This page:

TOC/References - Notes - Comb - Ion - Cavity - DAQ

Archived notes: index_2019 index_2020 index_2021 index_2022spring index_2022fall

News and notable results

Experiment/Programming TODO:


KSK105-TE3 sensor close to wall (input side of 674nm cavity) - larger airflow here from vents in floor?

100MHz from distribution amplifier 2m + 2m aircell7 cable -> +0 dBm @ 100MHz on spectrum analyzer
maybe check input-level of distribution-amplifier at some point, increase gain of amp, and/or input

TF OSA-measurement of M-NIR comb output. 1541nm seems to have more power than 156x nm.

Try slow integrator on FALC - seens to lower in-loop sidebands seen on PDH error-signal (around 25 MHz EOM frequency)
Falc slow integrator pos6: 400S/s 3min 2023-03-29_run1.txt
Falc slow integrator OFF: (large maybe 330Hz sidebands on PDH-signal around 25MHz??) 2023-03-29_run2.txt
Falc slow integrator pos5: 2023-03-29_run3.txt
wider view of beat-note shows sideband at ca 523-528 Hz also?
Falc slow integrator pos4 2023-03-29_run4.txt
pos 3 2023-03-29_run5.txt
attempt at SLI pos2 causes laser to unlock
FLI pos8
FLD pos6
laser unlocks also at SLI pos3, leave at pos4?


461nm thermistor 4.642 kOhm, current 199.82mA
check against helper-dispenser - old values are OK against dispenser-fluorescence

13:48 load with 1.75A
14:11 no ion loaded
B-beam hits outer shield hole - so either beam or shield has moved?
move outer shield towards helres (steel valve limits movement) - so B-beam has moved?

14:14:30 try load again? with 1.75A
14:21 no ion
check camera alignment

remove outer shield
remove mid shield
align B-beam based on electrode shadows/diffraction on B-output side

16:34:30 try to load with 1.75A
got diffuse ion loaded
increasing V34diff from 66V to 96V makes ion-image sharp again
align all beams A,B,C onto ion

re-install mid-shield and outer shield
refocus camera - but no clear ion?

17:13 try to load again
17:29 no ion.. stop loading
turn off lasers

To avoid removing camera-objective, install XYZ-stage on a linear rail that allows moving whole assembly in/out?
Currently XYZ-stage mounted to upper breadboard on 19mm high 1" posts (two posts diagonally)
XYZ stage is Thorlabs MT3/M (?) 12.7mm travel
MT401/M base plate with M6 bolts on lowest stage for mounting in bolt-pattern: 25.4mm x 74.4mm nominal
MT401/M base plate width is 108mm

3" rail mounting-plate e.x. XR25-RP1/M has suitable 25x75mm M6 thread pattern?
-- rail-plate height 23mm

3" rail 150mm long XR25DR2
-- rail height is 12.5mm

total height 23+12.5 = 35.5mm, but platform mates with dovetail to reduce this slightly

09:45 comb to H-maser lock

1348nm fiber-noise servo at side of cavity:
75MHz AOM shifts away from SMA-connector, so frequency shifted +75MHz for light towards comb(?)

178.5MHz fiber-noise for 674nm to trap
- old VCO output ca +5.5 dBm
- after attenuator box ca +2 dBm
- from new 19" rackmount VCO main output +2.1 (+2.3 with CPL unterminated) dBm at 178.5 MHz  (-5.9 dBm from CPL port with no connection on main-out, -7.5 dBm with termination)

674nm cavity mode for clockrun, old value ca -15.4 MHz
with +75MHz shift, should be at +59.6 MHz?

beatnote 87.08 MHz, 
higher piezo voltage, now 87.75 MHz

mode close to wavemeter 1348.052nm has beat-note at 59.68MHz?

change f0 sign: f0-beat moves lower makes beat-note nove lower -> f0 has positive sign (?)
now beat note at -10.3 MHz (-70 MHz from previous - OK)

NOTE: tracker bandwidht of 960 kS/s which works when comb in optical lock seems to not work with H-maser lock
with low BW tracker beat1+beat2 sanity-check sum is wrong by some random amount e.g. 17Hz 170 Hz or similar
Increased SDR bandwidth to 2 MS/s -> now get valid data with beat1+beat2=frep sanity-check as usual

Valid data from new setup ca 13 local onwards
With this setup, frequency from comb is computed with: (frequencies are relative to AHM3)
    frep = 249999896.0
    ncomb = 889558
    f0 = +35e6 # positive f0, AW2023-03-28
    f_fibernoise = 75e6  # reference at 150 MHz (check exact DDS FTW from ARTIQ! quantization error at 4e-6 Hz level)
    return f0+ncomb*frep-fbeat-f_fibernoise  # 1348nm frequency, drift-compensated
    # 674nm fiber noise +178.5MHz + double-pass Zeeman AOM to reach ion SRS


09:40 relocked new cavity, beatnote 13.615 MHz

13:23 relock laser again, 13.6175 MHz beatnote

19" racks above big optical table, ca 54cm high, depth 67-68 cm

TF work with 2um output from comb M-NIR
re-aligned 1348nm beat-note optics in menlo box (some optimization possible still, ca 900uW from 1348nm laser, maybe 350uW before fiber-collimator to beatnote detector)
beat-note now at 8.1 MHz (comb-mode changed?)

Try blocking light-path towards ion (optical feedback from splitter, retroreflector, AOM etc. in ion-branch, side of cavity?)

blocked ion-branch: 300S/s 5min 2023-03-27_run1.txt


16:35 lab T-sensor unboxed.

19:05 new clocklaser relocked. seems fiber-noise servos were locked, old laser was locked, and comb was locked ok.
beat-note and tracker ca 13.62 MHz 

laser stays locked only <60 mins!?


11:05 local. T-sensor placed in a Finnfoam box (30 mm wall thickness) next to T110.


16:13 local. Temperatures up to 26C detected in the lab (using IR thermometer). Moved the in-loop T-sensor back to the optical table next to T110.

10:20 local AC in-loop T sensor moved to right above an inlet vent and surrounded by FOBS (Feedback Optimizer / BBR shield)


DM Technologies MCS test. 3 channels (1-3) in parallel, in 4mA range to give total of 11.49375 mA through 50 ohm resistor.
Data: 2023-03/2023-03-21_CH1_thru50R_1d.txt


DM Technologies MCS test. 11.49375 mA through 50 ohm resistor.
Data: 2023-03/2023-03_CH1_thru50R_1d.txt

Rsync mirroring partly fixed. There are now 1.3G of images the script has to go through every minute.
Not good. Moved monitoring figures (DAQ, comb, ion) to a separate script that now runs every minute.
Rest of the data is not yet automatically mirrored
11:05 local: moved AC in-loop temperature sensor from ceiling next to T110 on optical table.


problems with mirroring data from strontium website to monitor website?
- killall mirror-to-monitor
- killall rsync
(this reduces number of processes from 7000 to 300)


old cavity unlocked sometime during night (too high PDH-signal saturates detector?)
reduce EOM drive to -14.2 dBm
relock laser and comb
9:22 ok beat-note data collected again, beat-note 12 939 650 Hz
09:24 old cavity power stabilization on, set-point 3.5 V

10:40 new cavity drift down to -110 mHz/s
old cavity drift estimate +18.4 mHz/s

10:50 open old cavity doors
acoustic sensitivity high close to 674nm input-fiber / output-end of cavity

1348nm power at comb, 5.9 mW (before fiber-collimator for beat-note detector)
doubler temperature setting +48.7 C
0.78 mW 674nm output out of doubler (FC-connector on fiber-output of doubler)
doubler temperature-controller supply was siglent psu above trap, +5VDC-10.5

Fiber-noise setup (1348nm fiber to comb)
AOM is AA MT80-A0.7, nominal RF-power 2.2W, max 2.5W, diffraction efficiency 70-75%, variable frequency shift 80+/-21 MHz
AOM PA is AA AMPB-B-24-10.500. supply 24VDC 490mA, Gain 25 dB nominal, 34 dBm output at 1 dB compression
VCO-output attenuated 3+3 dB and low-pass filtered Mini-Circ SLP-100+

initial fiber-noise setup built.
1348->674nm doubler intput fiber + doubler + output-fiber remain unstabilized.
split same 150 MHz reference from artiq dds to both new cavity and old cavity IR-fiber stabilization setups.
20:10 close cavity doors.
new fiber-noise servo AOM PA is inside cavity-box, hopefully heat-budget still ok.
old cavity transmission slightly lower now, power servo on with 2.5V set-point

beat-note from new cavity ca 14.274 MHz

SDR data with fiber-noise, doors closed, new cavity fan on: 300S/s 3min 2023-03-17_run1.txt
fan off 300S/s 3min 2023-03-17_run2.txt
20:24 new cavity fan back on


new EOM-resonator was tuned on frequency yesterday, then coils epoxied for mechanical stability.
Try new resonator!?
resonance still needs tuning +400 kHz (tuning was needed +2 MHz yesterday)

09:50 Set all Vescent slice-qtc channel set-points to -1 C

15:00 kaukolmpökatko ohi

new EOM-resonator
higher drive level from Toptica windows program ca -10dBm, later reduce to -12.2 dBm
same LO angle seems optimal 218degrees
SDR data with new EOM-resonator, 300S/s 5min 2023-03-16_run1.txt
No clear improvement from new EOM resonator
beat-note while scanning tone from computer loudspeaker -> beat note is quite sensitive to acoustic noise!
sdr_32k_2023-03-16_run1.dat 32 kS/s sdr-data during loudspeaker tests
try again with beat-note better centered for audio-band sdr_32k_2023-03-16_run2.dat 32 kS/s sdr-data during loudspeaker tests


test new EOM resonator?
before tests sdr ch1 20 000 796.64
ch2 17 474 383 Hz

Toptica EOM drive 25MHz, -16.2dBm, input level -10dBm, phase 218.6deg
unlock comb, unlock laser, test new resonator with EOM -> resonance ca 23MHz, reduce inductance?
back to old EOM-resonator, lock laser, hard to find same cavity+comb mode?

new rep-rate measurement 20 000 822.993
new beat-note 11 732 500 Hz (drift is negative)

16:15 2nd iteration of eom-resonator
16:25 relock laser and comb with old resonator


09:18 sdr-data 1000S/s 5min: 2023-03-14_run1.txt

Note some aliasing of >150Hz noise when SDR-counter at 300S/s

10:30 restart driftcomp, limits are [5, 30] mHz/s - this should avoid problems with zero drift
11:21 fix sdr-tracker rep-rate measurement, correct reading is ca 20 000 796.6 Hz
13:30 fix further problems with 1000S/s ZMQ-output (rep-rate measurement had errors)
18:00 back to 300S/s counting and zmq-output ( 1000S/s was causing glitches? )

Measure seismometer X, Y, Z spectra with FFT-analyzer.
SDR data during seismometer measurements, 1000S/s 3min: 2023-03-14_run2.txt

18:05: SDR data after change to 300S/s, 5min: 2023-03-14_run3.txt


13:23 flange temperature ca +21C
drift now low, ca 300 mHz/s
13:24 turn off fan, to measure beat-note
fan off, 5min, 300S/s 2023-03-13_run1.txt
collect also direct RF-samples at 32 kS/s (via audio output program), mixed beat-note ca 2.85-6 kHz
(modified SDR bandwidth to 960 kS/s, decimation by 30x gives 32 kS/s)

