I'd like to know the expected time accuracy of the T7-Pro RTC using NTP. With normal NTP, I would expect <10ms time error. Does the T7-Pro achieve this, or is the implementation a "simple" NTP that only receives the time and is susceptible to network latencies (and therefore errors >1s in laggy networks)?
The T7 uses SNTP.
Thanks for your response - I'll assume this means that the time accuracy is limited by the network. Are there any plans in the near future to support NTP or PTP on any of your models to achieve better time synchronization?
No plans at this time. The T7's firmware uses a 3rd party network stack which does not have any plans to update the time protocol.
Sounds like you are talking about network connected T7s. If you are trying to figure out how to synchronize data from multiple T7s, we do have a suggested technique to do this if 1 common host computer can talk to all the T7s. Basically, at some interval the host reads a clock value from the T7 and notes the system time on the host. You should be able to do this in about 1 ms to get ~1ms synchronization. Then as I recall the math works out where based on the possible clock error of a T7 you need to do this about once per minute to maintain 1 ms synchronization.
Thanks for the tip. That synchronization sounds very achievable when wired directly to a single computer.
My application is a multi-switch/hub LAN with T7s potentially connected around the LAN with other instrumentation. A time server provides time and NTP services to connected devices. All instruments should be time synchronized to < 10ms. The LAN can experience large latency fluctuations, so we have generally avoided SNTP devices due to the lesser algorithm abilities of those devices to estimate and correct for network latency. An NTP (or PTP) enabled LabJack would be great for us, though I understand there are challenges in implementing it.
Externally clocked stream may help: https://labjack.com/support/datasheets/t7/communication/stream-mode
In that mode the T7 will sample a specified list of inputs whenever an edge is detected on CIO3. We still need to get the synchronization signal to each unit, maybe the unused lines in a cat5 cable can be used.