Milagro Collaboration Meeting 2/26/98
Index
Thursday Morning
Gus Sinnis - Overview of Installation Plan
Tony Shoup - Cable & PMT Status
Linda Kelley - Electronics Update
Don Coyne - Cover Modifications
Peter Nemethy - Pointing / Surveys / Strings / Pedestals
Jim Ryan - Igloo Status
Andy Smith - Computing for GEANT
Thursday Afternoon
Richard Miller Online DAQ
Richard Miller Calibration
Neil Thompson - Cable Pull Box
Jordan Goodman - Budget Considerations 
David Williams - muon layer versus an expanded hadron layer
Stefan Westerhoff - Monte Carlo simulations of muon and hadron layer
Rich Miller - more on hadron versus muon layer
Jim Ryan - effect on solar physics of not having the muon layer
Todd Haines - concern about changing things at this late date
Wystan Benbow - Baffles
Dave Williams - Motivation for Baffles
Cy Hoffman - NIM-type Article in Astroparticle Physics - or in NIM?
 
 

Friday Morning

Richard Miller - Milagrito Operations Summary
Todd Haines - Water System
Michael Schneider - Water System
Andy Smith - Calibration

Friday Afternoon

Peter Nemethy - Pedestals
Tony Shoup - Moon Shadow from Milagrito
Morgan Wascko - Moon Shadow Simulations
Andy Smith - Markarian 501
Gus Sinnis - Markarian 501
Jim Ryan - Solar Physics
Jordan Goodman - Data Analysis Plans and Discussion
Baffles and Igloo Decisions 
 

Gus Sinnis - Overview of Installation Plan

PMTs -

weíre a little behind, but things will go faster

last PMT will arrive at the site in mid August

Cover Work

coming along

Igloos

decision has to be made now

procurement took a little longer, but still on time

looks like a 3 week gap in July while we wait for the igloos to arrive

Electronics

hv and tdcs not ordered yet, no money

Water System

reconfiguring of Pub still necessary

Site Preparation

canít drain the pond until the snow is gone and the ground is dry - so 4/15 may be too optimistic for this

Installation plans

about complete

Important dates - list leaves time for water fill

start 4/15 draining

end 10/15 water fill

application to state to dump the water - written -

application to dump backflush from media and charcoal filters - written -

what to do with the softener backflush? into 1MGal pond? 50kGal? 3-4% salt -

we need to pump water 24 hrs/day to fill by the cut-off time

Return to index
Tony Shoup - Cable & PMT Status

Short cables (pub to counting house)

use 100' cable (regular RG-59), two labels on each end

tested - pull, continuity, isolation

have 790/820 cables done (supplying 10 spare) already

Long cables (pmt to pub)

not done at UCI, but at TMA (company in ABQ) with complete testing

Fischer water-tight connectors

450' length, 500 being done now, 50 more later, takes 30-45 days

pressure and hv test vessel built by UCI, coming to LANL

we will do the testing here - are there safety issues?

Inventory

250 at Fenton Hill, 385'

45 at UCI, 385'

485 at TMA, 450'

65 at TMA, 385'

==================

845 total

PMTs

Inventory

260 at Fenton Hill -these need new bases!

345 at UCI now -60 of these need new bases

273 on order, to arrive UCI mid-March to July

~2 at UCSC

=================

880

Assembly

20/day solder bases

50/week acceptance test - this is limiting factor, because of facility

60/week encapsulate & test

black disk - to prevent light leaking out - can be done easily, except that these arenít sufficiently light-tight for igloos -

 Return to index

Linda Kelley - Electronics Update

ECL boards all assembled

Analog boards - surface mount all done

everything to be completed by end of April - a little late, but not too bad

crate backplanes all done

purchased 1 pair of power supplies for testing, money exists for the rest

fan trays and support hardware are under construction

We will have 61 ECL boards, and 63 Analog boards

total of 976 channels available for the pond ( = 61 * 16 )