89/91 Hz is KSK302 blade-pass-frequency, 239 Hz is KSK105 blade-pass-frequency

RF-beatnote PSD vs PSD( f-counter ) comparison. RF-data at 32 kS/s, counter data at 300S/s
Dividing PSD(f-counter) with f**2 gives similar PSD

mount fan to cart, not top surface of AVI-platform
15:50 drift changed sign and now negative?
16:55 -268 mHz/s (from 2 min plot)

Try added weight on top of cavity-box, to damp vibrations?
before weights, 3min, 1000S/s 2023-03-13_run2.txt
repeat 3min, 1000S/s 2023-03-13_run3.txt
before weights, gnuradio beat-note spectrum: carrier -35 dB, 90Hz peak -47dB maximum, varies a lot
17:30 add two stone-plates on top of box, ca 70cm x 50cm x 5cm and 70x50x3 cm

SDR data after stones 3min, 1000S/s 2023-03-13_run4.txt
repeat 2023-03-13_run5.txt
not much difference between no weights and with weights
add more M6 screws to plywood panels, tighten screws (input end and one side still missing screws, fix problem when box is open next time)
tighter M6 panel screws: 2023-03-13_run6.txt
collect more SDR data at 32 kS/s via audio-program


linux-computer was at 100% cpu - kill all old processes.
problem probably bartington voltage-plotting, too many points in one week plot. plotting every 10 minutes overloads machine?

reset driftcomp-program - it has problem when drift goes to zero -> update interval becomes (infinitely) large
(causes driftcomp aom, and comb + beat-note to jump slightly)
reset gnu-radio sdr tracker
stop bartington logging/plotting

12:20 reduce set-point by -2C to -0.95C
beat-note 17.309 MHz


09:47 reduce set-point by 2C to +1.050C. Vescent cools with max power 16.5W

KSK105 in -3 floor, VFD 35Hz, ca 2100rpm (- slip maybe 2-3%?), dominant sound maybe 239 Hz - could this be vane/blade passing noise as for KSK302?


SDR 300S/s data, ca09:45 2023-03-09_run1.txt
visually high peaks now at 83-84 Hz - does this mean KSK302 is running on new lower frequency?
10:40 KSK302 undergoing repair currently (KSK302 off, PKxxx two motors also off)
10:41 more SDR data: 2023-03-09_run2.txt

try also with new cavity cooling-fan off 2023-03-09_run3.txt
fan back on 2023-03-09_run4.txt
With new cavity fan OFF, quitest beatnote PSD we have seen?
Effect of new cavity fan quite large (but not at 89/91Hz where usual KSK302 peaks are?)
Compare to 2023-02-20, 2023-02-18, 2023-02-16 (run5 KSK302=35Hz)

11:40 turn off fan, try to identify remainng peaks?
fan off (no other changes, similar to run3 above), 2min data at 300S/s, rigol-oscilloscope on box is off 2023-03-09_run5.txt
turn off siglent siggen + second rigol scope on top of box: 2023-03-09_run6.txt
close computer-cabinet back door (directs sound towards cavitybox??) 2023-03-09_run7.txt
close door between small lab and big lab? 2023-03-09_run8.txt
11:52 new cavity fan back on

14:50 fan off, for testing
KSK302 now back on? peak at 90Hz? 91Hz
input-side door open: 2023-03-09_run9.txt
Coolinglaser vescent PSU has fan, close to clocklaser ECDL. try with vescent off. 2023-03-09_run10.txt
vescent back on 2023-03-09_run11.txt
15:42 new cavity fan back on
new KSK302 VFD on 40Hz, motor rpm 800, ca 6 A and 248V -> 1.5 kW power!
17:28 fan off, KSK302 on at 40Hz: (5minutes) 2023-03-09_run12.txt


10:35 local: 120 mm fan now cools the new cavity chamber (V = 7 V)
12:55 +3C set-point reached
steel flange reads +21C with IR-thermometer (orange tape is measurement-spot)


log bartington Z-direction with DMM.

13:19 lower set-point by -3C to +3.050 C. Increase max current to 2A, increase max power to 12.5W. Vescent displays current 1.24A after set-point change.
beat-note now at 10 MHz. should move < 2MHz/C so stay below 16 MHz?
Change analog-out and analog-in offsets to be 0 V.
16:53 increase vescent max power to 16.5W
steel flange reads +29C with Tenma IR temperature-meter
Date          Predicted 0-CTE
2023-03-03    -3.52 C
2023-03-04    -4.20
2023-03-05    -3.23
2023-03-07    -2.32


TEC in new cavity: Stonecold TEC1-07103, 18W, 8.5V, 3.3A, max delta-T 68C.


16:30 set-point to +6.050 C. Vescent current-limit at 1A, +8C set-point TEC current was ca -580mA.



17:10 turn off coolinglaser, no ion/experiments over weekend

ca10:30 SDR data: 2023-03-03_run1.txt
KSK302 now fixed 40Hz (?)

09:30 slope -21.3 Hz/s, beatnote 1.82 MHz. TEC power ca 1.64-1.66 W, current 430-434 mA
10:37 slope -20.3 Hz/s
13:20 slope -18.8 Hz/s
14:47 set-point to +8.050 C. beat-note at 1.44 MHz, will go through zero at some point.
Test with AVI-table off: 2023-03-03_run2.txt
AVI-table on: 2023-03-03_run3.txt
(see similar result 2023-02-07)


no ion? loaded Friday 17.2, lost 28.2 or 1.3
11 try to darken PMT with alu-foil, so that lab-lights don't give much counts
1.4(1.2) and 1.5(1.2) counts in 10ms now (N=100 measurements), front lab lights on
0.8(1.0) and 0.7(0.9) front lights off, back lights on
0.6(0.8) and 0.6(0.8) front and back lights off
A-beam 422nm causes some scattering/stripes on camera image? beam misaligned?

0-CTE is maybe 2-3 degrees away now?
14:23 slope -39 Hz/s
15:50 -37.4 Hz/s


09:10 beatnote, before new KSK VFD, so KSK105 runs on direct 50 Hz. 2023-03-01_run1.txt
10:19 possibly KSK105 off now? 2023-03-01_run2.txt
13:35 VFD-replacement still ongoing, lab temp rise.. 2023-03-01_run5.txt
50Hz vs. OFF
temperature close to normal around 17, new VFD on 35Hz setting
17:01 SDR data with new VFD 2023-03-01_run6.txt
35Hz vs. 50Hz
35Hz vs. OFF
VFD replacement took about 4..5 hours

ion went dark sometime during the night
09:26 coolinglaser max power, all beams
no ion with usual recovery procedures..

09:12 slope -32 Hz/s
change slice-qtc settings for analog in/out for colder temperatures (2023-01-25 settings Out1: 1V/C -20V offset, InA: 1V/C +20V offset)
Now -10V and +10V offsets, no change to gains: Out1: 1V/C -10V offset, InA: 1V/C +10V offset. This changes effective set-point slightly.
adjust temperature range for all sensors to match colder values, range +5 to +15 C - this gives more resolution for the measured values?
09:52 lower set-point by -2C to 10.050 C
13 laser was unlocked at some point ca 12:30
13:27 relocked, beat-note ca 8.3 MHz
TEC current 684 mA, power 4.08 W
16:43 relock again, beat-note 7.72 MHz, slope -53 Hz/s

Change Bartington orientation to Z. Display 45.94 uT, voltage 0.4595 V
Log to same database, but plot new figures
2023-03-01_Bartington_1w.txt Stop measurement 2023-03-03 afternoon 16:30 local.
Silver 5mm L-shaped allen-key is significant source of B-field! remove completely from lab these??
also rainbow-colored Menlo allen-keys


evening: lab lights on, big PMT dark-counts..
try 25dB/90ms clockrun, n_shots 15, g1=0.1, cycle time 24s
change to g1=0.5 and n_shots=30 - runs show large EQS now, why? locked to wrong peak, sideband?
restart with reset peak-positions

recover ion with usual procedure
try 27dB/160ms clockrun, n_shots=15, g1=0.4 ?

24dB/80ms spectra, with sidebands?
RID26407 24dB/80ms, 1 Hz step, 200 Hz span
SDR data during C1+ scan 2023-02-28_run1.txt
KSK103 now on 'max' 50Hz
08 local temperature rise (central heating cut?)

13:10 slope -61 Hz/s
17:00 slope -55 Hz/s


13:30 clockrun still running, servo-cycle 3033
stop clockrun to measure clearout etc.
B-counts 50.6 / 3.9 bright/dark in 10ms - but clockrun histogram only at win2=25, win1=30-35 ??
repeat B-counts 44 / 3.9
B-beam 422nm maybe 0.2-0.3 uW (hard to measure..)
ASE Repumper 11.9 mW
ASE Clearout 52 uW
ASE-alignment gives increase of 100 counts from 500/100ms to 600/10ms
B-beam alignment doesnt improve counts
PMT/Camera alignment gives very small count improvement

13:45 try 20dB/30ms clockrun, now with usual setting of 5 ms clearout - to see if win2-counts imrove?
bright/dark threshold is THR=6 counts. Count < THR is dark, Count >= THR is bright.
14:01 no win2 counts in 15minutes, 36 servo-cycles
14:03 switch to 25dB/90ms clockrun, 5ms clearout
14:40 restart with new PMT-threshold = 13 (based on B-counts 64.6/2.8 (default setting is 100 repeats of 10ms gate)

Try to take also continuous driftcomp value from ion clockrun data
Reads ion drift values from database during the last 10 minutes, sanity-checks values, and uses this data (if available)
If ion data is not available, fallback is 1h drift against maser
18:00 clockrun with 25dB / 80ms
18:14 24dB better attenuation for 80ms?
18:25 holy-blanket on top of PMT to reduce dark-counts?

