Hi Jim,
On 2/6/2014 3:32 PM, Jamieson (Jim) Rowe wrote:
Hi again folks,
You may (or may not) recall that a month or so ago, I asked for any information
that might be available regarding how to fix a ‘used’ FE-5680A rubidium module
from China (via ebay) which was tested by the supplier in China as working OK,
but would not seem to lock up to rubidium here in Sydney. There wasn’t a great
deal of info available, it seems, so I kept on checking ideas myself – mostly
with no luck. The module would never lock, but kept cycling back and forth
between about 9.999770MHz and 10.000036MHz – ‘searching’ for a lock, but never
finding it.
Anyway, a couple of days ago I was reading more about the operation of rubidium
vapour oscillators, and noticed that the ‘filter cell’ is very sensitive to
magnetic fields – hence the mu-metal shielding case, and also for the
‘C-tuning’ coil. And I wondered if the main reason why the FE-5680A had
apparently worked in China, but wouldn’t lock up in Sydney (Australia) might be
caused by the fact that Quangzhou (China) is in the northern hemisphere while
I’m ‘down under’ in the southern hemisphere – where the earth’s field is
presumably somewhat different, in terms of both strength and direction.
So I decided to test this in a crude way, by inverting the FE-5680A and seeing
what happened. And – lo and behold – it locked up within 2.5 minutes, and
stayed locked until I turned off the power and let it go cold again. The next
morning I applied power again, and within 3 minutes it locked up again with no
problems. And it’s been locked up now for over 48 hours...
My first thought was to make a typical 'down-under' joke and suggest you
run the 5680A upside down, but you beat me to it! :)
So it seems that the different magnetic field here may have been the problem –
either that, or it may have received a ‘jolt’ in transit, which prevented in
from locking unless it was inverted.
But how do I tell which of these explanations is right, without ‘opening her
up’ again and looking for some kind of subtle physical fault?
Another idea: perhaps the mu-metal shield case had acquired a small dose of
magnetisation in transit (via a physical shock, or from a strong field metal
detector). I guess in this case that I would have to remove the two halves of
the case, and bake them in a furnace to demagnetise them again.
My very limited knowledge regarding mu-metal is that it is so
magnetically 'soft' that it can't be magnetized. If it was somehow
magnetized, inverting the unit wouldn't make any difference, would it?
Or should I just run the FE-5680A upside down permanently – the simple but
‘crude’ answer?
I don't believe for a second (pun intended) that the earth's magnetic
field has any effect on the locking of your 5680A. It's just not strong
enough. The same applies to the C-field which can only nudge the
frequency one way or another by a small amount.
Since flipping the unit DID make a difference, my money would be on a
trivial, boring mechanical issue inside the unit. Could be a bad solder
joint, broken wire, floating piece of debris, or something like that.
Worst case might be a broken glue joint somewhere in the physics
package. That could be ugly. I would definitely open it up and see if
anything falls out.
I’m not sure if this FE-5680A has the ‘C-tuning’ gizmo fitted, or wired up. Am
I right in thinking that another approach might be to try varying the tuning
via the RS-232C serial port? Does this work via the C-tuning coil anyway, or by
tweaking the DDS?
The RS-232 commands affect the DDS - assuming your unit does have one.
It will have no effect on the locking, only on the output frequency.
A few years ago I bought a dead Datum SLCR Rb standard. It's a cousin
to the LPRO. I thought I'd learn some things by trying to fix it. The
problem was intermittent. I tore it apart and found that one of the
legs of the crystal had never been soldered! Never overlook the obvious.
I hope a much more experienced time nut can provide a few answers, please.
Maybe the blind leading the blind is a closer description.
Ed
Jim Rowe
_______________________________________________
time-nuts mailing list -- [email protected]
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