We will need to do some work at the site this summer, re-working existing boards

and will need storage at the site for power supplies, crates, and boards, bringing stuff to site by the end of April

 Return to index

Don Coyne - Cover Modifications

2 large and 4 small winches in hand, rope is at site (T-900, 2400')

reworked engineering design done by DC and Red Mountain Engrs

P&M will do for about $16k

Don will survey the pond layout for the locations of the new cover piers next week (3/2-4)

Hope to dig and pour concrete in March and April, mount and test in May (concrete needs 30 days to cure for applying pressure)

Catenary -

big tabs installed, and web also (?)

ropes are being ordered

winches are in hand

coupling hardware still being investigated - what do you connect to what? - and do you make it useful in the inflated mode as well as in the catenary?

winch mount - taking a lot of work

engineering check in progress

2 of them fabbed, to be tested at site next week

mass produce at UCSC in March, and install all components when snow is gone

Return to index

Peter Nemethy - Pointing / Surveys / Strings / Pedestals

superb time resolution on tdcís

we need to point correctly to within 1/10 degree - need direction of vertical especially -

check will be done on Monday by Galen & Ken to see if day-time and night-time surveys agree

suggestion of survey inside the pond by Cy - use with cover down - see grid

suggestion of using water level by Peter - but is probably too late in the process -

so check using water level on the way down!

need full survey of grid points and then manufacture the strings accordingly -

clips for attaching pmts to the strings - quicklink hard to undo under water - cĖlinks (split rings) virtually impossible to undo when under tension - carabiners also considered, but hard to keep the orientation in one-handed operation - but a variation on a carabiner, with a wire close - this seems to work, and will be ordered now

the string is woven kevlar - parallel kevlar was considered, but is difficult to work with, and the woven stuff works fine

knots unravel when not under force, use fishermanís glue to fix

use a float, that both the string and the coax cable get threaded through- $3 each (1 lb buoyancy, which is twice what we need to hold the string and cable) from a company that makes floats for fishing nets - but not known if the Fischer connector fits through the holes in the float

there will be enough string for the laser balls as well - ordered plenty -

pedestals -

Return to index

Jim Ryan - Igloo Status

Things look pretty good -

2 soft-igloo water tests, refined the design, need full-blown prototype test - including plumbing, light leaks, ballasting/buoyancy etc

soft igloo has 2"od pvc frame which sits inside the Milagro grid (and is attached by kevlar strings?), igloo attaches with long tabs - folded and collapsed, can be carried by 3 people - frame pvc has holes in it so it will fill with water - used styrofoam to provide bouyance in the frame

findings: stitching works; independent upper and lower frames, with the lower one secured to the grid, and the upper frame has (dark, lobster-trap) floats for buoyancy;

but - the PMT is far too buoyant, and stresses (almost tears) the fabric - you want it to be the high point of the igloo, but not too much - so may need some ballast at the PMT

no need for significant overpressure, can live with 1 gph under the floor

sleeves work better than tabs for attaching to frame

delivery in kits - need students for kit building, $600 per igloo looks ok, prototyping uncertain - need time and $$

building two per day is relatively easy, three per day may be possible with shift work

could do a trapezoidal profile, with the top bigger than the bottom so you donít lose as much inactive area (Gaurang estimates 15% inactive) - but this has other disadvantages, like loss of access to the grid itself

 Return to index

Andy Smith - Computing for GEANT

PC with 300 MHz PII ($2.5k) versus SGI O200 ($13k)

PC rates a 4.0, O200 rates a 3.2

Data rates - use MC to scale Grito to Milagro, just 500GeV to 10TeV -

expect VME rate to be 4.6Mb/s at 100PMT triggers, 6.8Mb/s at 50PMT trigger - event size in Milagro is much bigger, so data rate goes up a factor of 10 or so

Archiving at 3Mb/s is 8 DLT tapes/day ($230k / year) == canít write raw data!