13:28 slope -19.1 Hz/s
13:29 reduce set-point by -2C to +12.050C.
beat-note at ca 17.47 MHz
14:37 set-point reached (some overshoot)



KSK103 failed again, switched to 'direct drive' on 50Hz (old VFD bypassed) - howling sound in lab...
Cavity frequency drifts up by ca +50Hz. Maser and ion agree, so this is real drift of cavity+clocklaser seen by both
16:52 Collect Bartington Y-direction data with DMM. to see possible correlation with clockrun B-field (B4 or weighted mean)
DMM reading -52.09 mV, Bartington screen -5.2 uT (meter in 1x mode)
17:55 servos lost lock, restart with comb-linecenter. now 10ms clearout
15dB / 9ms clockrun to find B-field and center servos?
try 9ms and g1=0.06 to center servos before 90ms run start?
18:30 try 25dB/90ms with updated B-value. 47.99 s cycle-time with 10ms clearout.
2023-02-25_Bartington_1w.txt (2023-02-27 14:15 fix 100x-multiplication to get uT, in figure)
driftcomp has problems when cavity drift is close to zero - time interval between DDS steps becomes large, and between drift-value updates at least 20-39minutes which is too large.
19:46 reset driftcomp-program, but this resets drift-DDS and need to find zeeman components again
19:56 restart clockrun

08:20 slope 102 Hz/s
16:23 slope 83.9 Hz/s
beatnote now at 24.5 MHz, so can be measured directly without down-mixing (LO was at 55 MHz)
16:32 switch to direct measurement of 24.5MHz of beatnote, sign of drift should also change, now -82 Hz/s


09:24 ion decided to show up for work after being dark from 15:20 yesterday..
13:00 switch to OD1 filter and only B-beam (was full power, all beams) - if this stores bright ion better?
check uMotion: A 2.9%, B 5.9%, C 8.35%

Run uMotion optimizer

ca 15:40 temperature rise in lab (why?) but KSK103 seems on and at 40Hz setting (new VFD delivered but not isntalled)
Final voltages [-6.47 0.2382 21.5617 67.1439]
uMotion after optimization: A 2.7%, B 3.1%, C 2.9%

with old I2-driver and old -11.4939mA setting, DMM shows -571.3 mV over 50R resistor. Aim for same with new driver
300us peak-search (why C2- missing??)
Disconnect coild I1, I3 from old driver
B1 estimate = 4.822399692372849 (initial values for C2 etc. were probably missing, hence missing spectra)
B2 estimate = 4.813312142290789
B4 estimate = 4.814168233908341
25dB/90ms (is it asking too much to find the carrier in 25 point-scan of width 44 Hz optical = 5 linewidths ?)

try 15dB/9ms clockrun instead, while laser-drift settles?
If the new coil-driver is better, should see this in B-values from ion, with a probe time of 80-100ms?
Data from 2023-02-20 clockrun
19:55 switch to 20dB/30ms clockrun
21:30 90ms clockrun

Start of latest step has steep slope, because max TEC current was raised from 0.2A to 1A. 09:30 slope ca -74.3 Hz/s
15:49 lower set-point by -2C, to 14.050 C


09:00 Ion is still stored.
15:20 ion went dark
all beams, clearout, 405mn on. try detuning scans.
remove OD1 filter -> max coolinglaser power, detuning scan with 405nm on
no ion..(loaded 6 days ago 2023-02-17)

new cavity laser out of lock - try to relock?
09:20 new cavity relocked, beat-note now at 39.8 MHz, mixed down with 55MHz LO -> sdr measures 15.2 MHz
Note this is a beat-note with different sign, frequency now drifts up. Slope ca +17.4 Hz/s
jumped out of lock sometime mid-day
13:17 relock, looks like same mode ase earlier
13:20 lower set-point by -2C to 16.050 C
beat-note 15.434 MHz, expect maybe +10 MHz from the -2C step?

sdr data after relock 2023-02-23_run1.txt

stop constant-current DMM-logging, try to modify circuit and measure again. Datafile 2023-02-22_CCTest_1d.txt (measured voltage ca 500mv over 50R resistor, current ca 10 mA)
log also old I2 current driver voltage over 50R resistor, with black 34465A DMM at 50S/s (19ms aperture). ca 376 mV over 50R, artiq settings is I2 = 7.5667 mA

KSK103 turns itself off randomly, all A/C stops, and lab temperature rises suddenly
end of 2023-02-22, close to UTC midnight. AW manually turned KSK103 on again, and set to higher freq 35-> 40 Hz ca 08:45 local (2023-02-23).

Two events on 2023-02-21, both required manual turn-ON of the VFD to recover.


Test LT3045 based ConstantCurrent source - is it better than our current coil-driver source??
Measure voltage over 50R termination resistor, log with DMM that measures ca 2S/s, into database with tag "CCTest".
Voltage adjusted to ca 500 mV, corresponding to 10 mA current.
See 2023-01-09 for current-stability measurements, Inoise (amplitude) PSD 25 nA / sqrt(Hz) / sqrt(f)
LT3045 datasheet has output noise spec of 0.8 uV RMS / RSET, we use RSET=33 Ohm -> 24 nA RMS (but in 10Hz to 100kHz bandwidth - we care about << 1 Hz?)
measurement started ca 17:37.
previous 1 hour
previous 4 hours
previous 1 day

11:30 ion went dark?
- all beams, clearout, 405nm on, detuning scan
- ion recovers some minutes after detuning scan, while clearout + 405nm were on. (OD1 filter for 422nm in use, no need for higher power this time)

11:50 verified that KSK302 is now in automatic-mode, switching between VFD setting 39 Hz and 40 Hz 
(see 2023-02-16 for VFD frequency vs. vibration frequency numbers)
25dB 150ms spectra, RID26272 0.6Hz step 200Hz span, 333 pts/peak?
(150ms+45ms)*100shots*333pts = 80 min/peak

SDR data during C1+ scan 2023-02-22_run3.txt
RID26270 50ms spectra (start ca 11:47) - morning scan showed reduced sidebands for C4, why?

09:20 slope ca -38 Hz/s, 15:30 slope -31.5 Hz/s

sdr data (start of scan) 2023-02-22_run1.txt
sdr data (end of C4+, start of C4-) 2023-02-22_run2.txt

sidebands with bad vibrations? 20dB 50ms spectra?


Evening, maybe temperature-problems gone now?
RID26252, 0dB 500us
RID26257, 5dB 1ms
RID26260, 10dB 2.2ms
beat-note shows peak jumping between 89 and 91Hz again..
rabiflop with 15dB attenuation doesn't show oscillations
sdr data: 2023-02-21_run2.txt
large 89/91 peaks are back..

attenuator setting can be checked with web-browser:
Automate span and step settings for peak-search. Use 5 pts/FWHM, and 25 pts span.
Testrun with 10 dB RCDAT and 5 ms probe, RID 26222

tau_pi vs RCDAT once more.. C1+
0 dB / 15 dB    1016 Hz -> 492 us ok
5 dB / 15 dB    503 Hz -> 994 us
10 dB/ 15 dB    219 Hz -> 2283 us ?
15 dB / 15 dB   incoherent result - something wrong? large temperature swing in lab again, similar to morning!?
(but no activity in lab for many hours??)

around 17:40 reset gnu-radio (graphics was glitching)
KSK105 (-3) was off !?! turn back on, 35 Hz setting 

RID26217 peak-search with tuned attenuation for each pair?
15/17/21 dB, 0 dB RCDAT, 500us probe
C1+  75.650032, 816 Hz, h 48.1
C1-  75.623238, 842 Hz, h 39.4
C2+  75.663271, 750 Hz, h 37.7
C2-  75.609891, 696 Hz, h 47.0
C4+  75.703401, 783 Hz, h 38.6
C4-  75.569783, 786 Hz, h 39.4

RID26149 peak-search 0dB/300us
C1+  75.64987310928274
C1-  75.6232179209672
C2+  75.66326124610376
C2-  75.6098921652114
C4+  75.70343830545512
C4-  75.56982012201244

if we want equal rabi-frequencies for all components, attenuate AOM RF for pairs differently?
artiq DDS has 0.5 dB resoltion, mini-circ RCDAT-3000-63W2 resolution is 1 dB

flop on C4+
RID    RCDAT/artiq att    R0
26150  0 / 18             1790.8 Hz 
26160  0 / 19             1670.8
26152  0 / 20             1242 Hz 
26162  0 / 21             1116.9
26154  0 / 22             883.3 Hz
26164  0 / 23             764.6
26156  0 / 24             627.7 Hz
26158  0 / 26             555.6 Hz

26196  0 / 15             2444 (this should be saturation point of RF PA, accounting for RCDAT insertion loss 3.3 dB in datasheet)
26194  0 / 16             2229             
26190  0 / 17             2096 (previously 18 dB was close to saturation of RF PA, with RCDAT we have some extra attenuation before RF PA)
26166  0 / 18             1858.2 (repeat of first line above)
26169  1 / 18             1504.0
26172  2 / 18             1331.3
26175  3 / 18             1136
26178  4 / 18             877.3
26181  5 / 18             713
26185  6 / 18             609.6
26189  7 / 18             472.6

26198 flop on C2+ with 0dB/15dB setting
26201 flop on C1+
C4+ 2444 (above)  use 3.8 dB to get C1 freq?
C2+ 1379          1.4
C1+ 1114

Can we get equal rabi-frequencies with tuned dds attenuation?
C4+ 0/19 dB     ->  1564 (RID26203)
C4+ 0/20 dB     ->  1301 (RID26209)
C4+ 0/21 dB     ->  1106 (RID26211)
C2+ 0/16.5 dB   ->  1247 (RID26205)
C2+ 0/17 dB     ->  1168 (RID26213)
C10 0/15   dB   ->  1046  (RID26207)

26149 0dB/300us

422nm power
- maybe only ca 0.4 uW in A-beam on breadboard, typical coupling to B-beam is 10%, so only 0.04 uW?
- try fiber-alignment at laser (work with piezo-slider in/out caused misalignment?)
- after alignment ca 1.8 uW in B-beam before B-fiber (using OD1 filter at 422nm laser end)
B-counts 67.5 / 2.8
C-counts 73.9 / 2.9
A-counts 29.8 / 3.3 (LC 13Vpp), adkust LC-voltage for max: 33.4 / 3.0 with 9Vpp

check uMotion RIDs 26114,5,6
A 2.6%, B 11 %, C 9.6%

detuning-scans (some modifications to lineshape-fit function now in use)
A 26117
B 26122
C 26124

repeat with more points, and further to blue side
A 26126
B 26128
C 26130

run uMotion minimizer
missing last voltages? last run is maybe 25631 (2023-02-10)
[-6.47 0.225 21.75 66.6]
26134 is 15 iterations of Nelder-Mead optimizer
final voltages: V = [-6.4700 0.2334 21.6523 66.9688 ]  (modify program so these are better available, saved to h5 file)
after optimization uMotions 26135,6,7
A 3.1%, B 3.1%, C 1.3%

better lineshape now?
A 26139
B 26141
C 26143

LC-scan after lunch 26147

new detuning-scans after uMotion optimization

after optimizaiton

LC driven by siglent siggen, 2 kHz square wave. Voltage setting is Vpp, so Thorlabs datasheet Vrms = Vpp/2
positions of zero-fluorescence move, contrast changes, etc.
max slope was maybe -80 Hz/s, now slowing down
recent FWHMs compared to Fourier limit

Beatnote data from morning: 2023-02-21_run1.txt
Big rapid jump in lab temperature this morning - have the KSK machines been reset to automatic control?
Compare to 2023-02-20:
-- ionclock lab peak 34 now at 39-40Hz? (but in -3 floor VFD is still on 35 Hz)
-- no 100.8 Hz peak anymore?


RID25065 0dB/300us peak-search with old B-field direction

C1+ 75.650310 FWHM 1308, h 35.9
C1- 75.623064 FWHM 1377, h 29.1
C2+ 75.663655 FWHM 1405, h 33.0 
C2- 75.609699 FWHM 1324, h 40.2
C4+ 75.704256 FWHM 1260, h 33.3
C4- 75.568999 FWHM 1298, h 43.8
B4 estimate =  4.832471 (see also clockruns from 1-2 day back)

usual B-beam clockrun currents = [-0.0408, -11.4939, -0.7741]
set A-direction currents [I1, I2, I3] = [8.2497, 7.5667, -2.3043] mA

C1+ 75.650031, 1451, h 19.3  (-280 Hz AOM?)
C1- 75.623272, 1262, h 19.1  (+210 Hz)
C2+ 75.663430, 1460, h 25.2
C2- 75.610005, 1406, h 28.0
C4+ 75.703468, 1349, h 47.5
C4- 75.569874, 1142, h 53.8

B1 estimate =  4.768028
B2 estimate =  4.774902
B4 estimate =  4.773065

RID26068 flop on C1+ tau_pi ca 760 us
RID26070 flop on C2+ tau_pi ca 550 us
RID26072 flop on C4+ tau_pi ca 350 us 
A-direction Bfield, 300us,
RID26077 A-beam LC-voltage scan with new B-direction

RID26078 longer 1000ms gate and more points
schedule LC-scans with 1 hour intervals, to see if results are stable?

