IRIG-B optical fiber delay measurement

Hardware Setup

1) Fiber Interface Board of TrueTime XL-DC Receiver in Radon Hut

The new home-made module (see photo or schematics) has an additional optical fiber receiver sensor (HFBR-2416T, also of the Hewlett Packard HFBR-0400 series) for the feedback signal from the SK center hut. In addition, it has drivers and LEMO jacks for both, the source IRIG-B signal and the feedback signal for easy oscilloscope access (see photo of cabling setup).

2) Fiber Interface VME Module for TrueTime VME-SG in SK Center Hut

The new home-made VME module (see photo or schematics) has, like the new XL-DC interface above, an additional optical fiber transmission LED (HFBR-1414T) to re-transmit the IRIG-B signal back to the radon hut via the spare fiber. Also in addition, it has a driver and a LEMO jack for easy oscilloscope access of the IRIG-B DC-shift signal (see photo of VME cabling).

Measurements with Digital Oscilloscope

IRIG-B output versus feedback delay at XL-DC receiver in Radon Hut

1. (photo 1) made at 10 msec/div horizontal resolution:
   • channel 1: 1PPS output of XL-DC receiver (TTL level)
   • channel 2: IRIG-B DC-shift signal of XL-DC receiver before converted to optical pulse sequence at optical fiber LED.
   As shown, the leading edge of the second 8-msec pulse of the IRIG-B "frame" ID (double 8-msec pulse sequence) coincides with the 1PPS leading edge.
2. (photo 2) same as above, but zommed in at 10 nanosec/div horizontal resolution at leading edge of 1PPS pulse and by using "infinite persistence" display option. As shown, there is just a minimal (neglectible) propagtion delay of approx. 6 nanoseconds between the 1PPS and second "frame" pulse of the IRIG-B and no visible jitter.
3. (photo 3) made at 10 msec/div horizontal resolution:
   • channel 1: IRIG-B DC-shift signal of XL-DC receiver before converted to optical pulse sequence at optical fiber LED.
   • channel 2: IRIG-B feedback input signal right after conversion from optical fiber receiver sensor
   This just shows that transmitted and feedback IRIG-B signals are identical, just slightly delayed (not noticeable at this horizontal time resolution).
4. (photo 4) same as above, but made at 5 µsec/div. Here, the delay between output IRIG-B and its feedback signal coming from the SK center hut is already highly noticable at approx. 20 µseconds +/- 20 nsec.
5. (photo 5) same as above, but zoomed in at 50 nanosec/div, centered around the channel 2 leading edge. The leading edge of channel 1 is moved far left ouside the screen area, but the oscilloscope's internal measurement option allows here for very precise resolution of approx. +/-2 nsec (including signal jitter within the TTL-to-fiber conversion electronics).
   As shown, the signal delay between the original pulses and its feedback over the whole fiber link (2-way) is 20.0162 µseconds (+/- 2 nsec). Thus, the one-way delay over the 2km fiber from Radon Hut to SK Center Hut (including the conversion electronics) is half of that number, or:

   10.008 µseconds +/- 2 nanoseconds

   Taking into account the electronics propagation delays in the signal chain [TTL/RS422 driver(s), TTL-to-optical, and optical-to-TTL] which account to approx. 80-100 nanoseconds, then the optical propagation delay in the optical fiber can be estimated to as 4.954 nsec/meter or 20.186 cm/nsec.

IRIG-B signal versus 1PPS signal of VME-SG module in SK Center Hut

1. (photo 1) same as the first photo above, just to show how the VME-SG's 1PPS output and the IRIG-B pulses from the XL-DC receiver are related.
2. (photo 2) almost same as above, but zommed in at 100 nanosec/div horizontal resolution at leading edge of 1PPS pulse and by using "infinite persistence" display option.
   • channel 1: 1PPS output of VME-SG receiver (TTL level)
   • channel 2: IRIG-B DC-shift signal of XL-DC receiver after converted from optical pulse sequence from 2km fiber from Radon Hut.
   Here, the VME-SG's 1PPS signal is clearly not stable compared with the IRIG-B pulse sequence (or vice versa). Notice the almost discrete 100 nsec steps in the signals at -300,-200, and +400 nsec offset from the horizontal center. This indicates that the VME-SG module is generating the 1PPS pulses only slowly synchronized by the IRIG-B signal from the remote XL-DC receiver by using 100 nsec steps from its internal 10 MHz clock.
   This is good enough for 1 µsec resolution, but too jittery for our 100 nsec goal for K2K. Alternative solution: instead of using the VME-SG's 1PPS pulse, I'm planning to add a circuit to the optical fiber interface that extracts the 1PPS pulse directly from the IRIG-B sequence by identifying the IRIG-B "frame" leading edge. This can be done with a minimal jitter of just approx. 5 nsec worst case compared to the IRIG-B signal. Downside: If there's a power outage in the Radon Hut, then there's nor free running 1PPS pulse either ...