[time-nuts] Measuring Rubidium frequencies

[Magnus Danielson]

From: "Jim Robbins" <jsrobbins at earthlink.net>
Subject: [time-nuts] Measuring Rubidium frequencies
Date: Fri, 6 Jun 2008 09:01:48 -0400
Message-ID: <00b601c8c7d5$7b8a45c0$6401a8c0 at dell8200>

Jim,

> I have a problem which I am unable to understand and need some help.  I have two M100 Rubidiums, a Racal 1992 counter, a factory standard T'Bolt and a Symmetricom Starloc II.  When I set up either GPSDO as the External Timebase for the 1992 and measure either Rubidium, I get the same reading of 9.99998278.  When I use the 1992's Internal Timebase, I get 9.99999995.  All equipment has warmed up.  The antenna is a Symmetricom HP 58532A on the roof.     On a few occasions I have seen the frequency drift up to 0.00001722.  I would have expected that both Rubies would have read 0.00000000.  Any ideas would be much appreciated.

Do you have propper levels at the reference input?
If you have a too weak signal you may miss a cycle every now and then.

The same can happend if your measurement input has the trigger point in the
wrong spot. However, I suspect the reference input is much sparser in design
than the measurement input. This put more requirements on the signal and
"should not be a problem".

This would however make your counter show larger numbers.

The best thing you can do is to take a RF generator, dial it to 10 MHz and
dial the amplitude to different levels while feeding the reference input.
Watch for the same kind of anomalies.

Another thing which may kill you is interference. Additive interference can add
additional cycles. Now, for a frequency counter and its reference input
it produces the interesting feature of showing lower numbers than it should,
since your actual reference runs at a higher frequency than you think.

So, given that it is the external references which produces lower numbers, I'd
bet that you have additive interference on the 10 MHz signal. I'd suspect a
lower frequency than 10 MHz. Putting an RF-choke on the cable may help.
If you have a spectrum analyzer at hand, use that, but for the signal to have
impact it needs to be fairly strong, so scoping it may do the trick. However,
you must scope it with the counter connected to recreate the same environment.

It may very well be a switching supply that upsets you. One trick would be to
isolate the parts so that they are not part of a larger interfering
environment. Another trick is to just turn off everything not needed for the
measurement. Regardless, it seems like additive interference.

One counter-action you can take is to convert your 10 MHz to a squarewave.
The slewrate of the square leaves less time for the interference to interact
with the signal. RF chokes in addition may help.

Cheers,
Magnus