High resolution timestamps for packets details Topic is solved

[johncom]

Can I ask for an exact explanation of the process of assigning timestamps to captured packets using the Ethernet debugger (where the timestamp is performed, what timestamps are recorded by the wirehark, what affects the stability of the timestamp, ...)? We would like to capture the transit times of specific UDP packets (UDP packet stream 150 Mbps) simultaneously at two locations in the LAN. Capture would be performed by one computer with two connected Ethernet Debuggers. We are interested in the travel time of each UDP LAN packet with an accuracy of tens of nanoseconds - the time difference between two simultaneous measurements (associated by the packet ID) is needed, no absolute time is required. Statistical evaluation (min, max, mean time) of captured data (* .pcapng file) will be performed by our own script. Is Ethernet Debuger a suitable device for this application? Thank you in advance for your answer.

[Daniel (Intona)]

The timestamps are taken from the internal 64 bit counter that is incremented by 10 at 100 MHz, so the resolution 1 ns with 10 ns quantization. That 100 MHz is generated by the FPGA's PLL, clocked by the on-board 25 MHz oscillator (stability: 30ppm).

The timestamp is caught exactly at the time when the RX_DV signal of the Ethernet Phy rises. This is the earliest possible moment for the hardware to detect an incoming frame.

The host software saves the system time when the first data is received within a session:

[External Link Removed for Guests]

and adds this to the timestamps before passing to Wireshark:

[External Link Removed for Guests]

So you have a roughly natural absolute time with accurate frame-to-frame timing.
For me, it looks like the Ethernet Debugger is a perfect match for your application.

[johncom]

Thank you for your prompt and comprehensive response.