Ice. Cube <-> AMANDA Clock Integration: Keeping Time Lawrence Berkeley National Laboratory Gerald Przybylski March 21, 2005 Ice. Cube Collaboration Meeting
The Goal • Synchronizing Ice. Cube and AMANDA events to 10 ns or better (do we all agree 10 ns is a good goal) • Retain a strategy for recovery from fault • As Simple As Possible • Robust
Multiple GPS Clock Limitation • Spec: +/- 150 ns WRT UT (40 ns RMS) - Propagation path variations - Signal Reflections - Satellite clock fluctuations - Constellation variations across multiple clocks => won’t make 10 ns ‘requirement’
Signal Distribution • From (permanent) ICH, Send 10 Mhz, 1 Hz (or 1 pps), and Serial String • In MAPO, Regenerate TTL 10 MHz, 1 Hz, and RS-232 Serial in MAPO (GPS clock equivalent)
Distribution paradigm features: • Signals from Ice. Cube look like signals from a local GPS clock => easy switch over to local back-up GPS clock in case of outage in ICH. • Easy to understand debug. • Fixed time offset between MAPO and Ice. Cube counting house => constant in analysis software (the time offset of AMANDA hits not free fit parameter)
MAPO-to-Ice. Cube • AMANDA sends a trigger system signal (which it also time-stamps locally) to Ice. Cube • Ice. Cube triggers on and time-stamps the AMANDA signal with a DOM MB - Ref signal fed into a PMT signal input
Feedback Features • By itself, adequate for merging events - Interpolate time corrections between time events • Constitutes a Check on distribution system functionality and stability • Provide drift compensation data arising from temperature coefficient of fiber
DGPS? • Differential GPS can provide millimeter position resolution… Can DGPS provide nanosecond clock correlation between multiple GPS clocks at pole? • Used extensively in surveying, construction, road building… • Requires ‘base station’ with accurately known position (but pole station ice moves) • We’d require no custom time distribution hardware; no fibers needed, optical drivers, demodulators • 100% off the shelf hardware
DGPS Unknowns: • Is a DGPS solution even feasible? - No. There is no commercial system that does the corrections on the fly. • Will different brand/model GPS receivers be required? - 15 ns RMS seems to be the best you can do for phase error correcting GPS clocks. • Will there be any objection to placing a DGPS beacon? (it’s a radio source) - Moot point • Will DGPS be too costly? - Moot point
Action items • Specify distribution hardware; Start building **Miteq – SLL - 0. 005 -3 GHz Link - $2600 per Tx/Rx Pair *Fiber Span - AC 300 - 0. 1 -2. 2 GHz Link - $2700 per Tx/Rx ? Fiber-Span - AC 102 - 5 -500 MHz Link -$? Is there a product for DC to 500 MHz? 3/27/05 still looking *Some evaluation done by RICE collaboration ** Adopted by RICE
Additional action items: • Confirm installed fibers -BK thinks they are already installed -Darryn probably knows whether they are installed + Christian Spiring confirmed 12 multi-mode + many single-mode • Confirm that 1 Hz (or 1 PPS) isn’t too slow for fiber -Typical fiber link 50 KHz and up: Modulated? • Confirm that 9600 baud 8 N 1 isn’t too slow for fiber -Typical fiber link 50 KHz and up: IRIG- B better than serial overcoming fiber link frequency limitation
Commercial Fiber Distribution Products • Symmetricom - Model 6530 A and 6531 A Carrier Only, no 1 pps, no serial, no IRIG • Timing Solutions Corp -Model UTCG and TCT 10 MHz, 1 PPS, Serial, IRIG-B, <5 ns error, 200 ps jitter -Acceptable MTBF numbers -<$13000 including both ends -Track record with NASA Deep Space Network -Long commitment to this product line
UTCG + TCT solution Ice. Cube Counting House MAPO Symmetricom ET 6000 UTCG Not relevant 10 MHz + 1 PPS + IRIG-B Local GPS clock Backup AMANDA DAQ Components of sketch borrowed from Timing Solutions model 4300 manual LBNL GTP 5/28/05
Скачать презентацию Ice Cube — AMANDA Clock Integration Keeping Time
f9e8cc132d3d387ff8c62cbb5634033b.ppt