07 stop clockrun
try 30dB 270ms spectra?
RID25939 0.3Hz step, 3 Hz span, 11 pts/peak
RID25940 0.3 Hz step, 30 Hz span, 101 pts/peak
Fourier limit 1.48 Hz AOM
C1+ FWHM =  2.489 Hz AOM
C1- FWHM =  1.770 Hz AOM
C2+ FWHM =  1.916 Hz AOM
C2- FWHM =  1.834 Hz AOM
C4+ FWHM =  2.424 Hz AOM (lab temperature remains stable, but something disturbed clocklaser?, or B-field?)
C4- FWHM =  2.202 Hz AOM
are worse C4 results because of increased activity in building on Monday morning - or a sign of B-field noise?

08:55 new cavity drift now -4.8 Hz/s

collect 300S/s SDR data, during 270ms/C2+ scan. '2023-02-20_run1.txt'

New cavity beat-note was mixed down with 50 MHz LO, IF reached ca 800 kHz (RF beatnote on SA ca 49.33 MHz).
13:12 increase LO to 70 MHz, moving beat-note to 20.823 MHz

13:24 lower CH1 set-point by -2 C to 18.050 C

Try to measure LC-retarder voltage vs. ion fluorescence
- initially nice results, but lab-temperature swings causes problems ( A-beam polarization rotates or not pure?)
- drive LC-retarder form siglent siggen, controlled over Ethernet via vxi11-library
- much lower contrast between brightest and darkest setting when laboratory temperature fluctuates

check uMotion (RIDs 25982,3,4)
A 2.9%, B 6.1%, C 3.9%

try half-wave plate in front of A-beam?
initial contrast 220/60  (100ms gate, scan voltage through minimum and maximum)
230deg -> 240/77 (26006)
235deg -> 230/67 
240deg -> 240/59 (26004)
245deg -> 240/69 (26003)
250deg -> 240/100 (RID26001)
255deg -> 251/117 (RID26002)
260deg -> 240/120 worse..
270deg -> 280/140

around 240? (increase to 200ms gate)
242deg -> 469/144 (26007)
240deg -> 466/142
238deg -> 450/167 ?

set LC voltage high (zero retardance)
then adjust waveplate for max fluorescence
now measure contrast when scanning retardance?

re-align LC to 45deg from A-output polarization?

LC on rotation stage after A-objective, search for best contrast?
scan voltage 2.2V to 5V in 50 steps (this should include the two minima we want?)
contrast is max(pmt)/min(pmt)
45deg -> contrast 3.1 (RID26045)
40deg -> 3.6 (26046)
35deg -> 9.6 (26051) 1st minimum lower
32.5d -> 9.2 (26053) 2nd minimum lower
30deg -> 10.6 (26055, align beam before run) 1st min slightly lower
30deg -> 15.5! (26047) lower overall counts (beam misalignment?)
30deg -> 12.4 (repeat, 26050) 2nd minumum is low, 1st minimum not so low..
27.5d -> 33.2! (26057) 2nd minimum lower
25deg -> 9.8
20deg -> 4.4

longer runs, repeated
0.5V to 10V in 500 steps, 100+100ms counts
26058 (at end of run, switch to OD1 for cooling)
26059 OD1, 100ms
26060 OD1, 100ms repeat
26061 1000ms gate
Note switch to OD1 filter at end (before that some strange things also?)
lower counts because of OD1-filter


RID25940 40dB 270ms

CTE search so far

Joint I-term agrees with exp(-t/tau) model:


11:30 stop interleaved run

try clockrun with shared drift-estimate between servos
Dube2017 exponential averaging for drift, with update after each Zeeman pair and N=150
with cycle-time of 44s, this gives (150/3)*44 = 2200 s = 37 min time-constant for drift-estimate (not so different from 3600s maser estimate?)
but ion data has lower noise!?

Fix ion vs. maser frequency calculation, for comb in optical lock.
- ion now ca +330 Hz vs. maser

11:50 new cavity drift -9.4 Hz/s

ion went dark but recovers with all beams and 405nm
continue clockrun

13:08 Test how zeeman servos react when suddenly introducing +5 mHz/s drift
1h drift estimate from maser offset by +5 mHz/s, this should drive driftcomp AOM so that ion sees this drift?
13:36 ion drift-term reached 2.5mHz/s, half of applied offset in 18 minutes
(Note: residual drift plot has opposite signs for maser and ion data)
14:28 set offset to zero, maser and ion drift-estimates should agree now

try another +5 mHz/s offset test
set to zero after midnight


Continue with 90ms clockrun where we also store drift-estimate after each zeeman-pair and after a complete clock-cycle.
remove 1ms start_delay (200us delay is needed for artiq), and move TTL-trigger in parallell with probe. This cuts 1.8ms from dead-time, now 43.3 ms.
new field "drift_hz" in ZeemanPeaks database. After clockrun, compare against maser-based drift estimates, and determine best averaging algorithm?
14:14 leave clockrun running.

18:53 switch to interleaved run, with ExtraOrdinary clock using end_delay=0 ms. 80ms probe, g1=0.6, g2=0.1.
Cycle time 86.3s, dead-time 160 ms
19:57 EO-O difference -328(65) mHz (45 servo-cycles completed)

beat-note during 100ms C4- scan 2023-02-18_run2.txt

beat-note data 300S/s: 2023-02-18_run1.txt
morning: New cavity drift -23.5 Hz/s
13:30 drift -19.6 Hz/s
19:00 drift -16.4 Hz/s


inner shield sensors at ca 70% of -1 C step in ca 26 hours
19:48 drift -33.3 Hz/s and slowing down

09:53 drift now 41.8Hz - seems to be slowing down?

Measure old cavity frequency via comb rep-rate?
CH1 input of KK-counter in comb-rack split also to SDR CH1
CH1: comb-reprate measured as (4*frep - 980 MHz)  = ca 20 MHz, where 980 MHz is from AHM3
CH2: beat-note from new cavity

18:08 load with 1.75A
18:22 ion!
to 208MHz detuning, OD1-filter for 422nm
2023-02-17-run1  search for transitions, RCDAT 0dB, 300us probe
driftcomp had terminated (when restarting artiq?) start again.

B2 estimate =  4.830795
B4 estimate =  4.838361

RID25896 10dB/3ms peak search
C1+ =  75.650202
C1- =  75.623069
C2+ =  75.663679
C2- =  75.609598
C4+ =  75.704282
C4- =  75.568985

B1 estimate =  4.834592
B2 estimate =  4.833582
print:B4 estimate =  4.833893288636104

C1 Linecenter =  75.636636
C2 Linecenter =  75.636639
print:C4 Linecenter =  75.6366340926144

RID25897, try rabiflop on C1+, 10dB RCDAT, (spectrum was with 3ms)

RID25901 20dB/30ms peak search
NOTE: span is 100Hz AOM, so 200Hz optical -> we don't see +/-100Hz sidebands very well..
(previously with strong sidebands 1/50Hz = 20ms was a problematic probe length - check what FWHM we get now)
C1+ FWHM 13.8 Hz AOM (Fourier limit is 13.3 Hz AOM, so looks ok)
C1- FWHM 15.7 Hz AOM
C2+ FWHM 14.4 Hz AOM
C2- FWHM 13.6 Hz AOM
C4+ FWHM 13.4 Hz AOM
C4- FWHM gaussian fit fails

Try 25dB/90ms clockrun while we wait for cavity drift to settle
g1=0.6, g2=0.01, n_shots=30, cycle time 48.64, dead time 45.1 ms

RID25907 is 25dB/90ms peak search, 1Hz step, 300Hz AOM span
100ms and 80ms after this
RID25908 25dB 100ms
RID25907 25dB 90ms

RID25901 20dB 30ms

RID25897 rabiflop C1+
RID25896 10dB 3ms

RID25894 0dB 300us


KSK303 seems to be source of 89/91 Hz sidebands
- Auto-mode changes VFD frequency in steps between 48, 49, 50, 51 Hz

KSK302 (coordinate-room)? clear correlation between 89Hz and 91Hz sound measured with phone and VFD reading-changes 
run3: lower fixed frequency setting of 38 Hz on VFD

run4: Try a few different KSK302 settings, 

VFD (Hz)   Vibration (Hz)   ratio
35         80.06            2.2874
38         86.91            2.2871
39         89.19            2.2869
40         91.50            2.2875

dominant peak now 100.864 Hz, if this is a motor, VFD at 100.864/2.287 = 44.1 Hz ?

no changes to KSK303 2023-02-16_run1.txt
KSK303 manual 47 Hz 2023-02-16_run2.txt
KSK302 manual 38 Hz 2023-02-16_run3.txt
KSK302 different values tried 2023-02-16_run4.txt
KSK302 35 Hz 2023-02-16_run5.txt

16:37 Lower new cavity CH1 set-point by one degree to 20.050 C
with comb locked to old cavity, new cavity beat-note at 42.53 MHz (coarse SA value), mixed down with ARTIQ LO of 50 MHz, SDR measures 7.53 MHz

18:27 drift -15.3 Hz/s


Check ASE power (different fiber compared to usual ion-fiber)
measurement at fiber output (no objective as for ion), 10 mW initially, drops to 8mW in a few mintues

shortest pulse from SRS-shutter seems to be ca 5 ms - first tests with state-prep probably doable with this?

collect another 300s/s sdr-tracker dataset, 5 minutes
check with spectrogram how 89Hz disturbance comes and goes?

Piezo shutter notes:
B-coolingbeam passes ok through 2.0mm aperture, 2.5mm gives a bit more margin
A/B beam height above breadboard ca 32mm
ca 30mm extension needed on piezo-bender to give 3mm movement, use 35 or 40mm to have some margin?

89 Hz seems to shift to 91 Hz periodically?

34Hz is AC machine in -3 floor
87-88Hz could be compressor - close to cavity behind one wall
89Hz ??
91Hz ??
Y-axis is sqrt(PSD) so roughly RMS-amplitude of frequency-fluctuation at 222 THz (1348nm)


Ion dark early morning

stop interleave run
does not recover with 405nm, all beams, detuning scans

Try to verify clock-cycle timing

- TTL7 indicates start of clock-cycle
- TTL6 indicates PMT detection windows
- B-beam photodiode detects 422nm and 674nm light
- ASE-fiber InGaAs detector detects ASE+Clearout
test with latest clockrun-program, using 40ms probe
default settings for delays:
- cool 30ms
- repump 35ms
- clearout 5ms
- detect 10ms
- detect2 10ms
- start_delay 1ms
- clock_decay 1ms
- cool_shutter 5ms
- end_delay 7ms
Remove extra delays at start? TTL-trigger and start_delay
- Probe rise/fall-time 10-15us (? limited by measurement resolution)
- Coolinglaser shutter turn-on-latency 2.3ms, rise-time 0.4 ms
- Coolinglaser shutter turn-off-latency 2.8ms, fall-time 0.3-0.4 ms


more interleaved run with modified end-delay?
300us peak-search
C1 Linecenter =  75.63921880938959
B1 estimate =  4.821348723323228

RCDAT 10dB and 3ms clockrun to find coarse peak-positions?
20 dB and 30 ms?

