[EAS] [EXT] NPT via NPR

[Larry Wood]

Thanks Harold that is quite interesting.

I still wonder why, if Bill's EAS unit was monitoring both KCBS and KQED, it got the NPT from KQED. KQED was at least 13 seconds behind KCBS. I am assuming he is monitoring KCBS and was getting a clean signal.

Cheers,

Larry Wood,
KQED-FM

From: EAS <eas-bounces at radiolists.net> On Behalf Of Harold Price

EXTERNAL
Larry Wood asked how the SAGE unit polls the various inputs. Short answer is at about the same time, down to milliseconds, but that doesn't tell the whole tale.

The ENDEC looks at decoded data from all six inputs at the same time. Once a valid ZCZC frame is seen, it is compared to other frames from the same input in the last few seconds. Once two match, the time is logged. This is the time stamp you see in the date column in the log. If the third header is needed at one location and not at another, then the ENDEC log will show different times, one a header's length off from the other. ENDECs in the same rack will typically log at the same time. The first alert with two matching headers is marked as received, subsequent alerts with the same ZCZC (not including the LLLLLLLL id field) are marked as duplicates.

In Larry's data, we do see NPR 1 received at the same time on the two ENDECs.

Note, of course, it is possible to see two different seconds in the log for simultaneous events. The NTP (network time protocol) keeps the ENDEC synced to a few milliseconds, but 23.999 and 24.007 will show up in the log as 23 and 24, even though they are only 8 ms apart.

Note that, if you feed a single source into two inputs on the same ENDEC, you can't predict which one will be "first". One or the other ends up in the linked list first.

Once an alert is received, we wait for the end of the detected two-tone attention signal before the alert becomes sendable. You will typically see the length of the third header,with a one second pause before and after, then 8 seconds for the attention tone, and a second or two for alert setup, closing relays, etc., before you see the send log entry. That setup time will vary based on how busy the ENDEC is doing other things, such as polling cap, user input, web access, etc. In Larry's data, there were 13 seconds between logged receive and send on one ENDEC and 15 on the other, which is as expected.

An excerpt of Larry's data:
SAGE Log for KQED-FM

11:20:16 rx (KCBS 1)
11:20:18 rx Duplicate (NPR 1)
11:20:29 tx (KQED FM)
11:20:37 rx Duplicate (KSJO)

SAGE Log for KQEI FM

11:20:18 rx (NPR 1)
11:20:31 tx (KQEI FM)
11:20:47 rx Duplicate (KFBKAM)
11:21:06 rx Duplicate (KBEB)

A final note - the ENDEC uses NTP to keep the clock very close to the time source. Two ENDECs might be off by a few milliseconds,
If they are synced and in the same rack, the seconds update on each should look simultaneous to the eye. If your local displayed time is off by an integer number of hours you have your UTC offset set incorrectly. If you are off by more than a few seconds, you probably aren't synced. I see data in this thread from Bill Ruck with times of 11:35. Is this a typo, or is that EAS device not synced to a time standard? The ENDEC's tools page will show if NTP has ever synced.

Hope this was useful. I try to keep manufacturer-specific lengthy posts like this off the lists, but I had more than one request for this.

Harold

On Fri, Aug 13, 2021 at 12:24 PM Larry Wood <LWood at kqed.org<mailto:LWood at kqed.org>> wrote:
>In the interest of complete accuracy...

>KQED-FM received the NPT from KCBS. I'm not sure how the SAGE unit poles the various inputs, ...

__________________________________________________________
The EAS Forum Discussion List is hosted by the BWWG (Broadcast Warning Working Group). http://eas.radiolists.net<http://eas.radiolists.net>
Please invite your friends to join our Forum! The sign up is at: https://lists.radiolists.net/mailman/listinfo/eas<https://lists.radiolists.net/mailman/listinfo/eas>
___________________________________________________________