Solution is store processed data only on tape - 1 year proc is 50 DLT tapes - but put eah separate data product on a separate tape -

Current recon code does 800 Hz on 1 R10000 processor - with higher data rates, will need at least 4 more processors for kahuna -

General conclusions - we need lots of disks, 60 tape DLT library, fast calibration, archiving software -

Muon discrimination via multiplicity trigger - ? - or an outrigger trigger - ? -

Return to index

Lunch

 

Richard Miller Online DAQ

baseline daq complete; most runtime errors have been identified and solved

may need to increase the dma & fifo sizes for Milagro

& increase the number of processors (to 6?)

change the archiving scheme -

improve the error reporting to give the shift person more information

try doing real-time transient detection system -

dynamic multiplicity trigger -

Return to index

Richard Miller Calibration hardware

have 4 km of fiber at LANL, can do 25 laserballs (which is sufficient if no baffles; with baffles, need 50)

need to prepare the fibers, make the laserballs & floats, and test - 2 weeks

for installation, recommend a PVC conduit from the pub to the pond bottom and lashed to the grid

will do fiber installation prior to pmt installation, and allow 3 days, but should take only one day

should we instrument 10% of the igloos for calibration?

 
Return to index

Neil Thompson - Cable Pull Box

door under counting house has been made, cleaning up underneath

in the pub, working on patch panels for connecting short cables to long cables

two of them, parallel to each other, a foot wide and separated by 4'

cables into conduit go from inside of patch panels, strain relieved on ladder structure that bridges the two panels

must have to move some "furniture" in the pub in order to implement the new patch panel design -

Return to index
 
Jordan Goodman - Budget Considerations

computing - cpu's - can buy 8 more for 80k, but no money in this yearís budget

we do have 50k for a tape library and 36k for analysis computers & disks

muon boxes - we have 85k on hand, spent 36k on material, have 49k left

shower array - 20% of funds in hand - for outrigger tanks, filtration, deployment, fence, cables, etc

miscellaneous - baffles 15, vme crate 2.5k,

deferred - needed in early fy99:

80k cpu's, 24k analysis machines, 97k shower tanks, contingencies 15k, subtotal is 216k - we may need to write a separate proposal for this stuff - but what should we time this for?

if we want to have the cpu boards now, we have to rethink our priorities for other things - e.g. the muon layer, the shower array, etc

Return to index

David Williams - muon layer versus an expanded hadron layer

if we bring the hadron layer up to 286 tubes, we might be able to detect muons as well or better than the muon layer with a larger collection area, but perhaps a lower efficiency - and we would have extra tubes (~60) left over for outriggers

the last row around the outside is 1.5m above the rest of the pmts

 Return to index

Stefan Westerhoff - Monte Carlo simulations of muon and hadron layer

late light - indicates attenuation length greater than 5m in the water?

pe distribution in the igloos - agreement with Milagrosim with 10m absorption -

can do gamma-hadron separation by looking at ratio of sum pes in the hadron versus air-shower layer - but doesnít work for events that have no muons that hit the pond

proton showers have isolated hot tubes (due to muons that penetrate), while in gamma showers if one tube is hot its neighbors are too

define a hot tube as one with > 30 PEs -

an extended hadron layer is almost twice as effective at detecting muons as the igloo layer is

can use neural network on a 3x3 input tube grid to train on muon-like (isolated hot tube) versus core-like events - works pretty well -

can count muons better as well -

per unit area, the hadron layer is ~90% as effective as the muon layer at detecting muons

Return to index
 
Rich Miller - more on hadron versus muon layer

can use additional tubes in the hadron layer to help the timing that the air shower layer accomplishes

Return to index

Jim Ryan - effect on solar physics of not having the muon layer

background versus effective area issues - if we buy area, do we lose signal-to-noise

