[time-nuts] HP 5370B low frequency modulation
magnus at rubidium.dyndns.org
Sun Jul 22 20:36:13 EDT 2007
From: Dr Bruce Griffiths <bruce.griffiths at xtra.co.nz>
Subject: Re: [time-nuts] HP 5370B low frequency modulation
Date: Mon, 23 Jul 2007 11:14:25 +1200
Message-ID: <46A3E4D1.2000900 at xtra.co.nz>
> Hej Magnus
> Oscillation/cycling of oven temperature?
> Oven controller oscillation?
Might be, but then my "loadingcondition" has caused the oven control out of
stability. This would certainly supprice me. The oven control in the 10811 is
linear and for most practical purposes behave like a first-degree linear
diffrential system. OK, there might be a second degree aspect, but the poles
should be fairly stable and it would be more a pure "amplitude" problem.
Now as I was down putting the hood on (and removing some of the residues from
the not so very smart packaging, using isopropanol) the down-spikes had started
to become fewer and rare, and it had 490 uHz of Allan Deviation, i.e. 4,9E-11
for tau = 5 s. I beleive it will considerably lower as the spiking stops.
The full 10811 seems powered during standby mode. Some solutions will only have
the oven powered in standby. What does change on powerup of the rest of the
counter is that you turn on alot more heat, current, the fan (which has a fat
metall wall between the flow and the 10811, but that piece of metal wrapps the
10811 so the cool-of-effect may be there never the less.
After putting the lid on and letting it sit there the spikes is still evident
and last reading gives 5,8E-11 @ tau = 5s.
A comperative measure with another 10811 (which I just *happend* to have a
cable for lying around on the table) but which was unheated also gave time for
a little experiment, looking at the heat-up performance. Now, it started some
200 Hz hot, but went for a steady linear dive downwards which at about 400 s
curved upwards and had a gentle overshot (i.e. lower frequency) until it
stablized. However, subsequent measure showed a slow but steady rise and ADEV
measures was surely limited by the driftrate. It was however much lower, such
as 1.2E-11, and that includes the drift-rate tainting while the display
indicated a much smoother curve.
As I don't have a wiring aisle for my lab-bench (beleive me, I start to
consider it as a must have even if it would be a squeeze-in variant of 3-4
dm or so, I could not get the heated 10811 in my 5372A. I really need to hook
the backside panels up to a BNC patch any decade now. Anyway, otherwise I would
have a well heated alternative to measure on.
I wish I get a phonecall with good news from a fellow time-nut tomorrow. If so
I drive down to him. :-)
More information about the time-nuts