The timing of NTP synchronization of the ION meters.

[Tomohiro-Okada]

Hi 

I have two questions about time synchronization of ION9000.

I did time synchronization settings of the ION9000 meter(sync with NTP server via Ethernet.)

and the time synchronization worked well.

The meter settings are below.

Metering Setup - Nameplate Info
FAC1 DeviceType 9000
FAC1 Revision 003.001.000
FAC1 Feature Set Standard

FAC1 Template 9000_V3.1.0.0.0

Advanced Ethernet - NTP
CL1 Enable NTP Time Sync NTP
CL1 NTP Time Sync Interval 1 day
CL1 Event Logging Threshold 0.02
ETH1 NTP Server 192.168.000.010
CL1 Time Sync Type UTC
CL1 Time Sync Source ETHERNET

but  only the cycle of the time sync looked random.

(In spite of the "NTP Time Sync Interval" setting was "1day",

 the Event history of the synchronization  with NTP server was  roughly about every hour.

About 20 times time synchronization was done in a day.)

Why was this happened?

The second question is the timing of the time synchronization.

For example, If I set the  "NTP Time Sync Interval"  setting to "1day" and the first time sync was 2022/06/30 09:00.

The next time sync will be 2022/07/01 09:00 I think.

But, the event history of the time sync was not so.

What factors decide the timing?

and can I define the timing of the synchronization?

Regards,

Okada

[DanL]

The unfortunately named "CL1 NTP Time Sync Interval" can only configure SNTP time sync intervals (not NTP).

Be sure to configure the "CL1 Clock Source" to "Internal" ("COMM" is also acceptable as with NTP enabled it will behave as "Internal"). Never use an external time sync source with the Clock Source configured to "Line Freq".

If you continue to have issues, please contact support for additional troubleshooting steps

[Tomohiro-Okada]

Hi @DanL 

Thank you for your reply.

So if the protocol is "NTP"(not "SNTP"), is the timing of the time sync random?

And if I change  the  "CL1 Clock Source" setting to "Internal", how long does the quartz crystal unit of the ION9000 or PM8000

can work correctly?

I'd like to know the product lifetime of the quartz crystal unit.

Regards,

Okada

[Tomohiro-Okada]

Hi @DanL,

I have one more question.

If the protocol is "NTP"(not "SNTP"), does the gap of the time between PQM and NTP server affects time sync interval?

What factors decide the execution timing of the time sync when the protocol is "NTP"(not SNTP)?

Thanks,

Okada

[Charles_Murison · Accepted solution]

Hello @Tomohiro-Okada ,

If the meter is set to SNTP, each time packet received by the meter should contain specific information, What time the meter sent the SNTP request, what time the NTP server got the request and what time NTP server sent the response. The meter would calculate the delta or transit time based on these three pieces of information as as well the local time on the meter when the packet was received by the meter. Based on this delta calculation and the SNTP/NTP servers' timestamps, the clock on the meter is adjusted.

NTP is significantly more complex. NTP uses complex statistical algorithms, aimed at filtering small discrepancies. NTP is intended to adjust the 'tick' rate of the device 'what is the length of a second' rather than just adjusting the clock time every x seconds. There are multiple variables that go into the calculations to determine how much to adjust a second. From time to time there may need to be a time step event to help the meter's clock adjust to match the NTP server's time. If this jump is greater than the threshold you will see the event in meter's event log, "time about to change" and "time change". Looking at the time stamp of these events you can calculate the approximate adjustment of the meter's clock.

"how long does the quartz crystal unit of the ION9000 or PM8000 can work correctly?" I am not 100% sure I understand the question. If you are asking what is the expected lifetime of the crystal, this will match or exceed the expected life time of the meter. Or if you are asking if okay to keep using the internal crystal, the answer is yes, this is okay.

Regards,

Charles

[Tomohiro-Okada]

Hi @Charles_Murison,

Thank you for your kind reply.

My question about crystal was asking if okay to keep using the internal crystal and I understand what you said.

I changed the setting from "Line Freq" to "Internal" and it works well until now.

Regards,

Okada

[Tomohiro-Okada]

Hi @Charles_Murison  and @DanL,

Sorry to bother you again,

 but I was asked from our customer to inquire  some questions.

I corrected the settings of PM8000 and ION9000 from "Line Freq" to "Internal" and the other settings are below.

Advanced Ethernet - NTP
CL1 Enable NTP Time Sync NTP
CL1 NTP Time Sync Interval 1 day
CL1 Event Logging Threshold 0.02
ETH1 NTP Server 192.168.000.010

Clock - Timezone
CL1 Time Sync Type UTC
CL1 Time Sync Source ETHERNET
CL1 Clock Source Internal  (changed from "Line Freq")

In this case, the time synchronization works well but only the timing of the NTP synchronization is still looked random.
(time sync was done randomly per one hour or 2 hours.)

Does it work correctly?
(I understand NTP is more complex than SNTP and the interval is decided by many factors from your answer,
 but we needs your official answer that it is working correct.)

If the setting of "CL1 Enable NTP Time Sync" is "NTP" (not "SNTP"), is the setting of "CL1 NTP Time Sync Interval 1day" completely not used ? or has some meanings?(for example, roughly sync with NTP server per "1day")

Which do you recommend, "NTP" or "SNTP"?

Thanks,

Okada

[Charles_Murison]

Hello @Tomohiro-Okada 

If the meter is set to use NTP, the interval is not configurable in the meter settings so 1 or 2 requests per hour is not unreasonable. One request per hour or 2 hours even is not a bad sign. The setting "CL1 NTP Time Sync Interval" is not used by the meter when setting is NTP. 

Regards,

Charles