Return to index

Todd Haines - concern about changing things at this late date

when muon rings start overlapping, simple algorithms donít work any longer - and this is a concern for showers that produce more than 1 muon per hundred square meters - at > 10 TeV

so weíll lose at composition studies at high energies

 Return to index
 

Wystan Benbow - Baffles

"dog cone" - don't allow horizontal or up-going light to hit the tube

to improve timing - a robust hardware solution -

pond test - 4 baffles installed last August and September - 2 white, 2 mirrored

baffled tubes get about 60% overall pes compared to un-baffled tubes -

light that is cut out is also light that is not used by the angle fitter - late light - as found by doing the t-chi cut

enables tightening up of timing on single tubes, especially at low pe levels, and presumably would improve the angular resolution for Milagro

can also lower the trigger threshold, since baffled tubes are hit only 1/3 of the time when the trigger threshold is down to 25 (they were hit 2/3 of the time with the normal trigger) - also in events that failed the fitter, the baffled tubes are hit only 2/5 of the time

Return to index
 

Dave Williams - Motivation for Baffles

eliminate data with little or no value - to minimize unnecessary deadtime, etc

simulations to optimize baffle geometry -

assume 88% reflectivity (either specular or diffuse) - the materials used generally degrade significantly with time in water

optimum baffle design has radius 12.5 cm (compared to tube 10 cm) and height 7.0 cm (compared to tube 7 cm)

efficiency improved most for particles near zenith

laser balls - with 24 laserballs, have every tube within 10m of at least one laserball, at 81 degrees -

 

Milagro Collaboration Meeting 2/27/98

Return to index
 

Richard Miller - Milagrito Operations Summary

7.264 billion events as of 2/25

386 DLT tapes

80% up time - but problems remain -

Operations Lessons Learned

MG Set & High Voltage - how to do reset remotely - is this wise?

DAQ - DC-2 mounting problem - foam rubber support! needs fixing

Web based experiment monitoring -

try to improve access to historical data, not just current run

Shift - define responsibilities better & provide more error notification

Remote site needs to be maintained as a working environment

DAQ is not currently reconstructing

worker process is crashing due to array bounds problem, also fifo is overflowing

Return to index
 

Todd Haines - Water System

System was designed for maximum flexibility - every component can be bypassed

Separate components for getting rid of large particles, organics, minerals, fine particles, and biological organisms

While "making" water from supply, use media filter and softener, but when recirculating pond water, these two are bypassed

Weíre now at 2m of water - ran at 1m until 11/96, filled to 1.5 by 12/22, and to 2m by 2/4 - this is final Milagrito depth -

Recirculation at 200 gpm since 1/97, turned over 110,000,000 gallons total - >50 times turned over

Problems -

1/97 pump motors were "dying" due to incorrect installation - fixed, got 2 new motors for insurance

3/97 pump cavitating, caused a leak - solved by lowering the pump

3/97 cartridge tanks found to have bad metal, sent back for repair - 2 of them are not back yet

11/97 discovered that salt had been introduced into the pond early on - like 2000 ponds

backflushing of water system into 1.5 MGal pond was disallowed for a while, now apparently okay

Operating modes -

have operated with different combinations of filters through the year; the softener is bypassed now, the carbon filter will be bypassed soon - like now - as a test to see if organics are building up

Start with water from well, with ~500 ppm total dissolved solids -

but samples taken in 11/97 had 950 ppm tds - derived from conductance - and 300 ppm of this turns out to be salt -

sample on 12/12 showed 770 ppm tds - going down, partly owing to the addition of new well water - but this came down further than expected!

Todd measured a sample on 2/6/98 and found 500 ppm - which is still lower than expected from the previous results -

Reverse osmosis system would give you 20 - 30 ppm tds - perhaps we should consider?