Interleaved clockrun with modified delays:
RCDAT 23dB, 80ms probe, g1=0.5, g2=0.01, n_shots=30
Ordinary: 0.2 ms start_delay, 7 ms end_delay
ExtraOrdinary: 0.2 ms start_delay, 0 ms end_delay
NOTE: comb is now in optical lock, does not provide drift-data for driftcomp AOM (driftcomp left running at last valid value from morning)
This means peak-servos need to correct for more drift, so we expect larger I-terms
Also store ExtraOrdinary servo line-center as "EC124" in database for real-time evaluation of interleaved run

initial drift -13 mHz/s
interleaved servos have long cycle-time of ca 80s
-> I term settle towards drift-per-servocycle = -13 mHz/s * 80s = 1040 mHz/servocycle

difficult to get servos locked, without driftcomp from maser
try 10dB and 3ms run?
0.2ms start_delay
Ordinary: 7ms end_delay
ExtraOrdinary: 0ms end_delay
g1=0.5, g2=0.02, n_shots=30

10:30 Comb back to optical lock, for further cavity measurements

stop driftcomp-program (leaves constant drift for 674nm cavity)

new cavity beat-note at ca 42 MHz, mixed down with 50 MHz from artiq

resulting beat-note at 7.62 MHz

collect 300S/s tracker-data (no shaking or disturbances in lab, hopefully)
to compare against ion spectra

beat-note data at 300S/s run1 run2

Ion went dark, but recovers during 2nd detuning-scan with all beams, clearout, and 405nm on

RID25798 measure 80ms spectra, 1Hz step, 120 Hz span, 121pts/peak, randomized order

RID25800 re-measure ASE-shift? 0 dB, 300us, 300Hz step, 8 kHz span, 27pts/peak
comb linecent =  75.6395344
C1 Linecenter =  75.6395339
C2 Linecenter =  75.6395732
C4 Linecenter =  75.6394949
B1 estimate =  4.838381
B2 estimate =  4.835149
B4 estimate =  4.833428

RID25801 is same 300us spectra, but repumper override ON
comb linecent =  75.639535
C1 Linecenter =  75.635325 (C1+ fit failed, not a good value..)
C2+ automatic fit failed again
B1 estimate =  3.752397484303563  ??? WHAT automatic fit to C1+ peak failed - redo manually

Matlab gaussian fit gives ASE-shift:
C1 -2279 Hz
C2 -2385 Hz
C4 -2010 Hz

09:52 stop Bartington DMM-voltage logging

RID25802 detuning-scan B-beam
Bartington Y-direction B-field logged during interleaved clockrun: 2923-02-15_Bartington_1d.txt
Compare with B from ion?

100ms gate
100ms gate, 100 points:
extend to more blue side
10ms B-counts 49.9(7) / 0.8(0.9)
100ms B-counts 487(25.7) / 8.1(3.1)

RID25798, 80ms
RID25800, 0dB, 300us
RID25801, 0dB, 300us, Repumper ON


ion went dark
clearout on, 405nm, ion recovers with all beams and detuning scan
08:46 conitnue with new bug/nromal run

comb plots were stuck, comment out failing code (ion vs. cavity adev and similar)

09:30 went dark again

16:48 ion still dark? to max coolinglaser
405nm on, detuning scan
set detuning to loading setting ca 190 MHz
empty trap with rf off/on
16:59 load with 1.75A
17:13 ion! pmt counts during loading RID25765
208MHz detuning and slider pos2/OD1
uMotion A 2.25%, B 5.7%, C 4.3% (RIDs25767, 8, 9)
A-beam quite low fluorescence (optimal LC-retarder voltage changed? or alignment changed?)
B-counts 53.9 / 3.5

peak-search with RCDAT 0dB and 300us probe, 300Hz step, 10 kHz span, 33 pts/peak
comb linecent =  75.639537
C1 Linecenter =  75.639564
C2 Linecenter =  75.639558
(turned off 461nm mid-way through C4+ scan)
C4 Linecenter =  75.639613
B1 estimate =  4.841307
B2 estimate =  4.835069
B4 estimate =  4.836053

RID25778 peak-seach with clearout override on, aka. ASE-shift
comb linecent =  75.639533
C1 Linecenter =  75.638495 (-2138 Hz optical ASE-shift)
C2 Linecenter =  75.638342 (-2432 Hz)
C4 Linecenter =  75.638427 (-2372 Hz)
B1 estimate =  4.809794
B2 estimate =  4.852170
B4 estimate =  4.835871
ca 6.8 mW measured with Thorlabs slim-sensor on 500mW setting at 1092nm

Log Bartington B-field, with probe in Y-direction, measure with DMM (NPLC=10), database "Bartington"
DMM reading -54.7 mV, Bartington display -5.469 uT
(previous measurement along clockrun B-direction was 2022-12-16)
toolcart 2nd lowest drawer:
- open:  5.447 uT
- closed 5.470 uT

chagned comb rep-rate SYCNRO I-gain to lower value 12Hz - this maybe improves invalid counts by sdr-tracker
(less back and forth movement of beat-note on SA screen)

try to start clockrun
- 15 dB and 8ms low-resolution clockrun? servocycle ca 20s
- after 20 servocycles peak-servos seem locked, Dsum bounded, B-values agree
- at 18:24 open toolcart 2nd-lowest drawer?
-- bartington 5.441 uT
- no clear signal in ion-data B-estimates?
- at 18:29 close drawer
-- Bartington 5.464 uT

18:30 switch to RCDAT 25dB and 80ms clockrun
18:47 servos seem locked, Dsum reasonable, B-values agree
delay-line with leadscrew on cart: from 5.8 to 5.3 uT Bartington change when rotating cart through pi radians.
- 500 nT change in field seen by Bartington not easily visible with ion
- try 24 dB RCDAT, more counts?
18:53 reverse field by rotating cart? from 5.3 to 5.65 uT when turned by pi radians
no clear changes in ion B-field
- try 23 dB RCDAT?
18:59 increase set-point for cavity power-servo (might change cavity frequency slightly?)
(all servos show high R-counts after cavity-change)
19:03 leave lab and take big elevator up (change in Bartington-meter?)

19:43 try interleaved run with usual 7ms end_delay, compared to 0ms end_delay
Note: clock-cycle has a 1ms start_delay before probe-pulse, so effectively comparing 8ms and 1ms delays for ASE-shift
80ms probe, g1=0.6, g2=0.01, end_delay=7ms, end_delay2=0ms

RID25777, RCDAT 0 dB, 300us peak-search

RID25778, ASE-shift RCDAT 0 dB, 300us peak-search


08:30 ion dark, all beams, clearout on
405nm on, still dark
detuning-scan, still dark

reduce cavity power-set-point to 2.66V (servo could not reach old setpoint 3.0V, BOA PSU beeps when current 600mA? limit reached)
try relocking to Rb-line, but ion still dark
405nm on, 461nm on, detuning scan
ion still dark..
empty trap with rf off, on

17:39 load with 1.75A
initially with 208MHz detuning, changed to 188MHz detuning
no ion...
461nm current was 195.4 mA, check against dispenser? -> 195.2 mA

18:07 try to load again, with 1.75A
18:10 ion!  PMT counts during loading is 25683, but now got two ions...
18:11 try to load again
18:16 ion! counts during loading 25686?
to 208 MHz detuning

uMotion 25689,90,91 (max cooling power)
A 1.5%, B 1.8%, C 0.53%

B-beam counts 250.7 / 4.3 bright/dark
setting piezo slider fails..

pos2 OD1 cooling (T=10%)
B-counts 39.2 / 3.6 bright/dark

pos6 OD0.7 (T=20%)
B-counts 96.1 / 3.5 (bright/dark)

increase TPA drive to 2 A (from 1.9A)
B-counts 114 / 3.7
increase to 2.05 A
B-counts variable 108, 111, 114, 
to 2.1A
B-counts 114, 112, 117

switch comb to H-maser lock, usual rep-rate DDS setting of 19 999 584 Hz -> rep-rate is 250e6 - 104 Hz
artiq LO for tracker ch2: 215 MHz
restart usual artiq driftcomp-program (this resets drift AOM to old value, also AOM drive-level, can reduce BOA current now)
possibly right mode now at ca 15.327 MHz

RF-attenuator on 0 dB, max 674nm power
200us clock-transition search

turn on 674nm trap breadboard power servo, set-point ca 2 V
RID25710 is 300us clock-transition search with updated line-center
updated line-center from comb 75 639 556 Hz
B2 estimate =  4.826500000000882
B4 estimate =  4.826500000000216
(from old clockruns artiq setting was 4.8265 uT)

Notes on RCDAT AOM attenuation vs. probe length:
2022-10-22: RCDAT=25 dB, 80ms probe
RCDAT 0 dB, 300us (2022-10-17)
+5 dB attenuation should give OD1
+10 dB attenuation should give OD2 and 10-fold tau_pi?

RID25713 set 10dB RCDAT, 3ms peak-search, 25 Hz step, 600Hz span, 23 pts/peak
C1 Linecenter =  75.639530
C2 Linecenter =  75.639528
C4 Linecenter =  75.639531
B1 estimate =  4.832118
B2 estimate =  4.832293
B4 estimate =  4.831957

comb data shows lots of invalid points? (could be SRS-shutter shaking table/comb, or something else?)

set RCDAT 25 dB attenuation
RID25716 is peak-search with 80ms probe, 1 Hz step, 100 Hz span, 101pts/peak

try to start clockrun, with 90ms probe, 30 shots, g1=0.6, g2=0.01
servocycle length 48.64s, deadtime 45.1ms
servos seem to converge, Dsum ok

20:46 start interleaved bug/normal clockrun
80ms probe, g1=1.0, g2=0.01, 30 shots (same settings as ROCIT campaign?)
with previous normal clockrun, bright PMT couts ca 110/10ms
servo-cycle length now doubled to 90s

RID25707, 200us peak-search (probably centered on wrong line-center...)

RID25713, 10dB RCDAT, 3ms probe
RID25716, 25dB RCDAT, 80ms probe


Try to test LC-retarder with ion?

all lasers on, ca 190 MHz coolinglaser AOM
new Toptica TPA for coolinglaser, 1.8A seems to give strong Rb-signal and enough cooling-light?
10:42 load ion, 1.75A dispenser
- dispenser 2.2W initially, 2.258W after 1 minute (1.300 V from PSU)
- 10:53 ion! somewhat blurry, micromotion?
detuning to 208 MHz, brighter ion as expected

13:22 uMotion check, still OD0 highest cooling power
A 3.6%, B 8.6%, C 9%

Estimated coil-currents for B-field along A-beam
[I1, I2, I3] = [8.2497, 7.5667, -2.3043] mA
with abs(B) = 4.8 uT

Irises placed on output-side of A-beam, to hopefully get back to alignment easier?
Remove old A-beam mount, to move it backwards and make room for LC-retarder
old A-beam polarization roughly vertical
need to enlarge M4 holes on 1" post-mount, to fit Alu-plate with angled shelf + polaris 0.5" mirror mount for A-objective

modify A-objective so it can be focused further away
- made focusable/rotating lens holder thread 1mm shorter
- made custom 0.5" lens-tube locking ring thinner from ca 2.8mm standard part to ca 1.8mm

re-focus A-objective on ion
first 20-30 counts
objective outwards, 30-40counts
further outwards, ca 80 counts (B-beam saturates ion, ca 280 counts/10ms, this is target for A also?)
further out 100-120 counts
furhter out? 140 
A now 240-220, similar to B?

some issues with Piezo-slider for coolinglaser OD-filters...

16:22 run uMotion minimizer RID25631

after modifications A-beam polarization looks closer to horizontal now?
try to mount LC-retarder at 45-degrees

test LC-retarder with ion, PMT-program collects 10ms gate PMT-counts
siglent siggen drives LC-retarder with increasing amplitude from 0Vpp to 10Vpp
5 s delay between voltage steps
RID 25646 is PMT counts, record is 3000 bins, but voltage scan ends around ca 1400 (?)
at around bin 2200, manually set high fluorescence state, 4.7 Vpp from siggen (residual fluctuations from laser-power fluctuations?)
now ca 120 counts/10ms with A-beam (focus may have shifted from optimized focus earlier done today)

try blindly searching for clock-transition?? (but we don't know if 647nm cavity is on the right mode...)
old settings have C1+ around 75.78 MHz Zeeman AOM..
not clear if 674nm light reaches B-beam or not?

with clock-laser continuously on, change cavity mode until ion is seen blinking on camera - this is the right mode?
(beat-note against vertical cavity ca 19.4 MHz)
now repeat clock-transition search?

PMT-counts during LC-retarder voltage scan:


Testing LC-retarder, LCC1413-A
start at high voltage, retardance=0
6.2 Vpp square-wave into 1MOhm -> detectors equal, so circular polarization!? (lambda/4 retardance)
4.2 Vpp square-wave into 1 MOhm -> detectors reversed from start, so linear polarization, rotated 90degrees (ret = lambda/2)
3.46 Vpp ret = 3lambda/4, other circular polarization, detectors equal
2.93 Vpp ret = lambda, back to high-voltage detector readings

run amplitude-ramp with siglent signal generator, square wave
0V to 10V, 1001 pts, 1s sleep per point
0V to 10V, 101pts, 1s sleep per point
Thorlabs power-meter logging with 0.1 s sample-interval -> CSV file

Note retardance >2 waves, as expected from Thorlabs spec


beat-note at 300 S/s with some changes?
- run1: no disturbances, power servos locked
- run2: new cavity AV-table off?
- run3: AV on again (significant lag in live-plot on linux machine?)

Effect of new cavity AV platform ON/OFF is reduced noise from 6 Hz to 20 Hz


newcavityPD 1w
old cavity transmission photodiode ca 2.8V
09:26 turn on power-servo

Lab temperature jump (whole mikes building?) this morning.
674nm power servo was off, so probably see a mix of old-cavity 674-power, old-cavity fiber, and new cavity temp-response.
Usual temperature sensors in the lab seem slower to respond wrt. laser frequency with fast response?


New cavity output photodiode logged with DMM, while power-servo on:
newcavityPD 1d


shaking tests..

speaker on top of cavity, shaking vertically

5 Hz
70 mVpp -> +/- 4 Hz

log new cavity output photodiode with DMM to database, "newcavityPD"
start ca 18:05

old cavity power-servo off, input-side door open

cavityPD 1d

predicted instability from power measurement with servo OFF
tau_freerun = [  240.,   480.,   960.,  1920.,  3840.,  7680., 15360., 30720.]
adev_freerun = [1.04722469e-15, 1.48801454e-15, 2.55874362e-15, 4.01631998e-15,
       4.51634712e-15, 1.69872031e-15, 1.48093510e-15, 8.99771385e-16]


Old cavity power-stability seems ok over one day.
DAQ measures photodiode every 4 minutes, expected laser-instability from these measurements is low enough (below 1e-18 at 240 s).
voltage is 3 VDC, photodiode sensitivity 0.5 uW/V

Measure also cavity-output photodiode with desktop DMM that samples ca 2 S/s, starting ca 14:30, named "cavityPD" in DMM database
cavityPD 1d


ca 13:30
Connect PTB-box DDS in-loop for old cavity driftcomp-AOM (ca 80 MHz, -6 dBm drive level)

artiq dac ch0 test: ch0 1d


Try to measure power-shift for new cavity

power-servo off -> open-loop measurements
connect modulating signal-generator to RF-attenuator (final AOM before cavity)
200 mVpp sine-wave with 100 mHz frequency -> ca 40 Hz p-p sine-wave on laser frequency
Bench-DMM measures cavity transmission photodiode voltage, variations ca 1.243 - 1.321 V

DMM-data into database, starting 11:50 (modulation 200 mVpp 100 mHz)
SDR frequency data at 10S/s into file: 2023-01-31_powermod100mHz_200mVpp.txt  (2 minutes of data)

11:53 change to 50 mHz
sdr data: 2023-01-31_powermod50mHz_200mVpp.txt (11:54)

11:57 change to 200 mHz
sdr: 2023-01-31_powermod200mHz_200mVpp.txt (11:57 start)
slightly below 40 Hzpp

11:59 change to 10 mHz
sdr, longer file: 2023-01-31_powermod10mHz_200mVpp.txt

13:30 change to 500 mHz
sdr: 2023-01-31_powermod500mHz_200mVpp.txt
lower frequency modulation ca 20 Hzpp

13:37 turn power servo back on, set-point ca 1.8 V

old cavity servo was on during previous measurements, with set-point of 3.1V
connect DMM to old cavity photodiode
from ca 13:39

connect modulating signalgenerator to BOA-modulation on old cavity
100 mHz, 
200 mVpp gives 508-511 mA BOA-current
2.5 Vpp -> , photodiode ca 150 mVpp, BOA 499-520 mA
2 min sdr dataset: 2023-01-31_oldcavity100mHz_2p5Vpp.txt

13:51 5Vpp modulation, photodiode 300 mVpp, laser frequency 3 Hzpp, BOA ca 488-531 mA
2min sdr data 2023-01-31_oldcavity100mHz_5Vpp.txt

13:53 to 200 mHz modulation, maybe 2 Hzpp laser frequency
2min sdr data: 2023-01-31_oldcavity200mHz_5Vpp.txt

13:57 to 500 mHz
2min sdr data: 2023-01-31_oldcavity500mHz_5Vpp.txt

14:00 to 50 mHz, maybe 4 Hzpp laser frequency
5min sdr dataset 2023-01-31_oldcavity50mHz_5Vpp.txt

14:09 old cavity power servo back on, set-point ca 3 V
14:11 stop DMM measurement

DMM data exported to: 2023-01-31_newCavityPD_1d.txt

new cavity photodiode
2 V photodiode, 1.41 uW power-meter (Thorlabs S122C sensor, lambda=1380 nm)
1.5V photodiode, 1.03 uW power-meter
0.705 uW/V, 0.687 uW/V, mean 0.696 uW/V
compare to Thorlabs specs for photodetector: PDA20CS2 60 dB gain

old cavity
total power-output
ca 1.5 uW (before big cube), lambda = 674 nm
upwards beam to photodiode ca. 1.1 uW 
through cube (to camera) ca. 0.19 uW
photodiode ca 2.9 V
sensitivity 0.517 uW/V (total power-output of cavity)

18 old cavity had jumped to TEM01 ??
relock to TEM00, relock comb
old cavity power-servo on, set-point 3.02 V
new cavity power-servo on, set-point ca 1.79 V (red channel on scope)

18:07 test ARTIQ DAC-channel stability
ch0 set to 5.43211 V, measured with DMM, description "artiqch0"

Cavity power-shift data:
For old cavity measurements at 1348nm have been multiplied by 2, so they correspond to 674nm light in the cavity

89-fold difference in fractional-frequency/uW, 44-fold difference in optical frequency/uW.

DMM photodiode DMM
old 50 mHz old_50
old 100 mHz old_100
old 200 mHz old_200
old 500 mHz old_500
new 10 mHz new_10
new 50 mHz new_50
new 100 mHz new_100
new 200 mHz new_200
new 500 mHz new_500

spikes? once per minute?


09 drift now -528 mHz/s
new cavity temperature slopes ch3 -1.1 mK/day, ch4 -1 mK/day
re-tune tracker (glitches ca 8:55)

10 S/s live-plot on windows machine
frequency PSD shows peak at ca 1.6 Hz!

17 switch to another mode on the new cavity - more transmitted intensity -> better mode?? (drift is now positive)
power servo now reduces in-loop AC signal to ca 0.7 mVRMS

live plot on win-machine with 10S/s data:

10 S/s data 10sps_30min

10 S/s data 10sps_1min


14:50 drift down to -727 mHz/s
new cavity inner shield temperature slopes -3.2 mK/day and -3.3 mK/day (ca 4 days after switching on temperature stabilization)

15 collect 30min frequency dataset with 10 S/s

16:30 drift -706 mHz/s

19:08 reduce gain of new cavity power servo?
drift ca -680 mHz/s
19:12 laser unlocked for a short time..
try some comb-settings (glitch around 19:30)
19:32 leave lab
second 10 S/s dataset 10sps_10min_2

10 S/s data 10sps_10min

500 kS/s RF beat-note data from yesterday, mixed down to 1 Hz, with linear frequency-drift removed (quadratic phase drift)
downsampled 10-fold three times to 500 S/s. Compare to 2023-01-20 beatnote-spectrum collected with Siglent SA (3 Hz RBW)


lasers locked for 16+ hours
lab-temperature oscillations slowly decreasing

slice-qtc ch3 and ch4 now measuring with better resolution

beat-note ca 20653350 Hz, drift -1070 mHz/s

13:17 try collecting 1000 S/s data?
1000 S/s (usrp source overflow errors) this causes bias in sdr-tracker frequency value?? (also in the 1S/s data into database)
try 100 S/s (back to usual beat-note value for 1S/s data)
10 S/s

raw RF data collected with 2 MS/s, quickly a large file of 500 Mb!
14:20 try sdr-tracker lower RF-bandwidth of 500 kS/s, a few minutes into 500 Mb file
(store files on aika-nas in 2023-01-28 folder)
leave tracker running with 500 kS/s bandwidth

16:30 modify new cavity power-servo, by optimizing AC-signal on in-loop photodetector (scope ch4/red)
old settings had ca 3 mVRMS AC, new settings ca 0.8 mVRMS AC
PI-corner 30 Hz, gain-setting 0 dB
16:40 leave lab

1000 S/s data 1000sps_1min
100 S/s data 100sps_1min
10 S/s data 10sps_1min
10 S/s data 10sps_10min longer 10 min dataset
100S/s data shows aliased spectrum, therefore ADEVs too high?
10 S/s data, 10 minutes. something oscillates with 3-5 s period?
NOTE: this glitch due to running sdr-tracker at 1000 S/s, it drops samples, and this leads to frequency-bias!?

16hour plot ca 13:00


lasers seem to have stayed locked over night.
adev now reaches maybe 4e-16 at 100s (some short-term noise, bump at 7s and 2s ? vibrations, fiber-noise?)

ca 09 drift is down to -1371 mHz/s
new cavity inner shield temperature slow -8 mK/day and -7 mK/day (fix low resolution of these readings!? slice-qtc settings?
new cavity ch2 in-loop sensor shows +2 mK/day with ca 2 mKpp oscilations from lab-temperature oscillations of 1000 mKpp

old cavity inner shield shows 0.9 mK/day slope and ca 1...1.5 mKpp due to lab-temperature oscillations

ca 11 drift -1312 mHz/s
ca 13 drift -1238 mHz/s
ca 15 drift -1183 mHz/s
18:24 drift -1061 mHz/s

ca16:15 adjust SLICE-QTC channel temperature ranges and set-points to be identical for all channels 1,2,3,4
previously ch3 and ch4 showed quite low temperature resolution? maybe measurement has lower resolution when value is far from set-point?

testing old cavity by pushing on it
17:25:30 H1-direction (red trace, long side of cavity box)
pushing every 30s, with plastic cable-tie as soft spring

17:27 H2-direction (yellow trace, short side of cavity box)

17:28:30 vertical direction (blue trace)
(glitches in data, without any pushing.. seismic activity from outside the building??)
17:30 try vertical pushing again

Previous measurements with PTB-seismometer (located close to 674nm input-fiber)

Connect PTB Trillium cable to the VTT seismometer (located centrally, on opposite side)
level bubble on top of seismometer
both end-doors now open (maybe quad-photodiode cables etc. were touching the end-door previously?)
17:44 H1-direction
17:45:30 H2-direction

17:48 vertical direction

18:15 close doors, leave lasers locked, old cavity power set-point now higher
PTB-box is logging seismometer data with 10 S/s.

turn off lab-lights, close door between small and big lab - does this improve temperature-oscillations?

old cavity power photodiode power servo locked y-day evening set-point ca 2.8 V


Not clear why SDR-tracker had unlocked and shows noisy data from last evening and during night
new cavity seems locked, old cavity also, re-tuning SDR shows 3e-15 stability at 1s

beat-note now ca 60 MHz LO + 8.476 MHz, drift is +2.89 Hz/s
slice-qtc temperature controller shows oscillations with ca 5 hour period, and 1..2 mK in-loop temperature error

old cavity transmission photodiode shows large variations with lab temperature (is doubler temp-control working well?)

13:36 drift now -2.52 Hz/s, beat-note at 80.893 MHz

ca 14 try collecting new cavity transmitted photodiode signal with DMM, simultaneously with beat-frequency
valid data around 14:10 or so

fiber-noise detector DC-level 9V - not much room for an AC signal on top?
ND-filter in front of detector increases AC-signal significantly! with ND-filter DC-level ca 3.6V and AC-signal > +20dB increase

try first version of power stabilization.
NOTE: drift-comp AOM RF PA runs very hot! improve heatsinking and/or check RF levels?
new cavity has Thorlabs adjustable gain photodiode on output PDA10CS2 - used with 60 dB gain to get ca 2 V reading

also power stabilization for old cavity turned on.
new cavity transmitted photodiode
-1000 Hz / V frequency shift!?

new cavity slice-qtc temp-controller
old cavity transmission photodiode


'spikes' in new-cavity beatnote spectrum, seem to come and go with many hours time-constant

now ca 1e-15 ADEV floor

Vescent temp-controller stabilizes ch1-sensor, but ch2-sensor shows large oscillations that correlate with tec1 current

11:30 turn off Vescent ch1/tec1 control

Configure Vescent to take feedback from CH2 and control with TEC1
Output-1 configured to output CH2 temperature with 1 V/C gain and -20 V offset
Input-A configured as external temperature input for CH1 with 1 V/C gain and +20V offset

ca 13:30 turn on servo with CH2 as sensor and TEC1 as actuator. Limit TEC current to 0.2 A
set Integral time-constant to 500 s, D-control is off, P-gain 2

NOTE: PTB using digital temperature controllers from
ch2 sensor now used in-loop:


Vescent temperature controller PID-loop was turned on
stabilizes in-loop sensor ch1 with ch1-TEC, but this causes large variations in ch2 temperature


09 lab visitors? turn off comb lock, M-NIR and beat-note counting

11:45 lock comb again

tested KSK105 (-3 floor) off for a while - but no dramatic improvement in beat-note sidebands

issues with 'spikes' from new-cavity seed laser? (visible with comb blocked, only laser on beat-note photodiode)
issues with fiber-noise unlocked (need monitor-scope next to fiber-laser setup, or some alarm when fiber-noise not locked!)


ca 11:24 manual re-tune of SDR to avoid 500 kHz glitch in data-series

beat-note residual follows lab-temperature, from new cavity and/or unstabilized fiber?
drift +7.5 Hz/s

17:35 drift +7.317 Hz/s
Note new cavity ch3, ch4 temperatures have now stabilized and follow lab-temperature, with positive slope for past days
19:55 drift +6.998 Hz/s

16h plot ca 11:27

1h plot ca 19

telegraf system diagnostics during slowdown: (maybe because windows live-plot from database was left on on Friday 20 Jan)


some influxDB timeout problem during night?
tracker doesn't find beat-note, so maybe laser unlocked?

linux machine seems overloaded, 100% cpu and 35/46G memory.. (uptime 171 days)
lots of plotting scripts in memory (each ca 500 Mb memory), kill them with "pkill -f" and similar commands -> back to normal CPU-load and memory
18:09 restart SDR-tracker and database-writing, beat-note now ca 10.6 MHz
initial drift now +6.88 Hz/s 
18:12 turn on old cavity power-servo, set-point now 2.8 V

18:20 connect old cavity power PD to DAQ-channel
ca 2.4e-15 at 1s - similar ADEV as before (for de-drifted 5 min data)


10:26 drift down to 7.69 Hz/s
ch3 slope -103 mK/day, ch4 slope -109 mK/day

similar stability of beat-note as before, now with new-cavity fiber stabilized
old cavity power-stabilization was on, but not working very well

modify PI-board for lower overall gain
now seems to work in PI-mode reasonably well

new cavity beat-note now around 70.6 MHz (previously was 78 MHz? so a different optical mode?)
16:51 beat-note drift -6.31 Hz (so beat-note has different sign compared to previously)
1s adev goes slightly below 3e-15 now
16:59 close old cavity end door (it was open), power stabilization on with set-point of 2.9V

newer data, ca 16:50, with improved fiber-servo (old cavity one door open)2

with old cavity doors closed:

similar data with www comb-plots:
Compare this to ion vs. old-cavity from clockruns
(but servo-gains and QPN could give bias to ion vs. laser data, at least for small tau)
18:09 try how beat-note looks like without fiber-noise stabilization on new cavity (has 10+5m of PM-fiber for 1348nm to comb?)
for this test AOM-drive is from Rigol siggen 75 MHz 680 mVpp, with 10MHz maser-reference
this gives similar 2.1e-15 at 1s stability as with fiber-noise servo??
18:14 back to fiber-noise servo

try to see 50Hz side-bands on beat-note with spectrum-analyzer
18:20 new cavity temperature ch4 -33 mK/day, ch3 -33 mK/day

18:23 try with old cavity power-stabilization off (set-point was 2.9V, shows ca 3.3V with servo off)
ca 2.3e-15 at 1s without power stabilization
(down to 2.1e-15 and after waiting)

data during night:


drift down to ca 21 Hz/s
SDR re-tuning now happens less often, so e.g. 4 hour ADEV plot is reasonable.
inner shield ch3 1-day temperature slope -227 mK/day
ch4 slope -233 mK/day
(old cavity -1.2 mK/day)
11:19 drift 19.2 Hz/s
17:14 drift 16.1 Hz/s

first try at fiber-noise cancellation on fiber from new cavity to comb


beatnote data ok only until 22 UTC or so yesterday?

10:40 however everything seems locked and ok, after tracer retuning, beat-ntoe measurements looks ok
beatnote ca 14.8 MHz now (60 MHz LO from artiq)

power through old cavity dropped during night (with doors closed)
with power-servo off, now cavity PD ca 1.8 V
after alignment up to maybe 2.3..2.4 V
drift-AOM power was increased, now back to >3V on cavity-output PD

beat-note drift down to 38 Hz/s

comb-box optics for 1348nm beat-detection modified for fiber-noise cancelling, with free-space splitter and retroreflector 

19:20 leave beat-note measurement running
Similar ADEV between cavities as before. Old cavity power-stabilization has no major effect. Antivibration platform of new cavity ON/OFF also no major effect.


Measure old cavity PD with DMM

new cavity thermal noise limit maybe 2e-16

2022-11-29 results with BOA-modulation give 1.8e-13 / Volt on cavity-output PD
-> need Cavity PD to be stable to 0.6 mV or better to reach 1e-16.

13:24 use old system for cavity power stabilization, with feedback to BOA amplifier modulation input.
Cavity PD set-point ca 3V and DMM reading 3.004 V

14:30 drift now ca 63 Hz/s

add some alu-foil to optics-breadboard on side of old cavity
this causes transmitted power to fall (maybe 2.6V) - some component heats up and causes misalignment?
after realignmnent up to 3.4 V

17:15 close cavity doors
18:00 leave measurement running, with closed cavity doors, closed plexi-covers for comb (small gap for fibers)
old cavity power-stabilization is on, with set-point of 2.9 V
DMM measures in-loop cavity output power.
2.9V setpoint for overnight run

With BOA power stabilization for old cavity

no stabilization


14:12 stop beat-note measurement from new cavity vs. comb

usual Sr+ mode found for old cavity.
Lock comb to old cavity.
Needed some re-alignment to get normal amount of power coupled into/out of cavity
measure beat-note of comb vs. new cavity
beat-note ca 70 MHz, mixed down with 60 MHz LO from artiq -> 10 MHz beat-note measured with tracker
drift ca +102 Hz/s now

19:44 start toptica T-loop script to keep Toptica ECDL locked for a long time

ca 4e-15 ADEV from 1s to 60s based on 1S/s data
but this data has some glitches/spikes also?

new cavity plots as of 09:41 on monday 2023-01-16:


11:25 ion still dark

16:20 comb rep-rate to 19 999 909 Hz
valid data from 16:22
lower beat-note 6.49 MHz, drift <200Hz/s
New cavity frequency and drift during cooldown


17:56 previous comb DDS setting 19 999 754 Hz
17:59 change to 19 999 817 Hz. lower beat-note now ca 6 MHz
new valid data from 18:02

08 ion dark
all beams, slider pos1, detuning scan - no recovery
405nm on -> ion recovers!

check uMotion 25518, 25519, 25520
went dark during C measurement, recovers with B-beam
maybe C-beam misaligned (tested LC-retarder yesterday)

15:30 local parabolic fit to beatnote
slider pos2 (OD1) B-beam uMotion 25k (8.5s), 50k (15s), 100k, 200k (59s), 400k gives timeout

slider pos1 (max) B-beam uMotion 25k (1.9s), 50k (2.9s), 100k(5s), 200k (8.7s), 400k (17.1s)

Detuning scans with pos1/max power and with pos2/OD1:
ca 16:15 ion went dark after scan with full power? Rb-lock checked and looks ok
B-beam P422 ca 4.5 uW maybe. blue-detuned high power kicks ion out of trap?
leave full cooling power and all beams?
17:50 still dark.. ion probably lost?


Note: sat.abs. Rb-signal DC-level shifts maybe 1/3rd of peak-height when tuning AOM. Since lock is always to 0V, this creates nonlinearity in the real cooling laser freq vs. AOM-frequency relation.
New Fiber-coupled Toptica TPA for 844nm amplification
Using old 2012 PCB+mount (was in use with old Thorlabs TPA)

Issues with 461nm laser, because 922nm optical isolator was not fixed to optical breadboard
5A from helper dispenser
laser adjusted  I 195.36 mA, T = 4.642 kOhm - quite weak fluorescence seen on webcam..

