[time-nuts] Faster/more automated Rb oscillator calibration

[Gregory Muir]

Interesting you should mention this.  At the moment I have a calibration running in the lab that utilizes a Dataq Instruments (http://www.dataq.com/) DI-710 data logger unit that is being fed from the analog output of a HP 3575A gain/phase meter.  The DI-710 then couples to a PC via the USB connection.  Running the Dataq supplied WinDaq software gives me a simple chart recorder presentation on the PC where one can change the time base on the trace to extend from seconds to weeks in a single span.  In turn this gives me the sawtooth pattern as the 3575A rolls over in phase, direction indicating the high/low relationship of the frequency.

I start with a rough cut calibration using my 5335A then connect the unit under test to the DI-710.  From there I can watch the behavior over days and tweak as necessary.  The nice thing about this Dataq unit is that the 16 single-ended analog inputs (8 differential) can be put to other tasks as necessary.  I frequently connect the unused inputs to monitor DUT power supply voltages, lamp voltage, oven control, room temperature, AC line voltage, etc. that gives me a very complete picture of the effects of outside influences on the unit being calibrated.

In addition, the DI-710 also sports 8 bi-directional digital I/O lines that can be used for other tasks (time mark, external test hardware device control, etc.).  Options include the ability to stand-alone and write the data to a SD card, Ethernet capability, etc.  For a $600 investment, I have found it to be a real handy machine.

When sampling a single analog channel at a 1 Hz rate and using only a 512 meg SD card, you can continuously record to the card for 3000+ days before filling it up.  I often take it into the field to monitor unattended equipment in standalone mode (no need for a PC) in order to diagnose phantom anomalies that infrequently occur.  It is nice to let the unit run for a month or so then sit in the comfort of the lab to see what was happening.

Sometimes simple is nice.

Greg



On Fri, 31 May 2013 12:24:12 +0000, Brian Alsop wrote:

>This is probably nothing new, but looking over the archives, I didn't 
>see it posted in the last couple years
>
>I've had an LPRO-101 Rb oscillator for several years.
>It was first calibrated using the phase shift vs time method using a 
>GPSDO as the reference.  Stopwatch and oscilloscope.  Ugh!
>
>Needless to say it got really time consuming with full cycle times of 
>1000's of seconds.
>
>I've revisited this and come up with something much simpler but more 
>importantly automated.
>
>Simply input the GPSDO oscillator signal into channel A and the Rb 
>oscillator into channel B of my HP 5335A counter.  Tell it to determine 
>the time difference (1ns resolution).  Send these differences to a file 
>via HPIB.