1. Timing and synchronisation J. Jones, A. Moss and E.W. Snedden April/May 2015

2. RESULTS: DRIFT

3. Method Monitor relative phase between master oscillator (MO) 3 GHz and 81 MHz channels using 40GS·s -1 LeCroy scope; compare with laser signal from photodiode: – Trigger scope from photodiode, measure drift of MO signals as changes in time of first crossing point – Slew rate determines resolution: Laser – 3 GHz: ~10 ps shot-to-shot; ~2 ps with 5 minute averaging Laser – 81 MHz: ~20 ps shot-to-shot; ~4 ps with 5 minute averaging Photodiode used has lower bandwidth than 3 GHz diode used in previous studies (currently in use elsewhere), hence lower resolution (2 ps for Laser – 3 GHz possible with fast diode) Monitor temperature changes using localised air probes and MR’s Picolog recorder (always running) Measure VELA crest phase using standard operating procedures

4. Results Scope run 1: 23 rd – 27 th April Results: Friday to Sunday (midnight) – Significant, near-linear phase drift of laser- 3 GHz: ~3 ps·hr -1 ≈ 3° hr -1 200° change in two days! – Small drift of laser-81 MHz: ~0.8 ps·hr -1 ≈ 0.02° hr -1 – Day-night cycle accounts for variable 0.5- 1.0 °C change in room temperature Results: Monday morning – Phase jump at ~10 am in both laser-3 GHz and laser-81 MHz Sudden change in room temperature at same time Log notes laser was switched on at this time (laser room door opening?) – Phase jump at 3 pm in both laser-3 GHz and laser-81 MHz Sudden change in room temperature at same time Noted by operator (BM) as change in crest phase by 50 degrees, measurements at 11 am and 3 pm Mon. 27Sun. 26Sat. 25Fri. 24Thu. 23 VELA log 15:35 ”Something changed in RF system. Crest phase changed to 112 degrees.”

5. Results VELA crest phase: 27 th April Compare changes in VELA crest phase (monitored at regular points during operational shift) against drift of 3 GHz – laser Phase drift of VELA can be almost completely explained (i.e. within an order of magnitude or better) by 3 GHz – laser phase drift: – This excludes second-order effects from temperature changes to cables running to RF, drift of the RF system independent from the MO, which will exist but cannot currently be quantified from laser room – Only by eliminating 3 GHz phase drift can we begin to investigate smaller drift contributions

6. Results SLAP: evidence for 81 MHz – laser drift? Scope data indicates small phase drift of laser and 81 MHz, but magnitude is comparable to noise in measurement SLAP has ‘Fundamental Phase Error’ readout, derived from heterodyning the 81 MHz from MO and laser oscillator, but value is not calibrated – Error is used by the SLAP to tell if there has been a change in absolute phase to which the laser oscillator is locked – System does not explicitly treat drift, but will fall out of lock if the absolute phase changes by a certain (sadly unspecified) value. Fundamental error should therefore be representative of the relative phase difference between 81 MHz – laser could be a more precise metric if calibrated Record variation in error over hours which generally tracks room temperature; fast change in error when laser amplifiers deactivated – SLAP is enclosed in a rack also housing the laser amplifiers, which generate significant amounts of heat: turning the amplifiers off will change the temperature significantly Fri. 24 101112131415161718 22.8 23 23.2 23.4 23.6 Time / hours Temperature /  C 101112131415161718 0.04 0.045 0.05 0.055 0.06 0.065 0.07 Fundamental phase error (arb. units) Laser amplifier off: expect temperature in amplifier rack, where SLAP is, to change dramatically Laser room air temperature, not in amplifier rack: does not immediately respond to amplifiers

7. 024487296 -150 -100 -50 0 50 100 Relative phase drift / ps 3 GHz - Laser 81 MHz - Laser 024487296 22 24 26 Time / hours Temperature /  C Results Scope run 2: 27 th – 30 th April Results: Monday to Thursday (late morning) – Significant, near-linear phase drift of laser-3 GHz: ~3 ps·hr -1 ≈ 3° hr -1 – Negligible drift of laser-81 MHz – Small phase jumps on Tuesday correlate with sudden changes in room temperature (~0.5°C change) Decision made on Wednesday to move temperature probe closer to MO: – Now attached to same rack, but still only measuring air temperature – Air temperature almost 3°C higher at MO position: MO is directly next to oscilloscope, quantum composer, pulse generator: all electronics running 24/7 Fri. 01Thu. 30Wed. 29Tue. 28Mon. 27 Move temperature probe next to MO

8. Results MO rack temperature: 30th th April – 5 th May Monitor laser room temperature (air probe above laser table, close to amplifier rack), 1 minute average Laser phase measurement no longer available due to experiment conflict Average temperature varying by >1°C day- to-day, likely with ambient temperature (can’t find temperature logs for Warrington!) – Target should be <0.1°C – currently running an order-of-magnitude higher Temperature spike peaking Friday at 10 am: – Log notes laser turned on at 9:30 am – sudden rise due to laser operator moving close to MO? Fri. 01 Thu. 30 Mon. 4 Sun. 03 Sat. 02 Tue. 5 Wed. 29

9. Discussion What do collected results suggest regarding the current state of VELA? Clear: – 81 MHz – 3 GHz of MO are drifting by ~3-4° per hour Similar behaviour observed on ALICE, property of MO alone; ALICE solution was to implement feedback loop to stabilise drift Unclear what impact feedback will have on short-term jitter → should be investigated – 81 MHz – laser drift is <1° per hour, currently negligible in comparison to 81 MHz – 3 GHz – Operational drift stability of VELA is currently determined by 81 MHz – 3 GHz phase drift Smaller contributions to drift from RF, cables (etc.) can only be quantified once this primary contribution has been addressed – Sudden changes in laser room temperature (temperature in vicinity of MO and SLAP), such as those produced by entering or leaving the laser room, result in significant phase jumps (>10°) Further work required: – Limited evidence for drift between 81 MHz and laser oscillator, but quantitative analysis including calibrated phase measurement required: If drift contribution from 3 GHz is addressed, this may become the limiting factor Scope approach limited in temporal resolution due to low slew rate Error signal from Synchrolock is currently not calibrated but could be suitable