[time-nuts] Re: Spectracom 9383 w/OXCO

[Erik Oehler]

More information from Keith:

*The short/concise answer: *the NetClock’s oscillator frequency error
values being reported at the end of these routine/expected oscillator.log
entries are great*! *No need for any concerns.



*Now, the full/detailed answer:   *This was an often-inquired about topic.
The expected term of “*Freq Error*” being present in these routine
oscillator.log entries was often misconstrued as a problem with the
oscillator was occurring.   In reality, it’s only the values at the end of
this routine entry, not the wording of the entry itself, which indicates
whether or not there was a problem with the oscillator.



Though they are indirectly related, there is a very distinct difference
between “*Freq Error*” and “*Frequency Alarm*”. As described on page 15 of
the attached “NTP server Status and Log entries” document, unlike
“Frequency Alarm” (which are recorded in the Alarm log and are always a
Major alarm condition), “Freq Error” is not an abnormal event or an alarm
condition.  It is part of the normal reporting of the accuracy of an
oscillator.  It stands for “*Fractional Frequency Error*” which is a
mathematical comparison of the measured output frequency of the oscillator
versus the desired output frequency from the oscillator (in this case,
10MHz is the desired frequency). The result of this calculation is reported
in scientific notation, known as “*Freq Error*”.



All oscillators do drift (some more than others, depending on the type of
material inside the oscillator. Rubidium is an extremely stable material)
so the frequency is monitored/calculated on a regular basis in order to
discipline the oscillator.  Once the frequency error has been calculated,
the D/A value can be adjusted accordingly to correct the oscillator drift
that had occurred. The measured frequency is reported in the Oscillator log
in a couple of different types of measurements, including the fractional
“Freq Error” (in scientific notation) and a raw Frequency count (CNT) where
a count of 1,000,000,000 indicates the oscillator was measured to be as
close to 10 MHz as it can be measured.  This frequency error is reported as
FREQ ERROR=00E+00 (another indication that the oscillator is outputting the
desired frequency of 10MHz).  The closer the count is to 1,0000,0000,000,
the closer the oscillator was to the desired 10MHz, during that measurement
period.



In order to be able to accurately count the frequency of the oscillator, a
very accurate and stable 1PPS reference is required, in order to accurately
gate (start/stop) the oscillator’s output through the frequency counting
circuitry. The GPS receiver provides this stable 1PPS reference, while the
GPS receiver is tracking satellites.  For this reason, oscillator
disciplining only occurs when GPS satellites are currently being tracked.



As you will notice while reviewing the *Status and Logs* ->*Oscillator
Log* page
of the NetClock’s browser (as excerpted from the log entries you sent), the
reported frequency counts from this NTP server look very good (they are at
1,000,000,000 or very close to this count, indicating the oscillator is on
frequency and is very stable).  If these freq error values did actually
happen to be too large, there would also be corresponding Oscillator error
log entries being asserted in the *Alarms* *log*, as well.



I hope you find this info helpful~

On Thu, Jun 30, 2022 at 11:58 AM Erik Oehler <erik.oehler at gmail.com> wrote:

