[time-nuts] Z3801A EFC error

[Robert Benward]

To Jarl and others,

A few months ago I wrote about my Z3801A having a EFC error and an inability to generate a lock.  With the help of Jarl and others, I was able to correct the situation.  I simply disassembled the double oven (what a mess!) and made a manual adjustment to the frequency.  

Note the fix was simple, but the oven disassembled is a nightmare.  The connections to the internal oven is made via thin enameled wires laid upon what appears to be a long (18") foam rubber strip.  The enameled wires are suitably spaced and held down onto the gasket with beige polyester tape.  After years of use, the glue on the tape is failing and the gasket material loses integrity and rips as it is pealed back.  As one peals the gasket, there are exit points for the outer oven heater and sensor, and if not careful these connections can be broken.  The EFC and 10MHz connections are via very thin coax spaced among the enamels wires, and although more robust, I was still concerned about damage.  All the enamel wires terminate at the oscillator base via feedthroughs, not the typical PCB edge card one finds on the instrument style 10811 oscillator.  Probably an OEM style 10811.  If anything gets damaged, I don't see why the enameled wires could not be replaced with the Teflon variety, along with a silicon grade foam rubber strip.  The innermost insulation wraps the 10811 and looks like it should be a black foam rubber with adhesive back, but what I found looked more like black foam rubber with grease.  They slid all over the place and the "grease" was quite messy.  I don't know if the grease is normal or a breakdown of the old adhesive.  Other than using some new Kapton tape for reassembly, I didn't change anything else when reassembling.

During this fix, the 10MHz out was disconnected from the PCB and as I adjusted the output manually, I could see the EFC banging from one end to the other (between 0 and ~1,000,000 counts).  I also confirmed the oscillator's EFC range of +/-2Hz   I reconnected the 10MHz and made the manual adjustments.  Luckily, my frequency counter had an ovenized oscillator which I had calibrated against the Z3801A before it failed, so I know I had a reasonable reference.  I observed the EFC output and adjusted the frequency so I would get a middle range EFC number (~470K).  Per Jarl's comments I was worried about the EFC settling point after I reassembled the double oven.  Thinking about it, since the internal oven has a temperature control/setpoint, I thought adding additional heat via the outer oven should not change the temperature of the internal oven, but only change the duty cycle of the heater.  Once I reassembled the outer oven, I noticed no significant change in the EFC numbers.  Now I have a working Z3801A.  I watched the EFC for several days and noticed it leveled out quite nicely.  

My next experiment was to see what duty cycling the Z3801 would do to the overall EFC.  I wished to do this to see if I could have a locked 10MHz source without leaving the unit on day and night. This turned out to be a mistake.  The Z3801 worked for a few days, but then it went into a holdover mode.  I checked the output and it was gone,  it seems the oscillator output is no more.  After letting it cool down, the oscillator comes back, but dies within minutes.  Any ideas on this would be appreciated.

Off to my next problem: An HP E1938 hockey puck oscillator.  I will leave this to another email.

Bob