Trimble's Acutime 2000 compared to Arcron MSF

[Drk Ryan]

Are there ratings available for different reference clocks performance
with NTP?
I am looking at the offset for two stratum-1 servers, one uses
Trimble's Acutime 2000 and the other an Arcron MSF Receiver.
The performance is much better with the Trimble receiver. Has anyone
else experienced this? Or is there a problem with the MSF receiver?

[Jonathan Buzzard]

Post by Drk Ryan
Are there ratings available for different reference clocks performance
with NTP?
I am looking at the offset for two stratum-1 servers, one uses
Trimble's Acutime 2000 and the other an Arcron MSF Receiver.
The performance is much better with the Trimble receiver. Has anyone
else experienced this? Or is there a problem with the MSF receiver?

Yep the Arcron MSF receiver is a not very good. You can achieve
*much* better MSF results with some cheap homebrew hardware and
a thousand lines of C.

JAB.

[Piotr Trojanek]

Post by Drk Ryan
Are there ratings available for different reference clocks performance
with NTP?

when you use GPS and want performace, then I assume, that only PPS
signal characteristics is meaningful. so at first compare technical
specs of various GPS receivers. in example there are the Motorola OnCore
products, one expecially designed for timing applications, with hardware
algorithm to select the bests of avaiable satelites. this has much
better PPS accuaracy than the cheaper one:)

the next criterium is OS and hardware which GPS is connected to.

[Tim Shoppa]

Post by Drk Ryan
Are there ratings available for different reference clocks performance
with NTP?
I am looking at the offset for two stratum-1 servers, one uses
Trimble's Acutime 2000 and the other an Arcron MSF Receiver.
The performance is much better with the Trimble receiver. Has anyone
else experienced this? Or is there a problem with the MSF receiver?

Others have already chimed in with their own observations, which may
be right for them, but I have to ask:

What kind of performance are you looking for?

Microsecond-level PPS? Trimble Time Tag (which advertises that it
eliminates the latencies of PPS?) A couple milliseconds via serial data?

Most any serial link will need the serial line delays fudged... comparing
before doing such fudging is unfair IMHO. And if you only care to a few
milliseconds (a fair assumption for WAN links) then does any of it matter
to the application?

Tim.

[Paul Croome]

Post by Drk Ryan
Are there ratings available for different reference clocks performance
with NTP?
I am looking at the offset for two stratum-1 servers, one uses
Trimble's Acutime 2000 and the other an Arcron MSF Receiver.
The performance is much better with the Trimble receiver. Has anyone
else experienced this? Or is there a problem with the MSF receiver?

Drk,

Yes, some refclocks have better performance with NTP, others have
worse performance...

What exactly do you mean by performance? Does the ability of the U.S.
government to switch off GPS enter the equation? Or the fact that the
MSF transmitter is regularly switched off for maintenance? Do you need
to calibrate to absolute time? Do you mean long-term accuracy or short-term
jitter?

Do you have a PPS interrupt signal available?

The precision of the PPS signal from the Acutime claims to be 50ns (1 sigma);
you might expect an LF timecode receiver to be precise to a few milliseconds.
The LF receivers that I'm using seem to quantise their PPS output to 1ms.
Yes of course GPS is much more precise then MSF, doesn't everybody know that?

Much of this stuff is already available in the doc....

Paul