> Sam is correct! I work for Orolia (formerly Spectracom) and asked our
> support team about this issue. Keith from the team got back to me with the
> below explanation:
>
> I know “all too well” what this “commonly inquired about” NetClock 9383
> log entry means (☹)!   It means the NetClock is operating normally (😊)!!
> It was often misconstrued as an issue, but the “wording” of this expected
> entry is a “hard-coded” value which stands for “fractional frequency error”.
>
>
>
>
> All oscillators inherently have “error”.   This particular entry in the
> osc log is just reporting what amount of error there was, at that time.
> It’s the values reported at the end of this, and not the wording of the
> entry itself, that determines if there is an issue.
>
> On Tue, Jun 28, 2022 at 11:22 PM Sam Bach via time-nuts <
> time-nuts at lists.febo.com> wrote:
>
>> Yay, a question I can actually answer!  I also have one of these, though
>> mine has the TCXO option.  The FREQ ERROR is a measure of the output
>> frequency deviation from the reference, if I understand correctly.  It
>> doesn't indicate a fault with the unit, and is, like you figured, an
>> informational message indicating the status of the disciplining circuitry
>> for the oscillator.
>>
>> Sam
>>
>> On Tue, Jun 28, 2022 at 4:08 PM timenuts--- via time-nuts <
>> time-nuts at lists.febo.com> wrote:
>>
>> > Hi,
>> >
>> >
>> >
>> > Last year I acquired one of these from ebay – and it seems to be working
>> > fine.
>> >
>> >
>> >
>> > Having found this list while searching for something other – I thought
>> > this might be the place to ask 😉
>> >
>> >
>> >
>> > The oscillator log on the Spectracom shows this (and always have while
>> in
>> > my possesion):
>> >
>> >
>> >
>> > Jun 28 16:18:10 chronos spectracom: [system] FREQ ERROR= 0.00E+00 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 16:19:50 chronos spectracom: [system] FREQ ERROR= 0.00E+00 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 16:21:30 chronos spectracom: [system] FREQ ERROR= 0.00E+00 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 16:23:10 chronos spectracom: [system] FREQ ERROR= 3.33E-10 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 16:24:50 chronos spectracom: [system] FREQ ERROR= 0.00E+00 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 16:26:30 chronos spectracom: [system] FREQ ERROR= 3.33E-10 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 16:28:10 chronos spectracom: [system] FREQ ERROR= 0.00E+00 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 16:29:50 chronos spectracom: [system] FREQ ERROR= 0.00E+00 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 16:31:30 chronos spectracom: [system] FREQ ERROR= -3.33E-10 FREQ
>> > CNT= 999,999,999
>> > Jun 28 16:33:10 chronos spectracom: [system] FREQ ERROR= 0.00E+00 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 16:34:50 chronos spectracom: [system] FREQ ERROR= 3.33E-10 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 16:36:30 chronos spectracom: [system] FREQ ERROR= 0.00E+00 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 16:38:10 chronos spectracom: [system] FREQ ERROR= 0.00E+00 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 16:39:50 chronos spectracom: [system] FREQ ERROR= -3.33E-10 FREQ
>> > CNT= 999,999,999
>> > Jun 28 16:41:30 chronos spectracom: [system] FREQ ERROR= 3.33E-10 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 16:43:10 chronos spectracom: [system] FREQ ERROR= 0.00E+00 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 16:43:11 chronos spectracom: [system] AUTOMATIC D/A ADJUSTMENT.
>> > D/A= B08E MATCH CNT= 2755
>> > Jun 28 16:44:50 chronos spectracom: [system] FREQ ERROR= 0.00E+00 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 16:46:30 chronos spectracom: [system] FREQ ERROR= 0.00E+00 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 16:48:10 chronos spectracom: [system] FREQ ERROR= 3.33E-10 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 16:49:50 chronos spectracom: [system] FREQ ERROR= -3.33E-10 FREQ
>> > CNT= 999,999,999
>> > Jun 28 16:51:30 chronos spectracom: [system] FREQ ERROR= 0.00E+00 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 16:53:10 chronos spectracom: [system] FREQ ERROR= 0.00E+00 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 16:54:50 chronos spectracom: [system] FREQ ERROR= 3.33E-10 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 16:56:30 chronos spectracom: [system] FREQ ERROR= -3.33E-10 FREQ
>> > CNT= 999,999,999
>> > Jun 28 16:58:10 chronos spectracom: [system] FREQ ERROR= 0.00E+00 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 16:59:50 chronos spectracom: [system] FREQ ERROR= -3.33E-10 FREQ
>> > CNT= 999,999,999
>> > Jun 28 17:01:30 chronos spectracom: [system] FREQ ERROR= 3.33E-10 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 17:03:10 chronos spectracom: [system] FREQ ERROR= -3.33E-10 FREQ
>> > CNT= 999,999,999
>> > Jun 28 17:04:50 chronos spectracom: [system] FREQ ERROR= 3.33E-10 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 17:06:30 chronos spectracom: [system] FREQ ERROR= -3.33E-10 FREQ
>> > CNT= 999,999,999
>> > Jun 28 17:08:10 chronos spectracom: [system] FREQ ERROR= 3.33E-10 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 17:09:50 chronos spectracom: [system] FREQ ERROR= -3.33E-10 FREQ
>> > CNT= 999,999,999
>> > Jun 28 17:11:30 chronos spectracom: [system] FREQ ERROR= 0.00E+00 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 17:13:10 chronos spectracom: [system] FREQ ERROR= 0.00E+00 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 17:14:50 chronos spectracom: [system] FREQ ERROR= 0.00E+00 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 17:16:30 chronos spectracom: [system] FREQ ERROR= 0.00E+00 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 17:18:10 chronos spectracom: [system] FREQ ERROR= -3.33E-10 FREQ
>> > CNT= 999,999,999
>> > Jun 28 17:19:50 chronos spectracom: [system] FREQ ERROR= 3.33E-10 FREQ
>> > CNT= 1,000,000,000
>> > Jun 28 17:21:30 chronos spectracom: [system] FREQ ERROR= -3.33E-10 FREQ
>> > CNT= 999,999,999
>> >
>> > I don’t like the FREQ ERROR part – but seeing FREQ ERROR= 0.00E+00 FREQ
>> > CNT= 1,000,000,000 makes me believe it’s just rather informational?
>> >
>> > Or is this a case for concern?
>> >
>> >
>> >
>> > Kim Sindalsen
>> >
>> > _______________________________________________
>> > time-nuts mailing list -- time-nuts at lists.febo.com
>> > To unsubscribe send an email to time-nuts-leave at lists.febo.com
>> _______________________________________________
>> time-nuts mailing list -- time-nuts at lists.febo.com
>> To unsubscribe send an email to time-nuts-leave at lists.febo.com
>
>