainly there are lots of other cases where the impact was greater. All in all, they did a pretty good job of designing the telephone network...

ET

Reply to
Eric Tappert

Every "milliwatt" test tone I have seen is set at 1004 hertz. I am not saying there is no such thing as a 1015 hertz source, just that this is the first I am hearing of it. Maybe I just need to get out more.

Over the years I have heard at least three different explanations for why the 1000 hertz test tone was changed to 1004 hertz. It is certainly true that sampling a 1000 hertz tone at 8000 samples per second will result in the exact same eight patterns, repeating ad nauseum. [assuming that the test tone is stable]. This is one of the explanations I have heard, i.e., that it is not a rigorous enough test for the A/D converters and/or the actual line drivers. I have always found this somewhat dubious, as my experience with these things is that they either work or they don't. Timing slips and jitter were the primary problems in the days of true DS-1 signaling over copper, and testing often involved sending "all ones."

Another explanation I have heard is that 1000 hertz was never a problem until ESF came along. Extended Super Frame finds the framing bit by looking for the one bit out of 193 that consistently alternates between a "one" and a "zero" with each frame. Apparently the 1000 hertz tone would result in actual time slot bits doing the same thing, and bumping it up to 1004 solved the problem. I am not saying this is fact, just relating what I have heard over the years.

Jim ================================================== Speaking from a secure undisclosed location.

Reply to
Jim Bennett

I didn't look it up but I do recall that 1015 Hz was used, as was 1004 Hz (thanks for reminding me...). Perhaps the 1015 Hz was the European version. In any event, any small offset would do as all the bits in the encoding would change over a reasonable time frame.

As for the framing issue, to generate an error would require that a line error caused a loss of framing while the channel tone test was active, then the re-framing algorithm locked on to the channel signal instead of the "real" framing bit. This would cause, when the tone disappeared, another loss of framing error and recovery. That scenario is highly unlikely. However, if it occured the voice channels would be messed up until the tone disappeared.

As for ESF, it used a unique pattern of ones and zeros in the framing bit position to establish froming and determine which frame held the signaling bits. It is even more unlikely that a constant pattern in a time slot would mimic this pattern.

In any event, normal tolerances on test sets and link timing would likely mean that problems anticipated would rarely occur. I suspect the test frequency change was made according to the Bell Labs "belt and suspender" policy...

ET

Reply to
news

Yeah, I never liked the ESF explanation for the change to a 1004 hertz test tone. As you pointed out, ESF uses a unique bit pattern for the framing bits over 24 successive frames. If memory serves, it was the long-abandoned D1 framing [the predecessor to plain old Super Frame] that simply alternated the framing bit with each frame. Also, as you said, if a stable 1000 hertz tone results in the same eight patterns repeating, then the scenario seems even more unlikely.

So what was the reason? The "belt and suspenders policy" seems possible, but there must have been a reason for it, even it was based on a purely theoretical problem. Bill's original statement about "sampling ambiguities" is still stuck in my tired old brain - Bill, could you elaborate on this?

Jim ================================================== Speaking from a secure undisclosed location.

Reply to
Jim Bennett

The following article describes the T1 system in technical detail, including 'framing' and synchronization. On page 1440, they describe how they use the 1,000 Hz tone for testing purposes, and other test procedures and equipment for T1.

formatting link
The BSTJ says T1 was based on earlier experiments, including field trials, of the PCM (Pulse Code Modulation) system. A January 1962 BSTJ article provides a technical overview of experimental results:
formatting link
The January 1962 BSTJ issue has several PCM articles in addition to the above. Table of Contents:
formatting link
(Note that the BSTJ articles run from 12 meg to 25 meg).

Reply to
HAncock4

Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.