[time-nuts] Fine print on HP 5334B

[Rick Karlquist]

It's worth keeping in mind a couple of things:

1.  The firmware in the 5334B is identical to the 5334A except
for an extremely minor change to accommodate a different modulus
in the C channel.

2.  The entire program has to fit into 2KB (two thousand and
forty eight bytes) of ROM, and was written in assembly language and/or
machine language and is understandable only to its author (Ken T.)
It uses every last byte of memory.  The 2KB limit is due to
the legacy, discontinued microprocessor that was used (long story).

3.  The interpolator has somewhat questionable accuracy when you
get down to 12 digits.  This is hidden somewhat by the firmware,
but if you monitor the bits the interpolator puts out, you can
see what is going on.  To me, it is kind of amazing it works as
well as it does considering the design.

So please lower your expectations!

If you want a really good HP counter, move up to the 53181 which
is much more modern, even being 15 years old.

Rick Karlquist N6RK
R&D Project Manager of 5334B (circa 1986)




Hal Murray wrote:
> (Interesting timing given the SR620 discussion.)
>
> I'm (finally) getting my GPIB gizmo working.
>
> I have a HP 5334B running off its (good) internal osc looking at the
> output
> of a Z3801A.  Assuming the Z3801A is stable, this is a backwards way of
> looking at the internal osc.  (There is a shared in/out connector for the
> internal clock, so I can't see the internal osc if I feed it an ext
> reference.)
>
> I left it running for a night with the mostly default setup and 100 cycle
> averaging.  I think the default gate time is 0.3 seconds.  Answers ranged
> from:
>   F  +9.999999330E+06
>   F  +9.999999333E+06
>
> I've had this setup running for several weeks.  The bottom digit doesn't
> fit
> on the display, so I never noticed there was anything interesting going on
> down there.  The 100 cycle averaging has a button on the front panel.
> Without it, the display is one digit shorter.
>
> Yesterday, I got it running with a single 60 second gate.   Now it's
> wobbling
> between:
>    F  +9.999999339E+06
>    F  +9.999999340E+06
>
> I was a bit surprised there is such a difference.  Is this all obvious to
> anybody who has carefully read the manual?
>
> Of course, now that I think about it, averaging frequency measurements
> seems
> less helpful that simply using a longer measurement time.  But I'm still
> surprised the answers were so different.
>
> Another quirk I noticed.  They truncate rather than round for the display
> on
> the front panel.  9.999999339 shows up as 9.99999933 rather than
> 9.99999934
>
>
> Is there anything else like this I should watch out for?  Does anybody
> have a
> quick summary on how to get good results or traps to avoid?
>
> --
> These are my opinions, not necessarily my employer's.  I hate spam.
>
>
>
>
> _______________________________________________
> time-nuts mailing list
> time-nuts at febo.com
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>
>