Time syncronization of PM8000

[sesa71269_bridg]

Hello Guys,

May I ask what type (exact products type, reference number, ...) of time syncronization products can you offer to do syncronization of PM8000 meters? We want to realize a

± 1 ms syncronization and I need and exact product which we can use for this purpose.

Thanks, Laszlo

[Bojan]

Hi Laszlo,

PM8000 meter could be ± 1 ms synchronized by the following:

For example, SER-3200 would do the job.

Best Regards,

Bojan

[sesa71269_bridg]

Dear Bojan,

Thanks for giving fast respond. It is a good product if you have the meters in reasonable distance. Now we are looking for a solution that can be used in a huge industrial site whihc have a territory of 4 km x 4 km with more that 20 substation with PM meters all over this territory. the proble mis that DCF (Frankfurt time) is not really stable source which we can receive...

So we need a solution that can be used in such a huge territory and can provide really syncronized time for all PM meters all over this 4 km x 4 km territory. And we need ± 1 ms syncronization.

I believe SER-3200 is not able to do ± 1 ms syncronization with 2 wire RS485 over 4 km x 4 km territory, em I right?

Laszlo

[Bojan]

Hi Laszlo,

I would go with Fiber Optic Transceivers instead of 2-wire RS485 in your case:

[sesa71269_bridg]

thanks, looks good solution

[sesa82206]

This is how I have done it:

Regards,

[Bojan]

Belden EIA-485 Cables Provide Superior Electrical Performance, but the maximum distance achieved would be 4,000 ft. at 100kbps, and Laszlo has to cover 4 km x 4 km territory.

[sesa46696_bridg]

I would think the best solution would be to simply add additional time synch units from Cyber Sciences. It would be easier than trying to run fiber over an industrial site. Remember the GPS time synch is not from meter to meter, it is from the meter to an atomic clock. The meters do not have to be communicating to each other for their time to be synched

[Hal_Etheridge]

My suggestion is to have one GPS per substation as well.  I suspect that the actual accuracy will be better than the specs as the devices are only tested to +-1msec worst case absolute error but it should much better a majority of the time.

The only possible issue with this is that the device to device time difference could be +- 2 msec in the worst case.  Each device is +- 1 msec to the GPS standard so that means if device A is at +1 msec error and device B is at -1msec error (compared with the absolute GPS time) the overall error device-to-device is 2 msec.  How realistic this is in practice I am not sure about though...

You can test the accuracy in practice by having the digital inputs of two or more devices connected to a single cable that is periodically pulsed.  The difference in the timestamps is the difference in the clocks of the two devices.  This is similar to a test we have used internally to verify the timesync accuracy of the devices (no matter the timesync method used).

One thing to remember...  For sequence of event analysis you do not actually care what the *absolute* timestamp is long as all the device use the *same* or common time base.  It is the *relative* timestamps between the devices that is important in this case.  Of course if you are doing wide area SoE analysis the most convenient time base is to use GPS or another similar source. 🙂

I actually got a patent granted on a method to timesync devices across a wide area network without the need to use GPS but rather a single device as the "master" time source (which could be GPS sync'ed but that was not a requirement).  Existing PME communication channels could be used with their non-zero and random comms latency.  Unfortunately this idea has not been implemented in our devices yet.

[Bojan]

How complicated is to install a GPS antena from each electrical room? I guess, it has to be mounted at the top of each building?

[Hal_Etheridge]

My experience is more from GPS running watches than the ones used in this application but long as it has a reasonably good view of the sky then it should sync.  I would not be surprised if mounting it on an outside vertical wall would work most of the time.  You should just need a single satellite in view to get the time, while for detecting movement you need a minimum of three or four (more in view just makes the fix more accurate up to a point).  There should be about 8-12 satellites present if you have a completely unobstructed view of the sky (depending where you are in the world).

Note that I have no real experience mounting an antenna for this application so anyone else who replies and has done this should have a much better idea of the realities of the situation so use this at your own risk.  🙂

[sesa71269_bridg]

This sounds nice but I believe a GPS time sync product with all the feature will be around 2-3 000 USD, right? So such a solution in this industrial application with around 14 substation is quite expensive ... No any other cheaper solution?

[Anonymous user]

Hi Laszlo (and thanks to Ivan and Steve, too, for comments about Cyber Sciences!)

Your project with need for time-distribution over long distances is a perfect match for our new CyTime SER models with PTP (IEEE 1588).  As Hal said above, SNTP is not good enough for the accuracy needed, but PTP can achieve sub-microsecond accuracy over Ethernet.

For your project, just provide GPS time signal (IRIG-B or DCF77) to one CyTime SER3200, then this device serves as PTP grandmaster for all other SERs--anywhere on the Ethernet network.  Any PTP slave device (SER3200) can then serve as "time-sync master" for PM8000, with its ASCII RS485 output to PM8000 COM port.

Here are links to the Cyber-Sciences.com site and some key documents:

NEW—DataSheet: Precision Timing with PTP (IEEE 1588)

NEW—TN-100: Hi-Res Time for Modern EPMS Using PTP (IEEE 1588)

NEW—Customer Presentation: PTP-Enabled SER

One other comment regarding point raised by Hal about +/- 1ms errors possibly adding.  In fact, to ensure the end goal of +/- 1ms timestamp accuracy, we expect time-sync TO the devices to be on an order of magnitude better--thus to get 1ms, we need 100us or better.  This is exactly what the SER3200 provides to the PM8000 from its ASCII RS485 output.  And PTP allows this same goal to be met over Ethernet.  Even without any special Ethernet switches--no changes to the existing Ethernet infrastructure.

