[time-nuts] HP Stories: An architectural view of the HP 5060/5061 and awkward oscillator adjustments.

Sat Feb 23 12:48:46 UTC 2019

Hello Time-Nuts,

The 5060/5061 Cesium Beam Frequency Standard was an amazing product from an architectural perspective.    As a new engineer first seeing the 5061A, I had no appreciation for the brilliance of the overall system.    As Rick Karlquist noted in the story about battery chargers, the system architecture is the foundation of any good instrument.   The 5060A was 20 years old when I joined HP in 1984, so I have no personal insights into the development team or process.   But looking back at the 5060/5061 system decades later, now as an experienced system engineer, I have a great appreciation for the system approach.

Several things illustrate this:

The modular design, with reasonable portioning of functions and good interfaces between the modules, made it easy to break the instrument down into bite size pieces for design and troubleshooting.   This modular approach, with each sub system having a clearly defined purpose (and likely specific module specs), enables a team of engineers to develop in parallel, rather than relying on one guy to do it all.   (Based on Rick's stories, I can image Len playing sheep dog running around making sure that everyone did their job right, and holding the whole system together.)     From a troubleshooting perspective, problems are easy to isolate to a specific module, which then can either be swapped out, or repaired at the component or PCB level.

The front control panel shows great insight into what is needed to keep the instrument in good working order.   The built in meter, with sensors all through the system, allowed fairly in depth diagnosis and tuning, without removing the covers or needing any instruments or tools.   (See control panel images.)     The different modes offered from the master control switch were well thought out.   "CS off" to allow oscillator only operation, to save the life of the tube.   "Loop open" to allow tuning of the sub systems, etc.   The design of the panel was done with the end user in mind.

The 5060A and 5061A products stood the test of time.   The overall design of the 5060A was carried into the 5061A.  Start to finish, it lasted 25 years as the benchmark of commercial frequency standards, until the 5071A replaced it in 1991.    Electronics technology progressed tremendously during those 25 years, but the basic 5060/5061 design continued on.

It was fantastically reliable.   Only linear power circuits, with robust heat sinking of all power devices.   The legendary Len Cutler ban on aluminum electrolytic capacitors.    5060s were still in use in 1985, after 20 years of constant operation.  Likewise, 5061As were abundant in time standards for 25+ years until they were replaced by the 5071A in the 1990s.

The longevity presented problems for guys like me in manufacturing, as described in other HP Stories.   Because HP was committed to servicing products in the field, any module redesigns had to work in the old system.    The 5060A only had about 5  years of products life, and a fairly fresh start was used on the 5061A in the early 1970s.   But when I arrived to work on the 5061B, there had been 15 years of "de-evolution" to the 5061A.   The thrash around the oscillator, the heart of the Atomic Clock, was the most dramatic.

The original 5061As had a "105" 5MHz crystal oscillator, about the size of a small brick.   All was good.   But when the 105 was replaced with the 10811A in about 1980 (?), it got messy.    The 10811A was a 10MHz oscillator, but the 5061A was designed around the 105 oscillator's 5MHz signal.    The first task was to make a 10811 look like a 5MHz oscillator.   An auxiliary circuit board was designed that divided the 10MHz output by 2, and the resulting 5MHz was routed through the system.   Rob Burgoon, one of the mentors to the 5061B team did this design.   I recall him describing how he used an old, slow, "regular"  TTL flip-flop as the primary divider, because it had lower noise than the L, LS or ALS newer family TTL devices.    Sure enough, Rob's design made a 10811 look electrically like a 105.     Thus the 5061A "de-evolution":   The 10MHz oscillator was divided by 2 to make 5MHz, to source the unchanged RF Frequency Multiplier and Synthesizer, which together fed the Harmonic Generator to create the magic Cesium 133 frequency of  9,192,631,772 Hz.   But this same 5MHz signal is then multiplied by 2 to supply the 10MHz precision frequency front panel output.    This architectural travesty used to bother me greatly.   But to supply the 10MHz directly unmodified to the from panel would have required redesigning the Multiplier, Synthesizer, Buffer Amplifier, and Frequency Divider modules.   HP decided it wasn't worth the effort.

