[time-nuts] HP Stories: Frequency Counter business decline, Modulation Domain Analyzers, geeks as models

[Rice, Hugh (IPH Writing Systems)]

Hello Time Nuts,

Hopefully some of you will find this snapshot of HP's "Santa Clara Division", the home of Frequency Standards and Frequency Counters interesting.   Send me a note if you want me to keep writing old HP stories.


When I arrived at HP Santa Clara Division (SCD, Division 02, as in the second HP division ever) in 1984, the beginning of the end of the "frequency" business was starting to be felt.

This product line dates to the 1950's with the invention of the first frequency counter.   Breakthrough for its time, it enabled precision measurements of frequency, using "digital" techniques.   The basic idea:  Open a "gate" for 1 second, and count the number of positive zero crossings of a signal.  The count was the frequency to 1Hz resolution.    Longer gate times (10 seconds) would give more resolution, etc.   This required the invention of fast digital counting circuits (essentially a chain of flip-flops, likely implemented with the very earliest discrete transistors), some binary to decimal conversion for the display, and there you have it:  Something that can measure frequency.

Like all breakthrough measurements, the next 20+ years was spent refining the concept.    Better "front end" amplifiers, able to measure low amplitude signals.   Faster counters, that could capture higher frequency signals.    Microwave front ends, to either divide the signal by 8 or 16 or whatever before being counted, or using some other down conversation technique.    Somewhere in the early 1970's (guess), the reciprocal counter was invented.  Rather than open the gate for 1 second,  use the input signal to open the gate for one period, and count a very high frequency time base instead.    It was much faster, and for lower frequency signals, much much higher resolution.   Now, if you had a ~1KHz unknown signal, rather than only getting 4 significant digits in a one second measurement, if you counted a 100MHz time base for 1 millisecond, you would 5 digits of resolution.  Open the gate for a thousand cycles, and 8 digits of resolution were available.

With the reciprocal counter, it was essentially measuring a time interval (one period of the input signal), and calculating frequency.    The obvious extension was an instrument to measure time intervals directly, using two inputs.  Channel A to start and Channel B to stop.  And the "Universal Counter" was born.

All of these required an accurate frequency source to precisely open the gate, or to be counted in reciprocal counters or time interval measurements.   This is what drove the market for PFS - precision frequency sources.    A progression of improved quartz oscillators were invented, and then the cesium standards starting in 1965.   Now, from HP you could buy it all:  The perfect time base, and the high resolution counter.   All your frequency measurement dreams supplied from wonderful HP.       This was a great compliment to HPs frequency generator product lines.  (Remember HPs first product in 1939 was the 200A audio oscillator.)    A frequency counter, using a precision time base, was able to calibrate (or perhaps tune) an oscillator or frequency generator to create a complete system to generate very accurate signals.    Lots of great applications for all of this.   Growth, profits, happiness.  I'll guess that the mid 1970's were the peak of the business.    In 1984, SCD was a ~$100M business per year, with frequency counters being over half of it.   (SCD also made laser interferometers, which was about ¼ of the business)

But as the 1970s became the 1980s, things started to go flat.   Frequency sources (Cesium standards, etc.) last for a very long time.  Did everyone who needed one already buy one?    We were also running out of new tricks to put into frequency counters.   And they lasted for a long time too.   New products were mostly just modern implementations of older models, with better computer interfaces for automated control, and cheaper to build.   Every general manager knows that he'll never grow his business replacing older products with newer, but cheaper replacements.

To make matters worse, the cold war was ending, and D.O.D. spending was slowing down.
To make things still worse, the signal generators now were "synthesizers", and included their own precision time bases.   No counter needed to measure the output.
And even worse, digital oscilloscopes were getting better, and they measured frequency right on the front panel, to 3 or 4 digits.   Which was often good enough.

SCD needed a new trick.    Divisions that don't grow become crummy places to work.  Especially for upwardly mobile managers.     A huge investment was made in "Sandblaster", which was  wafer level IC tester.  This lasted a couple of years and was cancelled.   "Waveform Recorders" were also invented and brought to market.  I think they were essentially a very deep memory digital oscilloscope, back when lots of DRAM memory was really expensive.  I think Rick Karlquist worked on these for a bit.   I don't think they really took off, and probably had a "charter war" conflict with the oscilloscope division in Ft. Collins(?) Colorado.

The premium universal counter in the early 1980's was the 5370A/B.  It's specialty was extremely high precision time interval measurements, with resolutions down to 20 pico seconds.     An idea was made to extend the time interval measurement into continuous real time, and measure dynamic time intervals, as a function of time.   HP called this "Modulation Domain" analysis.   I think that David Chu, one of the senior scientists at SCD in the 1970s and 1980s was a key inventor of this.     The 5371A was the outcome of this work.    It was followed by the 5372 and 5373.    There also was a lower cost version call the 53310 ("Stonehenge"?).     The measurement was novel and unique.   You all know what oscilloscopes and spectrum analyzers do, but this was a new spin:  You could look at the phase of a signal as a function of time.

During the heyday of Modulation Domain Analyzers, HP would take promotional pictures for sales brochures.   The photographer would sometimes grab a HP employees to be in the photo shoot.   Jim Cole, now an inkjet engineer that I still work with was the model for the 5371A.   I stumbled across his picture in the Keysight company history web pages.  He had long forgotten it, and enjoyed seeing his  younger self from 30 years ago.    In about 1990, I was walking down the wrong hall one day, and offered a ham & cheese sub sandwich if I had a few pictures taken with the 53310A.   This photo of me with the instrument was up in the lobby of the SCD building in Santa Clara for several years.    A colleague rescued it from a dumpster during a renovation, and mailed it to me, as I had relocated to Vancouver WA.   It is now proudly displayed on the ceiling of my shop.   You can't see it this image, but the CD chosen by the photographer for the photo shoot always made me laugh:  An album by "Twisted Sister".

For a number of years, Modulation Domain analyzers really helped the business.   I recall that the 5371 sold about 80 units a month, for about $25K each, or ~$2M of revenue a month.  For a division with less than $10M/month revenue, this was a really big deal.    A glimmer of hope for the division.      We just needed to keep coming with more ideas like this.      But this didn't come to pass.   Modulation domain analysis, with all its novelty, just didn't need to be done that much.   And digital oscilloscopes kept getting better.  Slowly, sadly, precision frequency measurement was becoming a rarely needed thing, and there were already thousands of good HP counters out there that worked just fine.

I was just down in the EE lab at my HP printer R&D site in Singapore last week, and found a 5314A and a 5315A counter on the shelf.    The lead tech thought they were pretty cool, but it was obvious they hadn't been used in decades.    I told him that I use to manage the engineering team that manufactured that product 30 years ago.    In his polite Singaporean way, he acted impressed.

HP growth was eventually found in other businesses (like inkjet where I went 25 years ago), and a colleague (John Flowers) that has stayed with Agilent and then Keysight said that Div 02, SCD was disbanded some many years ago.   It was a great place to start my HP career 35 years ago, but time and technology has marched on and left it behind.

Hugh Rice.










-------------- next part --------------
A non-text attachment was scrubbed...
Name: 5371.jpg
Type: image/jpeg
Size: 50538 bytes
Desc: 5371.jpg
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20190121/0b5f4fc2/attachment.jpg>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 53310 Poster.jpg
Type: image/jpeg
Size: 143125 bytes
Desc: 53310 Poster.jpg
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20190121/0b5f4fc2/attachment-0001.jpg>