Thx for the great post!

[Marvin_Foster]

I see that the PM8000 can take in a IRIG-B signal on an Digital Input or an ASCII signal on it's serial port.

If have a SER3200 is there a preference of either of these solutions?  I have 3 PM8000 near by that we want to time sync.  If we go with the IRIG-B signal, would you use the 5V or the 24VDC version of the PLX module?  I can't find anything in the PM8000 manual about what type of IRIG-B signal it will accept.  B006 for example.

[Anonymous user]

Hi Marvin,

This is another excellent question!  Cyber Sciences is a third-party supplier of SER and time-sync products, but most of us re ex-Schneider people and our factory is near the Schneider Electric PowerLogic metering plant in LaVergne, TN (near Nashville, TN USA).

So, I don't want to speak too much for Schneider Electric, but we do work closely with our many =S= friends, and we have tested both PM8000 and ION7650 with our "PTP-enabled" solution using SER-3200-PTP.

We find that the accuracy from our SER-3200-PTP RS-485 output (Arbiter ASCII protocol) to the PM8000 is similar to accuracy of our IRIG-B signal to PM8000.  From the SER-3200-PTP end, the RS-485 output is built-in (standard feature), and so this is the least cost.  To add IRIG-B output, we offer a small wiring accessory called PLX-5V which outputs the standard IRIG-B signal (5V DCLS, code B006) to the PM8000.  The only drawback to using ASCII/RS-485 to PM8000 is that the PM8000 has only one COM port.  So if you already need that COM port for other purpose (e.g., Modbus gateway), then you should choose IRIG-B.

In the case of ION7550/7650, the only option is ASCII/RS-485, so the choice is clear.  And this meter has more than one COM port, so there is usually one available for precision time sync.

In general, think about a time-sync architecture in which you set the time of the first SER-3200-PTP (PTP master) by GPS (either IRIG-B or DCF77), then all other SERs on the same Ethernet time-sync automatically via PTP (PTP slaves).  In turn, any SER-3200-PTP slave can serve as a "time sync master" and output ASCII, IRIG-B, DCF77 or 1per10 (used by Sepam).

Please see links in earlier posts to Cyber Sciences Tech Notes for more details, or please call (615-890-6709) or email me (Robert.kennedy@cyber-sciences.com). Also, here's a link to our Tech Library, which is an easy way to find more documents:

www.cyber-sciences.com/library

I hope this is helpful--best wishes for success!

[Anonymous user]

PS. Regarding your question of using PLX-5V or PLX-24V, I forgot to explain:

Normally, PLX-5V is used for IRIG-B. (up to 7 devices)

For long distances and/or greater number of devices (8-32), the PLX-24V can be used. This outputs IRIG-B at 24Vdc plus control power to one or more Cyber Sciences STR-IDM (IRIG-B Distribution Modules), each STR-IDM has 8 conventional (5V DCLS) IRIG-B outputs.

For your project, with just 3 PM8000, you sould use either ASCII/RS-485 or PLX-5V to distribute IRIG-B to PM8000.

[Yut]

I have just finished one project which the GPS Clock source to the last meter (ION8800) is 3 km.

I install the converter that convert TTL(DCLS, IRIG-B) to FO (single mode) the at the end I use FO to TTL(DCLS)  converter convert signal back to meters.

See Meinberg website for more.

[Anonymous user]

Hello Yut,

Running separate cables (copper or fibre) for IRIG-B used to be the only option. Today, there are other solutions that may also be interesting.  In the case of very long distances (e.g., 3 kilometres in this example), two other possibilities:

1) Install separate GPS antenna and receiver, as suggested in an earlier post. The price of gps receivers from Arbiter, Meinberg, Masterclock, Tekron, etc. has made this affordable. The antenna is mounted on a roof, with a good view of the sky.  A surge suppressor may be added if desired.

2) If the devices are already on the same Ethernet network, then the solution is much simpler:  sync a Cyber Sciences SER-3200-PTP unit at one end via IRIG-B, then install a second SER-3200-PTP at the other end, equipped with PLX-5V IRIG-B output to meters which support IRIG-B (e.g., PM8000).  The SER-3200-PTP also outputs ASCII/RS-485 for PM8000 or ION7550/7650.  The first SER serves as "PTP Master" (PTP= Precision Time Protocol, per IEEE 1588), and any other SER-3200-PTP unit on the same network will sync automatically--within 100 microseconds.  At least 200 SER-3200-PTP slaves are supported.

[Yut]

Hello Robert,

I did the project like that base on availability of GPS Receiver and existing SM FO network this is the cheapest way to distribute IRIG-B signal to all meters.

Sometime we have to move the antenna and receiver far away when all meters are in the switch yard due to interference from big transformer  then fed signal via FO.

All the receiver's brand you mentioned work well with ION meter (we use the Masterclock most).

[Anonymous user]

Hello Yut,

Thank you for your interesting feedback on projects.

The PTP solution (precision time-sync over Ethernet) that I described is mainly for projects where there is already an Ethernet network connecting all the meters.  In this case, the time signal can take advantage of the same LAN for time that is already used for data.

I will see the CEO of Masterclock next month so I will tell him about your good experience with his devices!