On Tue, Feb 28, 2012 at 11:24 AM, Pete Lancashire
pete@petelancashire.com wrote:

I think in practice you are going to need to mount the 5680 to an
aluminum plate at least twice the size of the 5680 and then keep a
moderate airflow over the top and bottom.  Perhaps mount the  plate
using standoffs.  It's clear to me from seeing all those screws that
the 5680 needs to be firmly connected to some kind of heat sink even
it only a large plate.

Yes you pretty much have to calibrate the 5680.  But what do you mean
"if that changes"?  the 5680 will drift away from any calibration.

Chris Albertson
Redondo Beach, California

With the FE-5680A mounted to a 4x6x1" finned aluminum heatsink as shown here:
https://picasaweb.google.com/109928236040342205185/FE5680A#5685636673930493138

the opposite side of the case reaches 55 C in a 20 C ambient, with no fan whatsoever. Running a 40mm 12V fan at 5V aimed at the heatsink drops that to 45 C. At 5V the fan is nearly silent, and the airflow is quite weak.  Despite the internal temperature compensation via DDS offset updates that the unit apparently does, I measured one of my units to have a 7E-12 per degree C temperature sensitivity, larger than I expected. So I do want to use a thermostat on the fan.

The Symmetricom XPRO manual http://www.symmetricom.com/link.cfm?lid=7082  (which I believe is a somewhat similar Rb design) estimates that the lifetime of that unit doubles when baseplate temperature is reduced from 60 C to 40 C (MTBF: 243k hours => 591k hours).  Note that 243k hours is about 28 years.

On the other hand, my assumption is these surplus FE-5680A units were cheap because they were considered end-of-life already.  Maybe for something somewhat fixable, like the 60 MHz VCXO center frequency having drifted out of a spec window.

bad english on the 'if that changes'. Eventually with both in the same
box have the
Rb tweeked by the GPS. Just some other project for fun.

On Tue, Feb 28, 2012 at 11:42 AM, Chris Albertson
albertson.chris@gmail.com wrote:

Hi Pete;
I saw a paper from on of the "Blood Suckers" that said the cooler the better, and the effect of temp on life was huge. I looked on my computer and couldn't find the paper again. I now think that data may have been in the hard copy of a manual for one of the telecom shelves I have worked with. If I find it, I will pass the numbers on. But needless to say, for long life keep it as cool as reasonably possible.
Best Wishes;
Thomas Knox

On Tue, 28 Feb 2012 11:42:51 -0800
Chris Albertson albertson.chris@gmail.com wrote:

It could also be that those screws are used to hold the top
and bottom part of the cover firmly together. Dont forget that
Rb's are sensitive to magnetic fields and that the case is mostlikely
made of some high permeability material.

		Attila Kinali

--
The trouble with you, Shev, is you don't say anything until you've saved
up a whole truckload of damned heavy brick arguments and then you dump
them all out and never look at the bleeding body mangled beneath the heap
-- Tirin, The Dispossessed, U. Le Guin

I am fitting one of these into a HP 5335A counter.  The project became a bit
more involved than it first seemed and part of that is how to deal with the
cooling.  The manual shows an AC fan but it had been replaced (sloppily, I might
add) with a DC fan which was terribly noisy.  I found a much smoother one with
even more airflow in my junkbox.  The 5680A is mounted near where the fan is so
it gets plenty of airflow, but what about when the unit is in standby?  What I
did was put a little relay in which is powered by one of the DC supply outputs.
When the unit is off, the relay opens which places a resistor in series with the
fan (which is now powered by the small laptop 15 volt supply I installed).
Thus, when in standby the fan drops to a lower speed and acceptably quiet
operation and the 5680A doesn't get too hot.  I determined the resistor value
experimentally via my super duper accurate
run-unit-for-a-while-remove-cover-feel-5680A-case-change-resistor-repeat
method.  Ok so I'm not looking for supreme accuracy of temperature, just good
operation.

Nifty little project to get great accuracy for half the price of a good used HP
oven oscillator.

Peter

On 2/29/2012 6:03 AM, Attila Kinali wrote:

That is one of the better uses for a 5680.  I have a counter that is
non-function because some one salvd the HP OCXO from it and I'm
thinking that a 5680 would fit inside.  But then I thought again and
the 5680 can fit outside.

One problem I think with your setup is that the 5680' temperature will
change abruptly when you take the counter off standby.  You might
have to wait 15 or 20 minutes for it to be stable.    It think it is
easy to just run the fan at the exact same minimum rate all the time
I put a temperature sensor in the 5680 and use a comparitor to turn
the fan off and on.  The 5680 needs 15 volts so I dropped some volts
with an LM317 regulator  I adjust the regulator voltage so the fan
does not make to much noise and the lm311 swiches the fan on/off as
required.  So there is never an abrupt change.  and the fan run no
more than it absolutely needs to.  I had to add a positive feedback
resistor to reduce the cycle rate.

On Wed, Feb 29, 2012 at 4:15 AM, Peter Gottlieb nerd@verizon.net wrote:

--

Chris Albertson
Redondo Beach, California

Hi

If it is a normal FE it is pretty close at 15 minutes and quite good after an hour or so. It certainly will beat the original OCXO at at hour. You might just run it when the counter is on and forget about running it in the standby mode.

Bob

On Feb 29, 2012, at 1:15 PM, Chris Albertson albertson.chris@gmail.com wrote: