Funky Gravity Data Micro-g LaCoste Derek van Westrum Or How I Learned to Stop Worrying and Love the Bomb...
Funky GravityData
Micro-g LaCoste
Derek van Westrum
Or How I Learned to Stop Worrying and Love the
Bomb...
Outline
Good Data: Everybody wants some... How to “read” data quality
Drop & Set Scatter Residuals Drop Fit Sensitivity or “Truncation Tests”
Common Hardware Setup Problems Common Software Setup Problems Hardware Failure Symptoms
How to “Read” Data Quality•Drop Scatter should be as small as possible. In a lab, <20uGal for an FG5, <50uGal for an A10•The value depends on the site, but it should at least show no overall trends up or down
(these data are really good, by the way...)
How to “Read” Data Quality cont...•Set scatter should also be as small as possible, with no overall trends. •Typical values should be less than 5uGals for an FG5 and <10uGals for an A10 or FGL•For an ML1 laser, the reds and blues should be self-consistent
How to “Read” Data Quality cont...Residuals are the difference between the raw data (time and distance) and the best least-squares fit:
MeasurementsBest Fit
Residuals Note: vertical scale exaggerated, normal residuals are approximately 1nm.
0 2 4 6 8 100
20
40
60
80
100100
0
t2runif 1 4 4( )0 100
t2 100
t2 t
2runif 4 4 1( )0 50
100 t
How to “Read” Data Quality cont...Here’s an example of good residuals. The single drop RMS should be small (<1 nm) and flat – no low frequency noise...
How to “Read” Data Quality cont...Fit Sensitivity or “Truncation” Tests.
•Plot the sensitivity of the gravity value as a function of the fit times (at the top of the drop and the bottom)•The measured gravity value should not depend heavily on what fit times you choose!
How to “Read” Data Quality cont...Fit Sensitivity or “Truncation” Tests Cont…
Common Hardware Setup Problems Common Simple Setup Mistakes Superspring Issues
Ringdown (measuring too early) Coupled System
Interferometer Small Fringes Rubidium Clock Unlocked
Dropper Issues Hitting the side/top/bottom Poor vacuum
Laser Issues Laser Unlocked (ML-1) Wrong/Missing Modulation Frequency (WEO) Laser “Peak Hopping” (WEO)
Software Issues PC Clock wrong
Common Simple Setup MistakesOr How to Avoid the Dreaded & Cryptic “Time Out - Check Trigger, Fringe Amplitude, or TTL Cable” Message…
This message simply means that the software did not acquire the expected number of fringes. Period. It might have received zero or 99%, but it needs all of them. It will wait about 20 seconds and then “time out”…
Common Causes: Bad alignment No trigger cable Alcohol pool in interferometer Beam blocker engaged Dropper in reset mode Dust cap on somewhere …
Superspring Issues
Patience! Often large scatter is simply due to taking data too early. Get a cup of coffee, and start the measurement 5 minutes later!
“Ring down” – movements of the spring as it settles down look like changes in gravity…
Superspring Issues Cont…
Check the residuals – they might tell you something…
Is the system still coupled?
Superspring Issues Cont…
But definitely check the fit sensitivities – they will tell you for sure!
Is the system still coupled?
Superspring Issues Cont…And yes, earthquakes anywhere will show up in the drop scatter! With luck they’re far away, and you can just wait it out…
Interferometer
•If the alcohol spot is 1 diameter off, that is a negative 4uGal error•2 diameters is -16uGals!•The error goes as the angle squared, so things can quickly go wrong…
•Note – a non-vertical beam can never make the value too high!
First of all – your beam has to be perfectly vertical!
Ryan Albery checking verticality in Prudhoe Bay, Alaska, 2003.
Interferometer
•Note, at “normal” scale, the residuals don’t tell you too much…
Small Fringes. If the amplitude is too small, the software will “miss” a fringe, and this throws the timing (and thus the gravity value) off considerably…
Interferometer
On auto-scale, you can see the “integer” step, where the system misses one fringe, throwing things off by 316 nm.
Make sure your fringes are maximized:
300-400 mV for FG52-4 V for A10/FGL
Small Fringes. If the amplitude is too small, the software will “miss” a fringe, and this throws the timing (and thus the gravity value) off considerably…
InterferometerSometimes getting the fringe amplitude increased is easier than others…
Getting a new dropper aligned to the interferometer in “clean room” conditions in Prudhoe Bay, Alaska, 2005.