Attenuation length results -

in 5/97 measured ~ 4m attenuation lengths - then in 6/97, after changing to better filters, got ~8m - but in 10/97, back to ~4m - in 1/98 ~6m

results are difficult to understand

why does the measurement change so much?

what is our real attenuation length??

are samples contaminated between LANL and Santa Cruz

Trigger rate

roughly constant 2/97 to 11/97 at 275 Hz, increased to 325 Hz during December fill, increased again to 400 Hz during final (Jan-Feb) fill - until yesterday, when the threshold was cranked up to get the daq system back

Scaler rates

low threshold: light leaks obvious, especially starting in October - then rates go up with the fill events again - apart from those considerations, this remains mainly constant, currently at 6 MHz

high threshold has other structure, more difficult to interpret

low threshold shows about 20 hits per particle - and high is 2 per particle -

Results from triggers and scalers suggest the water is not bad, and not getting worse -

But it really pays to have really good water - twice as much light is like doubling the number of pmts

How to improve -

ultrafiltration ~$40k

reverse osmosis ~$50k -- downside is 30% waste on fill, 5% on recirc --

deionization ~$500k (full system) -- downside is environmental issues on regenerating the deionizing medium

can do r.o. + de-i. in combination for $130k ($50k + $80k incremental de-i system, not full) - but still probably too expensive

Propose

invest in r.o. for the fill

use of a "floating floor" to eliminate light scattered from the air/water interface

Return to index
 

Michael Schneider - Water System

Essentially no change in the pond water over the course of one year of operation - sometimes looks better, but may be systematic errors

We really havenít changed the system at all - except for brief & inconsistent tests

Weíre running the charcoal filter at too high a pressure

We should use the right filters (more expensive) and run things at appropriate pressure, we could do much better than we are

Return to index
 

Andy Smith - Calibration

problems with tot-to-pe conversion - of long standing

bridging the transition region between low threshold measurement and high threshold method is difficult

systematic effects of gain across pond could produce offsets of 0.2 degrees in ra and dec in source positions due to calibration problems

Return to index
 

Peter Nemethy - Pedestals

is the speed of light a universal constant?

from laserballs, measured 6/97 11.06 cm/cnt; 12/97 11.02 cm/cnt - not consistent

temperature differences in index of refraction? possible - but tds may also affect n

difference does not presently affect the calibration of the time pedestals -

good enough

do laserballs swim?

some are good swimmers, some like to "go home" -

at least two lines seem to be cut or free (untied?)

but others (three that had drifted before) seem to be closer to their nominal positions

do optical fibers shrink?

if you cut out the outliers, data on fiber delay is consistent with no changes

except for laserball 0, which seems to have shrunk by 1.2 counts -

is water temperature different across the pond?

many of the outliers are associated with a single laserball

are the pedestals consistent or not?

pedestals from June and from December are consistent with each other to within about half a count for most tubes

there are 5 outlier tubes in this distribution

Return to index
 

Tony Shoup - Moon Shadow from Milagrito

more data under varying operating conditions

but using same calibrations & algorithms for all the data sets

fit 6 free parameters - shadow position (2), width (2), occultation fraction, and (linear) declination slope

3 data sets - with 1m depth, 1.5m depth, 2m depth (last only 5-6 days)

first two give 5 sigma shadow, last one just 2 sigma, but limited data

data set 2 gives offset (-0.25, -0.25) in Moon minus true position

narrower shadows are produced by more stringent fitting cuts

width of shadow is 0.5 degrees in data set 2 with nfit > 50

Return to index
 

Morgan Wascko - Moon Shadow Simulations

tracking cosmic ray motion between moon and earth

assume earth is a dipole field

tracked moon over the course of a day, found systematic differences over the course of a lunar

Return to index
 

Andy Smith - Markarian 501

Hegra finds Mrk 501 is about 2 Crab - so what should we see?

depends somewhat on the spectrum - which is flatter than the Crab

we see ~4 sigma, giving s/n ~ 1/140 - observed

we would have expected about half this, implying that Mrk 501 is actually about 4 Crab with our energy response

a better analysis may give 6 sigma in 150 days - which is about 0.5 sigma per day

eventually, 10x data rate in Milagro, so weíll get 1 sigma per day

center of Mrk 501 signal is offset by (0.25,0.25)

Return to index
 

Gus Sinnis - Markarian 501

gets 1.7 sigma in one-degree bin centered on true center

daily excesses up to ~ 2 sigma

folded with 29 day period from Telescope Array observations, see some dependence on phase, but noisy -

