[lug] Time sync w/GPS or radio

[Nate Duehr]

Gary Hodges wrote:
> I'd like to set up a solution where I keep time set on a single linux 
> box with either GPS or the NIST radio signal.  I've found a company in 
> the UK that sells what I'm after, but I'm curious if anyone knows of a 
> U.S. based vendor?

Sounds like you got a lot of replies, but I've done both...

1. Set up NTP to be locally disciplined by a GPS hanging off of a serial 
port.
2. Buy a box that does this and use that.  I'd recommend one, but the 
company went under.

(They used to sell to the U.S. Navy, and had all of the crazy 
certifications that made the government folks happy.  Boxes worked great 
and did NTP, rdate, some other time protocols we didn't care about... 
and also could clock local telco gear to Stratum 1 quality after a 30 
day sync period.  Their on-board clock source was a Stratum II.

Pretty nice.  They must not have charged enough for them...)

The second option is a LOT more expensive, but... it does have the 
advantage in that they're often built to properly drive a remote LNA 
(low-noise amplifier -- basically a "powered antenna" in layman's terms) 
that can be mounted outside on the roof of the building, the receiver in 
the unit includes a voltage source that can power an LNA a couple 
hundred cable-feet away (typically).

Having an outside antenna (in either case) greatly increases the chances 
of seeing enough birds in the GPS cluster to keep the time set 
correctly, but both NTP and these "boxes" have a "drift" mode which has 
a decent clock that will be re-disciplined (usually slowly) back to GPS 
time after a loss of signal event.  In either case, you want to build 
some type of alert to let you know if the GPS is falling out of sync. 
Don't buy a box, stick it in a rack, and ignore it... (like it sounds 
like your customer has.  Ha!)

Indoor antennas, RF interference from a data center environment, and GPS 
signals from LEO satellites, doesn't quite cut it, most of the time.

(Had one GPS receiver in a data center that had windows... the window 
let "enough" signal in that the GPS would lose sync with the GPS cluster 
when there were no birds in view out the window or directly overhead.) 
You can't count on this, though -- many office windows are covered with 
"metalized" films for tinting these days, which isn't as RF-permeable as 
regular glass.

Nate