15:39 try to load with 1.75A dispenser current
dispenser ca 2.3W after warmup
13:52 ion! 
detuning AOM to 208 MHz
ca 300 counts/10ms seems to be ion limit, independent of laser power and PMT-alignment
A-beam 300-310 counts/10ms
B similar 280-300
C 240 initially, 
ASE align? with B-beam, no big improvements from alignment

set slider pos2, OD1 for 422nm
uMotion A 34%, B 47%, C 37%

start from old Vbias voltages  [-6.47, 0.2351, 22.5187, 63.2567]
RID25434 try Nelder-Mead uMotion minimizer, 30kphotons/run, max_iterations 10
final measured contrasts C = 0.0518 0.0835 0.0672 RMS = 0.1191
final voltages V = [-6.4700 0.2960 22.0866 65.6435 ]

verify with manual uMotion measurement
A 4.9%, B 7.8%, C 7.5%

detuning scan, 100ms gate, -20 MHz to +5 MHz from set-point 208 MHz
A RID 25439 (no drop in flurescence on blue side?)
B 25441
C 25443

repeat with -20 MHz to +10 MHz and 1000 ms gate
A 25445
B 25447
C 25449 (flat result??)
C repeat 25452

Search for secular freqs with coarse tickler method?
Tickler has 20 dB fixed attenuator
dds-attenuation 10 dB, PMT gate 100ms, 0.84 MHz to 0.9 MHz 600 points sweep 
gate 20ms, att 5 dB? search for X and Y modes?
remove 20 dB attenuator, use 15 dB software attenuation?
maybe one mode at 910 kHz?!
rid25490 upper at 958 kHz

search for axial/Z mode, found at ca 1.962 MHz

now install 20dB attenuator again, for photon-correlation method
rid25495 is z-secular with correlation method, 17 dB attenuation, 10kphotons/pt, 1.95 to 1.97 MHz in 50 pts. A-beam
- narrow peak 1.96266, A cools Z-mode badly, so narrow peak..
rid25496 is B-beam scan 1.96 to 1.966 MHz 30 pts.
rid25497 is B-beam 0.956 to 0.96 MHz 30 pts, 7 dB DDS attenuation
rid25498 is B-beam 0.908 to 0.912 MHz 30pts, 7 dB att

rid25500 Leave uMotion minimizer, 20 iterations N-M optimizer, 50kphotons/pt, slider pos1 (max coolinglaser)
C = 0.0103 0.0115 0.0206 RMS = 0.0257
V = [-6.4700 0.2335 21.9985 65.7307 ]
verify with manual measurements 100 kphotons
A 1.0%, B 1.23%, C 1.57%

slider pos2 for lower cooling and longer ion lifetime?

leave secular Z-scans with different levels of attenuation?
rid25506  28 dB, 1.952 MHz to 1.972 MHz, 90 pts
25507 26 dB
25508 24 dB
25509 22 dB
25510 20 dB
25511 18 dB
25512 16 dB

100ms gate

1s gate
upper radial coarse tickler 20ms PMT gate, no HW attenuation 15 dB sw attenuation
lower radial coarse tickler 20ms gate, 15 dB dds att, 0.905 to 0.915 MHz 600 pts
coarse tickler 20ms gate, 25 dB dds att, 1.9 to 2 MHz 600 pts

A-beam B-beam

N-M uMotion minimizer

Secular Z-mode with correlation method and increasing level of tickler?

09:42 upper beat note drifted down to ca 180.4 MHz
lower beat note drifted up to ca 69.7 MHz

increase comb rep-rate to 19 999 871 Hz
lower beat-note ca 7.5 MHz
upper beat-note ca 242 MHz
valid data with these settings from 09:53

ca 09:53
ch3 slope -6.7 C/day, mean +32.8 C
ch4 slope -6.8 C/day, mean +32.9 C

initial beat-note drift ca +798 Hz/s 

TF installed fiber-noise cancelling on new-cavity->comb fiber
beat-note optical frequency now -80MHz from previous value 

17:00 new rep-rate 19 999 754 Hz
lower beat-note at 23 MHz now has opposite sign to previously, it drifts down
17:20 drift now -657 Hz/s


VTT press release on ionclock (YLE puoli seitsemän y-day 2023-01-11)

Connect Vescent SLICE-QTC temperature controller with USB to linux computer.
Serial 9600-8-N-1 communications, script logs ch1,ch2,ch3,ch4 temperatuers to database once per minute.

Inner shield sensors ch3,ch4 at +36.3 C show initial slope of -8.8 C/day

16:50 start measuring ca 9 MHz beat-note of new cavity 
CH2 is 2nd beat-note at ca 215 MHz (mixer LO) + 26 MHz = 241 MHz
comb is locked to H-maser, with usual DDS setting 19 999 584 Hz

Comb RF lock PI settings were wrong, P gain too high
with reduced P gain beat-note sum ch1+ch2 agrees with rep-rate again quite well!
better data from ca 17:15
initial drift of ch1 seems to be +1200 Hz/s !?


Stop new cavity bakeout
New Gamma plots at end of DAQ page, from new cavity ion pump

17:04 bakeout Pt100 sensors off
17:13 disconnect BNCs from rack panel
Does this have effect on GPIB timeout/glitches of cavity temp-control system?


11:05 stop Vref measurement

Summary of Vbias measurements
NOTE EMM-measurement via photon-correlation finds E with unc. of 10...2 V/m - so voltage-source instability 0.2 V/m is small contribution

Electric field instability from voltage data and trap geometry:

From Tomberg2016 we find the circuits used.
Use circuit-simulator to find noise-sources? Laser-current-drivers like Troxel2011, Erickson2008, Seck2016, Libbrecht1993 should work for (unipolar?) coil-drive.

2019-11 V3,V4 gain was increased from 10x to 17.7x by 15 kOhm resistors in parallell with R10 and R24 (10 kOhm).
Old gain was 1+(100k/10k) = 11x. New gain 1+(100k/(10k || 15 k)) = 17.7x
Corner frequency of OC18 trap-filter is ca 15 Hz - no effect on the results above?

If Vref voltage multiplied by ideal gain 500 mA/10 V we get a similar PSD!?b
Magnetic field noise from current noise measurement:
I2 PSD ca 25 nA/sqrt(Hz)  / sqrt(f[Hz]) = 625 nA^2/Hz / f[Hz]
A2 = 0.4134 uT/mA 
-> B2 = 10.3 pT/sqrt(Hz) / sqrt(f[Hz]) 

From ion data 10.6 pT/sqrt(Hz) / sqrt(f[Hz])  + QPN 245 pT/sqrt(Hz)
fit to ion data between 0.6e-4 and 1e-3 Hz gives 81 pT^2/Hz -> 10.6 pT flicker


11:07 change to measuring V4, reads +86.185 V

16:45 change to Vref at back of Vbias box +10.645 V





10:47 change to measure V2, reads -6.27591 V

16:37 measure V3, -41.45 V



17:00 stop bartington measurement

change to measuring voltage driver V1 - measurement via T-connector on driver output, cable remains connected: V1-out -> filter -> trap.
reads ca -6.6646 V
artiq settings V12cm -6.47 V and V12diff 0.19434 -> 
V1_artiq =  -6.66434
V2_artiq =  -6.275659

V34cm    22.3670 V
V34diff  63.8193 V


Tapiola trains: 2.5min, 3.75min, 5 min, 7.5min
Kivenlahti trains: 5min, 7.5min, 10min, 15 min

Summary of Bartington measurements (sensor on optical table on ASE-side)
Keysight DMM measures Bartington voltage output at ca 2 S/s
peaks at 1e-3...1e-2 Hz are train rep-rate?

14:30local stop RF on/off loop, log-file entries are:

UTC 2022-12-25 08:51:58.807997 LOCAL 2022-12-25 10:51:58.807998 SET 10.0 REAL +1.0000000000000E+01
UTC 2022-12-25 08:58:04.703273 LOCAL 2022-12-25 10:58:04.703274 SET 10.0 REAL +1.0000000000000E+01
UTC 2022-12-25 22:58:05.051472 LOCAL 2022-12-26 00:58:05.051475 SET 10.0 REAL +1.0000000000000E+01
UTC 2022-12-26 12:58:05.191314 LOCAL 2022-12-26 14:58:05.191317 SET 10.0 REAL +1.0000000000000E+01
UTC 2022-12-27 02:58:05.351460 LOCAL 2022-12-27 04:58:05.351463 SET 10.0 REAL +1.0000000000000E+01
UTC 2022-12-27 16:58:05.487463 LOCAL 2022-12-27 18:58:05.487467 SET 10.0 REAL +1.0000000000000E+01
UTC 2022-12-28 06:58:05.631455 LOCAL 2022-12-28 08:58:05.631458 SET 10.0 REAL +1.0000000000000E+01
UTC 2022-12-28 20:58:05.775460 LOCAL 2022-12-28 22:58:05.775463 SET 10.0 REAL +1.0000000000000E+01
UTC 2022-12-29 10:58:05.959463 LOCAL 2022-12-29 12:58:05.959465 SET 10.0 REAL +1.0000000000000E+01
UTC 2022-12-30 00:58:06.099461 LOCAL 2022-12-30 02:58:06.099464 SET 10.0 REAL +1.0000000000000E+01
UTC 2022-12-30 14:58:06.243470 LOCAL 2022-12-30 16:58:06.243475 SET 10.0 REAL +1.0000000000000E+01
UTC 2022-12-31 04:58:06.407464 LOCAL 2022-12-31 06:58:06.407466 SET 10.0 REAL +1.0000000000000E+01
UTC 2022-12-31 18:58:06.583469 LOCAL 2022-12-31 20:58:06.583473 SET 10.0 REAL +1.0000000000000E+01
UTC 2023-01-01 08:58:06.759485 LOCAL 2023-01-01 10:58:06.759489 SET 10.0 REAL +1.0000000000000E+01
UTC 2023-01-01 22:58:06.902481 LOCAL 2023-01-02 00:58:06.902484 SET 10.0 REAL +1.0000000000000E+01
UTC 2023-01-02 12:58:07.243466 LOCAL 2023-01-02 14:58:07.243469 SET 10.0 REAL +1.0000000000000E+01
UTC 2023-01-03 02:58:07.466522 LOCAL 2023-01-03 04:58:07.466525 SET 10.0 REAL +1.0000000000000E+01


10:47local adjust variacs, +2 DIVs on each
V2 to 44V
V4 to 56V
V6 to 56V
10:50local picture of Vescent thermistors
ch1 40.0915 C
ch2 39.7695
ch3 39.5827
ch4 39.5763
GP reads 3.3e-7 mBar

ch1 44.3032 C
ch2 41.4037 
ch3 39.9249
ch4 39.8546
GP 3.8e-7

15:04 Change B-field measurement to Y-axis, along long edge of optical table.
(no change in datalogging, goes to same file as previous X-measurement)
reads ca +24 mV = 2.4 uT now



12:25local stop Z-direction B-meter

turn B-sensor to measure along short edge of optical table, X-direction
12:32 restart measurement, reads ca +1.2 mV



10:58 local: Start 14-hour trap RF cycling (while True)


Stop coil-current and reference-voltage logging.

10:30local Continue with B-field measurement, now in Z-direction



try to fix cavity temp-control and logging script (gpib timeout problem)
run every 6 minutes from crontab now (was every 5 minutes)

16:20local close cavity doors (but many temporary cables prevent closing doors properly)

For 2022 Fall results see index_2022fall

For 2022 Spring results see index_2022spring

For 2021 results see index_2021

For 2020 results see index_2020

For 2019 results see index_2019