Emerson Exchange 365

Remote Stratum-1 NTP Time Server Accuracy and SOE cards

I'm having a lot of confusion surrounding DeltaV Time Sync with relation to SOE cards. I have the current architecture:

1.) DV network with 2 controllers and ProPLUS each with 1 SOE card attached to them.

2.) GPS based NTP time server ON A REMOTE NETWORK (i.e not connected to the ACN)

In this case, the PPLUS syncs with the NTP time server. My question is related to the "DeltaV Network Time Synchronization" white paper. It states the following:

The time synchronization accuracy for workstations, using 100Mbps ethernet, can be expected to be ±10ms in this configuration.
For DeltaV embedded nodes, the accuracy will be ±1ms (older DeltaV controllers running at 10Mbps are able to achieve ±10ms accuracy).
These accuracies can be achieved only when the Stratum 1 time server is directly connected to the DeltaV ACN.

Question #1: Related to the stated accuracy of +/-1ms when the time server is on the ACN, what is the accuracy when it is remote? Do you really gain anything by using a remote time server?

The "DeltaV Sequence of Events" white paper states the following:

Because SOE messages require more accurate synchronization, an NTP server that uses a standardized reference time - such as
Coordinated Universal Time (UTC) – is used. In this confguration, the reference time is typically communicated to the NTP server in
the DeltaV network using a radio or satellite receiver (eg. Global Positioning System—GPS) or modem. NTP servers that utilize this
method are referred to as stratum 1 NTP servers and provide the highest level of accuracy. These servers are capable of providing
time synchronization accuracy of +/- 1ms for controllers and of +/- 10ms for workstations. If SOE input cards are present in the
DeltaV network and the synchronization between controllers is less than +/- 1ms, the controllers will generate events indicating that
they are out of sync with one another.

Question #2: This last statement indicates that there will be errors is time sync accuracy is greater than +/-1 ms, but the Time Sync white paper states that you can only achieve down to +/-1ms. Can someone resolve these two (apparently) conflicting statements for me?

Alex Lutz

• 18 Jun 2019 12:29 PM
• Cancel

5 Replies

• Alex Lutz

  

  • 12 Jul 2019 12:00 PM

  anyone have any thoughts? The whitepaper describing the SOE accuracy with relation to the NTP time server is very ambiguous. Anyone else's interpretation would be much appreciated!

  • Cancel
• Alexandre Peixoto

  

  • 12 Jul 2019 1:12 PM

  In reply to Alex Lutz:

  Alex,
  
  in most cases the GPS Time Server is connected to L3/DMZ or other external networks to provide time reference to more than just DeltaV. A good example here is virtualization and security solutions like the SIEM for DeltaV systems. In this mode, the time reference will "cascade" through servers until it gets to the ACN meaning that the Stratum level will increase hence being less accurate if compared to when the time reference is directly connected to the ACN, but without SOE this lower accuracy is completely acceptable and simpler to set up in a layered network.
  
  If you have SOE cards, then we require that the GPS time server is connected to the ACN so that you get the most accurate time reference possible without any additional latency since you'll need that to really make use of the SOE functionality.
  
  The SOE card data sheet and the Time Server white paper are just confirming that you need a time reference directly connected to the ACN to enjoy the milisecond updates without synchronization issues. In your case I would not even be discussing the use of remote networks for time synchronization since you are using SOE.
  
  Regards,
  
  Alexandre Peixoto

  • Cancel
• Sal Salamone

  

  • 12 Jul 2019 2:17 PM

  In reply to Alexandre Peixoto:

  Alex,
  
  As Alexandre stated, a stratum of 1 means you are directly connected to the UTC source. Every jump through another NTP server increases the stratum. The Proplus is by default at a stratum of 8, I believe. It is a high number to it does not interfere with a better source. The network time protocol within the system will always use the lowest stratum number as its source for time.
  Answer #1: The best source for time is a stratum of 1, so a local UTC is best. When you are connected to a remote NTP server the best you can get is stratum 2.
  Answer #2: The statement, "DeltaV network and the synchronization between controllers is less than +/- 1ms, the controllers will generate events indicating that they are out of sync with one another."
  I believe this it poorly worded. I believe it would be better stated as, "DeltaV network and the synchronization between controllers is less than +/- 1ms, the controllers will generate events indicating that they are out of sync with one another OTHERWISE."
  If you look in diagnostics you can see how far out of sync each node is with the ProPlus.
  
  Note, though the ProPlus is the source node for DeltaV Network time, the ProPlus may be getting its time from another DeltaV node, like an application station where the Plant LAN is looking at an NTP server. I have a customer doing just that. There is a KBA that helps you set up a system this way.
  
  You may already know, that all SOE cards report to their controllers which take and organize the events relative to their localized time. localized meaning within the DeltaV Network. That is why it is so important for all the controllers to synchronized to less than +/- 1mS and hence why exceeding this should generate an indication that the controller is out of sync. I have seen this message before.
  
  Regards,
  Sal

  Sal Salamone

  • Cancel
• Alex Lutz

  

  • 12 Jul 2019 5:12 PM

  In reply to Sal Salamone:

  Sal/Alexandre - excellent responses and very insightful! i appreciate you shedding some additional light on this subject.
  
  After reading these writeups, are the following statements correct?:
  
  1.) if you have SOE cards attached to separate controllers and you want their synchronization to be < +/- 1mS to ensure accurate time stamping of events that are brought in by separate cards, then you must have at least a Stratum-2 GPS NTP server that is DIRECTLY CONNECTED TO THE ACN.
  
  2.) Connecting the ProPLUS to a GPS NTP server on the PLANT LAN does not directly improve the time synchronization on the ACN. It may make the ProPLUS time more accurate compared to some "truth time", but the ACN time syncing still behaves exactly the same (i.e. everything syncs up with the ProPLUS at the same rate, and no improvement in time sync delay).

  • Cancel
• Sal Salamone

  

  • 12 Jul 2019 6:19 PM

  In reply to Alex Lutz:

  Hey Alex,
  That's right ... I was wondering if you were thinking a NTP other than the ProPlus would improve accuracy of your SOE results.
  My understanding, and Alexandre correct me if I'm wrong, all the events will be placed into Event Chronicles in chronological order relative to local DeltaV system time. However for that to be accurate relative to external DeltaV system time both systems should be a close to a statum of 1.
  An example. The one customer I talked about who gets NTP through their PlantLAN does this so their ASPEN data is chronologically correct with their DeltaV events and trends.
  An SOE, to my knowledge, is just a datalogger operating at 0.25mS update rates. It caches the events in order until the Controller can collect the data. Then the controller(s) send their material to the Event Chronicle Historian ( usually the ProPlus).
  Make sense?

  Sal Salamone

  • Cancel