Homestake Meeting Minutes, 04/17/2014
Attending: Eric, Gwynne, Riccardo, Vuk, Victor, Tanner, Shivaraj, Michael, Nelson
Homestake wiki: https://zzz.physics.umn.edu/groups/homestake/home
1) UMN test setup
Tanner: Working on the GPS timing system with Liteway. Master digitizer produces digital GPS signal, converted to optical, then converted back into digital, and sent to the second digitizer in the slave mode. Everything seems to be working fine, no loss in SNR. Started running the system on a 12V battery: a digitizer, a baler, a seismometer, and a receiving module for the GPS (two units at the moment, but hope to reduce this to one in the final design). Currently testing how long we could run this system on the 12V battery (55 A-hours) - has been running for two days and still fine (expecting 6W consumption).
Vuk: currently using sealed-gel style of the battery, which is expensive, but also safe: as Gary told us, regular car batteries emit hydrogen, so in enclosed spaces could explode. However, in the mine these batteries will not be in enclosed spaces, so it may be acceptable to use deep-cycle batteries, but not sealed-gel. Will follow up with Gary on this.
Riccardo: suggest to record the voltage across the battery at mV level as a function of time, which will be useful diagnostic for later on, when we operate in the mine.
Shivaraj: did you verify accuracy of the timing system using the data?
Tanner: Have not checked the data yet.
Vuk: should think more about this, not sure that we have the information in the data files to do this (in the way Shivaraj did in the past). Also, getting the data from the baler is now a priority, since we have to return the demo in 3 weeks or so.
1) Homestake update (Vuk)
Vuk: based on the success in Tanner's test setup, prepared a plan for the GPS distribution, power supply, and network access at Homestake, at various stations. Draft is posted at: https://zzz.physics.umn.edu/_media/groups/homestake/array/gpsdistributionsystem_plan.pdf
Briefly, there would be two GPS distribution systems, one through each of the two shafts, in order to access all levels and minimize the number of split-offs. AC power should be available in all stations at 300, 800, 2000 levels, most stations at 4100 level (certainly 3, maybe 4th as well), and at least some of the stations at 4850 level that are close to Ross or Yates. The rest of the stations will have to be battery operated. Regarding network, in AC powered stations there will be a network switch, with a single fiber connection to the outside. In battery powered stations propose to use 3 separate fibers for 3 ethernet connections at each station, to minimize power consumption.
Shivaraj: how much power does the ethernet switch take?
Tanner: 12V and 1A, so 12W - this is a lot compared to the rest of the requirements.
Shivaraj: How about the alternative power?
Vuk: planning to directly convert optical to digital on each fiber, expect it to be small?
Shivaraj: need power to convert optical to digital ethernet signals even at AC powered stations.
Vuk: will have to look into this.
Riccardo: maybe worth thinking about stringing a small power cable along with the fiber, maybe this could be a good and not too expensive option compared to the batteries. Should ask what is acceptable in the mine - what kind of cable and what kind of maximum voltage they could carry.
Vuk: will try to look into it.
Riccardo: regarding the batteries: if you put many batteries in parallel, protect them with a diode each, so that if one shorts the system is not affected, shorts are more likely to happen if the battery is discharged to zero. If we will cycle them only very few times only, we may go for not deep discharge batteries and go for maximum amp/hour per dollar.
3) Analysis of existing data
a) Frames/miniSEED files (Shivaraj, Tanner, Michael)
Shivaraj: almost all frames are copied to the Caltech cluster now, should be done today:
b) Estimation algorithms (Shivaraj, Eric, Noah)
Eric: Looking in more detail at the turning depth: http://www.ligo.caltech.edu/~ethrane/stoch/turning_depth2.pdf Model the rock as 100 thin wafers, with refraction at each boundary. Snell's law applied at each discontinuity, using different wave speeds according to v ~ z^alpha, where alpha=0.27 comes from a different site (expect it to be smaller at Homestake). Find that waves turn around if the incident angle is larger than ~55 degrees (wrt the z-axis). For any alpha, there is a range of input angles that leads to turn over.
Vuk: would like to get better intuition about this - naively, expect small features not to affect much the waves with long wavelengths. In other words, would you expect the turning depth to depend on the frequency/wavelength of the wave?
Riccardo/Eric: this is similar to waveguides. The “feature” is small in one dimension, but goes across another dimension, so diffraction intuition is not applicable…
Vuk/Eric: observationaly, from geophysics is it known whether turning depth depends on frequency?
Victor: skin depth (associated with Rayleigh waves) does depend on the frequency (larger for larger wavelength), but not sure how this is related to the refraction argument here. Will try to follow up on it more.
Nelson: may have to go back to the Fermat's principle and solve the problem from scratch.
Vuk/Eric: it would help to learn what is known about this in geophysics, at least to get a sense of what we need to do. Victor will look into it.
Vuk: on the other hand, in the radiometer code, this should boil down to changes in phase, which should be calculable for a given model. So we should be able to correct it?
Eric: yes, we would have to add angle and depth dependent corrections to the phase.
Vuk: will contact Jaret to get more information about the wave speed at Homestake.