Time Base Cals on 90s vintage HP Freq Counters

[Hawaii596]

I've noticed this on a number of HP frequency counters (I believe 5384A, 5385A, and today, definitely 5334B, and probably others).

But in the performance checks, they don't test the time base.  There is an adjustment procedure for standard and option 010, but no performance check.  This has bugged me for years, and I normally just add it on anyway.

Anyone have any insight as to why HP (Agilent) did not check these in their procedures back then?  I was just auditing the accuracy of tolerance limits listed in my data sheet, which reminded me of this ongoing issue.  So thought I'd bring it up here - maybe one of you has read some tech note or bulletin in years gone by as to what is happening here.

[jimmyc]

  This has bugged me for years, and I normally just add it on anyway.

NOTE
Adjustment of the Time Base Oscillator is normal due to the Aging Rate of the
crystal . This is common to all Quartz Oscillators . The adjustment actions taken
during this calibration will ensure the greatest reliability of the TI by adjusting
the time base reference frequency to the nominal value each time it is calibrated.

cal intervals of counters are based on drift accuracies of the oscillator, and since the procedure is to optimize the time base at what would be considered "the right time" (pun for all the time nerds) it would only prove the age rate of the oscillator, which is already proven.  how do you add a spec that is not specced though?  wouldn't that be over specced and not therefore "legal"

from 33k3-4-451-1

Connect TI TIME BASE IN/OUT to the Electronic Counter CHANNEL A input.
4 .1 .7 Set Electronic Counter 50Q/1MS2 switch to 50f and GATE TIME as necessary for the resolution of the
following steps.
4 .1 .8 (STD) The Electronic Counter must indicate 9 999 982 to 10 000 018 Hz ±1 count of LSD.
4 .1 .9 (OPT 010) The Electronic Counter must indicate 9 999 999 .1 to 10 000 000 .9 Hz ±l count of LSD

[Hawaii596]

I have no problem with GIDEP (T.O. / Navair / etc.) procedures.  They've added in what the OEM for who knows what reason, left out.  So in my opinion, the various GIDEP procedures are fully adequate in how they do this test.

Time Base specs are typically in drift per day or month.  Therefore you would multiply as appropriate.  For example option 010 is 5 x 10 -10 / day after 24 hour warm up.  At a 12 month interval it would be 12 months worth of that drift.

To quote the specs from the HP 5334B counter, in the GENERAL specs, TIMEBASE section the spec is <3 x 10-7 per month, and option 010 is spec'd at <5 x 10 -10 per day after 24 hour warmup.  They are definitely spec'd.  Matter of fact I could easily write a procedure to properly check them including calculation of acceptable drift since last cal.

A point of trivia about quartz oscillators, every one has a different drift rate.  Every quartz oscillator when de-energized, allowed to go "cold" then re-energized and allowed to stabilize will have a different aging rate than it did when you turned it off.  Over time, the aging rate of a given quartz oscillator will change (not just it's value, but it's rate of change). 

In my compliance environment, my strict guidelines are to follow the full OEM calibration procedure (in HP (Agilent) lingo, "PERFORMANCE VERIFICATION.").  So I must violate my requirements to properly verify operation of these old HP counters. 

I just pulled the manual on a newer counter (HP 53131A) and found the same thing. 

Then for the heck of it, I just randomly selected a NON-frequency counter instrument with a high accuracy timebase (HP 8663A generator).  It has the same frequency accuracy specs as option 010 on the HP 5334B counter.  In it they spec'd a full aging rate check including calculation of drift rate of the time base. 

It just doesn't make sense to me that you calibrate a time base on a generator but not a frequency counter.  Again, GIDEP procedures (I just looked at 33K3-4-451-1 - the T.O. for the HP 5334A/B) and it checks the timebase.  It is a reasonable method.

Okay, enough of my rant for today.  I need to get back to work.  Thanks everyone for listening.

[mdbuike]

Ah now, if I mismember correctly, at least for USAF procedures, when checking the 8663A timebase, the counter is referenced to the lab time standard, and when frequency accuracy is accomplished, the counter and generator's timebases are tied together.

Standard practice where I am is that all the counter's timebases will be tied into the lab cesium when used for normal frequency measurements (i.e. an 8640B or 3325A, etc.) unless specifically directed by the procedure to tie the generator to the counter.

Mike