The new 10811 module had to fit mechanically too.  To do this, the 10811 was mounted to an aluminum sheet metal bracket, which provided mounting flanges that matched the 105B attachment process.   (HP always used aluminum sheet metal in their instruments.  Because it was just better.   In my budget constrained inkjet world, sheet metal is all steel.  As thin as possible.)   But there was still one more problem.   HP oscillators had a precision mechanical tuning capacitor, to get the free running frequency close before closing the control loop on the Cesium Standard.   (This was an awesome, air dielectric precision device.  See the picture of the inside of the 10811.)   The 10811A adjustment cap didn't align with the 5061A control panel access hole positioned for the 105B oscillator.

It is amazing how difficult it is to design a solution to turn a screw, when the access hole is one inch out of alignment.    The 5061A kludge soliton was an S-shaped flexible shaft, enclosed in a steel tube, with a screwdriver blade like thing fastened to the oscillator end, and a screw head object on the control panel end.    This created a customer problem though.   When turning the flexible shaft driven oscillator tuning cap, the flexible shaft acted like a torsional spring a bit, and sometimes would twitch and change the capacitor setting, causing a jump in the frequency and/or 1PPS outputs.  Also, being in contact with the capacitor all the time, mechanical disturbances could cause noise in the capacitor.    Customers didn't like this.    (Unfortunately, I couldn't find a picture of this system.  Any of you have a 5061A service manual and want to share a scan?)

I worked with a couple of mechanical engineers to fix this during the 5061B program.   The objective:  Find a way to tune the capacitor, but when the tuning was done, NOTHING CAN BE TOUCHING THE 10811 CAPACITOR!   Two solutions were invented, shown in all the pictures.  For the 5061A retrofit, a spring loaded, gear driven offset mechanism was invented.   One shaft aligned to the control panel, the other to the oscillator tuning cap.     For the 5061B, we invented a spring loaded adjustment screw mechanism in a machined plastic housing, that would spring back after the adjustment was done.   The 5061B access hole on the control panned was moved down an inch, and this was a straight shot.   It was fun working with the MEs on these designs, and trying them out to make sure they worked as intended.

Architecturally, the nice clean original 5061A design had degraded, due to module serviceability backwards compatibility constraints, into the mess we shipped with the 5061B.    But it all worked, and development and servicing resources were minimized, which was the bigger priority.

Whenever I see what looks like a crazy stupid design in a product now, I remember the 5061B, and consider: "Maybe this engineer wasn't really an idiot.   He or she may have had some outside constraint this isn't obvious in the finished product."

Hugh Rice    23 February 2019

-------------- next part --------------
A non-text attachment was scrubbed...
Name: HP 5061A control pannel.jpg
Type: image/jpeg
Size: 58723 bytes
Desc: HP 5061A control pannel.jpg
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20190223/ff166680/attachment.jpg>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: HP 5061A gear drive oscillator adjustment.JPG
Type: image/jpeg
Size: 109990 bytes
Desc: HP 5061A gear drive oscillator adjustment.JPG
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20190223/ff166680/attachment.jpe>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: HP 5061B 10811A.jpg
Type: image/jpeg
Size: 109355 bytes
Desc: HP 5061B 10811A.jpg
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20190223/ff166680/attachment-0001.jpg>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: HP 5061B Control Pannel.JPG
Type: image/jpeg
Size: 136026 bytes
Desc: HP 5061B Control Pannel.JPG
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20190223/ff166680/attachment-0001.jpe>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 10811_tuning_cap2.jpg
Type: image/jpeg
Size: 23211 bytes
Desc: 10811_tuning_cap2.jpg
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20190223/ff166680/attachment-0002.jpg>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 10811A.jpg
Type: image/jpeg
Size: 53662 bytes
Desc: 10811A.jpg
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20190223/ff166680/attachment-0003.jpg>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: HP 5060A Front-view.jpg
Type: image/jpeg
Size: 59595 bytes
Desc: HP 5060A Front-view.jpg
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20190223/ff166680/attachment-0004.jpg>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: HP 5061A _top view.JPG
Type: image/jpeg
Size: 23304 bytes
Desc: HP 5061A _top view.JPG
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20190223/ff166680/attachment-0002.jpe>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: HP 5061A 105B.jpg
Type: image/jpeg
Size: 227364 bytes
Desc: HP 5061A 105B.jpg
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20190223/ff166680/attachment-0005.jpg>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: HP 5061B Oscillator Adjuster.JPG
Type: image/jpeg
Size: 61695 bytes
Desc: HP 5061B Oscillator Adjuster.JPG
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20190223/ff166680/attachment-0003.jpe>