Interferometer
The clock being disconnected (bad cable) or unlocked is usually obvious. This is really the only thing that can make your gravity value off by milliGals!
Rubidium Clock Unlocked
Dropper
•Again, at “normal” scale, the residuals don’t tell you too much•On auto-scale, you can see the free fall parabola that comes to an abrupt stop when object hits the cart•Check dropper bubble levels are okay, and that dropper tuning (the “hover”) is okay.
Test mass hitting the side, top, or bottom of cart…
Dropper Cont…
•If the dropper controller gets an “extra” trigger, the software can fit the “gravity” value during the lift.•The value will typically be something like 300 or 400 Gals! •You will notice this right away - the residuals are also very bad.•If it is still running you might be able to manually provide another trigger to get back “in sync” (two wrongs can make a right…)
Dropper “out of sync” and gravity fit during lift
DropperAnd sometimes you’ll need a little more effort to get the dropper happy…
Fred & Ryan tuning droppers at an intercomparison while Derek wastes time taking photos. Luxembourg, 2003.
Dropper•When the vacuum is bad, the drag of the molecules on the test mass causes this characteristic residual shape. •After a couple minutes of this, the gravity value will plummet even milliGals•Check your ion pump voltage and current! And keep the turbo pump close by…
Poor vacuum
Laser (ML-1 for A10 & FGL)
•“Drifts” of 100s uGals are usually due to a laser that is unlocked•Make sure the laser is in REMOTE mode and that the software is set to “Alternate Red/Blue” modes•Note that the laser can “only” drift 1.4 milliGals. If your problem is bigger than that, it is probably NOT the laser!
Unlocked
Laser (ML-1 for A10 & FGL) Cont…
•If the Red & Blue wavelengths are swapped in the software, this is a typical sight•If the laser is locked in MANUAL mode and the software is set to Alternate, then only half of these sets will be measuring the correct gravity value – the others are off by 1.4 milliGals!
Locked on Wrong Peak
Laser (WEO 100)•If the Modulation Frequency is wrong or disabled, the residuals will look like this.•Note though that after 100 or so drops, the average residual is almost “normal” and that the mean gravity value is not actually greatly affected.•Best to have the right value in there though…
Wrong Modulation Frequency
Laser (WEO 100) Cont…
•Sometimes the laser will hop to an adjacent Iodine peak.•If the software uses the wrong values, the gravity values will appear to jump by as much as ~25 uGals. Possible causes:
•Wrong 1f values entered•Values changed as laser warmed up for an hour or so•Cable from laser to software not connected•Wrong setup parameters in software
Laser Peak “Hops”Cont…
Laser (WEO 100) Cont…
•The 1f voltages are measured to keep track of the peaks – the voltages and peaks are unique•The software monitors the 1f voltages, and knows which wavelengths to use in the fit – the resulting gravity values are consistent
Laser Peak “Hops”
Software Setup Issues•The PC Clock must be set to GMT (no daylight savings) at all times•If the clock is off by 1 minute, the tide corrections can be off by up to 1uGal•Here the clock is off by 5 minutes
PC Clock set wrong
•And here the clock is off by 1 hour
•Note that any Tide/Ocean Load setup problem can look like this: clock, location, using the wrong Ocean Load files, etc.•10-100uGal, sinusoidal set signals are always a Tide/Ocean Load problem…
Software Setup Issues Cont…Wrong Ocean Load Files
•In versions of g before v6, the default ocean load filename was always “OceanLoad.olf”•If you run with the OceanLoad file from another location, your set scatter will be in the 10s or 100s of uGals•Note – In g v6, the software suggests a unique string based on the location name.
Software Setup Issues Cont…And sometimes the PC itself is just wrong. Get some help!
Trying to find a dropped screw after reseating the Time Interval Card. Madrid, Spain, 2001.
Hardware Failure Symptom Summary•0-10 uGal
•Check verticality, setup heights & parameters•Check all below…
•10-100 uGal•Check Tide/Ocean Load Setup•Check Vacuum•Check if Dropper Coupled•Check all below…
•100-1000uGal•Check Laser Locking•Check all below…
•1000uGal and beyond…•Check Rubidium Clock
Conclusions…•If something seems weird, it’s probably best to assume it is. Stop taking data and figure it out!•Check the scale of the problem and eliminate impossible causes…•Always check & double-check drop scatter, set scatter, residuals, and fit sensitivities•Take your time, don’t panic, and you’ll get it!
Take Good Data & Go Home with a Smile…