Return to index
 

Jim Ryan - Solar Physics

looking at scalar data to see major solar event of 1997 Nov 6 - which produced GeV particles observed at Mt Washington, also Durham and Climax

in Milagrito? scaler rate efficiency will be comparable to neutron monitor

high threshold scaler rate starts to rise a few minutes after Climax onset, an hour and a half before sunrise (therefore not light leak)

trigger rate (100 pmt) also rises, harder to understand

after correcting for temperature and pressure (coefficients of Fowler & Wolfendale), still see a dramatic effect - rate increase of 2 parts in 400 -

is it real or some other artifact?

not a light leak, since it occurs before sunrise

aberrant behavior of one pmt?

rate increase not uniform across the pond - tube 103 sees it most (big effect), and neighboring tubes are also bumped

increase occurred in two steps, and first increase is seen in a different tube, later increase is seen in yet another tube - both of these are neighbors of tube 103

one tube is set off as a result of a high count rate in another?

possibly a "flasher" - one tube is caused to emit light - gas or base arcing -

this could explain why the 100-tube trigger rate goes up too

either we're really unlucky that the increase coincides so well in time with the Climax neutron monitor - or - ??

should look in patch-by-patch data for high threshold, should also look in igloos

Return to index
 

Cy Hoffman - NIM-type Article in Astroparticle Physics - or in NIM?

should include Moon shadow, Markarian, results -

description and diagnostics of the instrument

need someone to take control of the structure / shell of the paper and getting it through the process

Jordan volunteered to do this, and was accepted

 
Return to index

Jordan Goodman - Data Analysis Plans and Discussion

What should we discuss or reveal in public?

Moon

we are moving forward to publishing the moon stuff -

we have clear evidence of the existence of the shadow, in 3 independent data sets

offset is uncertain, but we have no clear idea where it should be

the width is also not a direct measure of our resolution, but an upper limit

therefore we can talk about the moon stuff in public

Tony will make up a set of standard plots - so this can be presented

Markarian 501

we probably have a signal there, but weíre still trying to understand it

significance much less, at different spots in the sky

also this is a physics issue - says something about the source as well as the detector

John Learned says we should announce that weíve seen the source - to help the field

but if we are seen to be struggling to see the strongest source in the sky, perhaps thatís bad too

Other analyses we should do with Milagrito -

when do we decide whether we should look at Crab, or Mrk 421, etc?

Letís spend the next few weeks fixing the calibrations

Ask Andy to be a corrections working group head to settle some of the issues of calibration - curvature, slewing, sampling corrections - have a web page to store the graphs and so forth

Then letís reprocess Mrk 501 - then thatís what we get

Then letís start looking at other sources - need to buy enough disks to store the reprocessed data

What we want to do :

sources of cr; agn; grb; pbh; TeV photon absorption

How to do :

cr sources: DC sky maps - do online

transient sources, agn - source files

grb, pbh : short time sky maps

composition studies: summary files, prescaled

Return to index
 

Baffles and Igloo Decisions -

Propose no decision about baffles right now - until detailed proposal is in

baffles in hadron layer still an open question -

need to do more simulations, need to do installation plan

but it looks like baffles in the air shower layer will do us good

Decide to put baffles on the air shower layer - yes

Muon layer - decision - how to make?

opposing the elimination of the muon layer are: Jim R, Galen, Todd, Stan, and (weakly) Brenda -

big plusses for elimination are: larger hadron layer, exterior outrigger counters -

big plusses for keeping it are: more definite discrimination for muons, easier trigger for the solar physics, better composition analysis at high energy

Vote on elimination of the muon layer: 4 opposed, > 10 in favor

Decide to eliminate the muon layer and go with an expanded hadron layer - yes

Now - should any muon boxes be included? -

keep 7 to 9 muon boxes clustered in some location in the pond - not a critical path decision -

 
Return to index