Slow monitoring and recording system summary

Updated January 2, 2008.


Magnetic measurements


(GAHS 2006/06/20)
PostgreSQL. See DC Slow MRS database information page for SlowMRS table structure, performance, and access information.
Data-supply clients:
Java, Python, Perl, ADA, and ROOT/C++ supported for sure; pure C, C++ and Fortran allowed, but more difficult.
Display and analysis clients:
Same as above.
See also slides 10-13 of March 13, 2006 presentation.


(GAHS 2006/06/08; updated 2006/09/07)

The basic plan for the slow monitoring hardware hasn't changed since it was originally proposed: Dallas Semiconductor "1-Wire" chips connected with CAT5 wiring over long runs [1].

(Quick summary of 1-Wire system)

The plan is to run most of the chips in "powered" rather than "parasitic" mode for fastest network performance. So four conductors will be used in the CAT5 wiring for our "1-Wire" bus, corresponding to pins 1-4 on an RJ11 connector, per Fig 4 of [2].

   Device count as of 2006-March-12:
   (from presentation at 2006/03/13 collab. mtg. at ANL, slide 5)
   LS/oil temperatures:  6 per detector (12 total)
   Temp/humidity:        3 per lab (6 total)
   AC line monitor:      2 per lab (4 ADC chips total)
   General purpose ADC:  20 discrete 4-channel boxes per lab (40 ADCs total)
   FE/WFD monitoring:    120 cards plus 30 spares
                         300 ADC chips and 150 thermometers
   Total number of 1Wire devices:  512  (256 per lab)

According to the manufacturer's reports and specifications, single-branch 1-Wire networks up to 300 m effective length will be reliable when some simple design guidelines are followed [1,3]. Each 1-Wire chip adds 0.5 m of effective length to the network. Additional capacitance (e.g., from circuit board traces) adds one meter of length per 24 pF. In principle, we could instrument an entire lab on one branch. The more conservative plan is to have at least two separate networks in each lab, one for crate electronics monitoring and one for detector monitoring, with branches in each network addressed through remotely-controllable 1-Wire "coupler" chips. [4,5] This improves reliability and ease of diagnosis or problems, and also lets any in-detector components be switched off during data taking if it proves necessary.

A few medium-scale tests with a dozen or so devices over 300 meters of CAT5e cable have been carried out. Undergraduate researcher Jim Black did many systematic tests in the summer of 2006, with encouraging results: see Jim's report and website. Among the recommended improvements are increased electrical protection for devices at the end of long cable runs and use of libusb-based linux drivers for better control of voltage slew rates and error reporting. A large-scale test with hundreds of devices will be carried out before the full system is bought.

Magnetic measurements

Magnetic measurements will be made using the Honeywell HMC 2003 three-axis magnetometer [6] interfaced to 4-channel 1-Wire ADC chips. The magnetometer has a sensitivity of 1 V/gauss, and produces an output voltage of 2.5 V + B*(1V/gauss) for each axis; the practical range is -2 to +2 gauss. Honewell claims an inherent sensitivity of 40 microgauss, corresponding to an output voltage precision of 40 microvolts. Dallas Semiconductor gives no noise specification for their ADC chips at all. Preliminary tests show RMS fluctuations of about 15 millivolts on the readout, well in excess of the digital resolution of the ADC. This is under investigation. Averaging 100 samples brings the RMS of the mean under 1.5 millivolts, corresponding to 1.5 milligauss (150 nT); averaging 2000 samples brings the RMS of the mean under 0.3 mV, corresponding to 300 microgauss (30 nT). No strap set/reset was performed [7]. These results should be regarded as highly preliminary. (2007/01/10)

For reference, the total "International Geomagnetic Reference Field" (IGRF-2005) is 48 microTesla (0.48 gauss) at Chooz (4.7894 deg W, 50.0900 deg N, el 121 m), with an inclination of 65.5 degrees and a deviation of -0.29 degrees; the magnetic field vector is (+20.0,-0.1,+43.9) μT. At Kansas State University (96.5831 deg E, 39.1931 deg N, el 333 m), the total field is 54 microTesla, with an inclination of 67.4 degrees and a deviation of +4.3 degrees; the magnetic field vector is (+20.6,+1.5,+49.6) μT. These IGRF-2005 values were calculated using the "USGS Geomagnetic field calculator" [8]. The coordinates for Chooz and Kansas State University (Cardwell Hall) were obtained from the GNV181 database, Yahoo Detailed Search, and NASA World Wind [refs to be provided].


[1] "Guidelines for Reliable 1-Wire Networks", Dallas Semiconductor application note 148, Nov 2001;

[2] Data sheet for DS9490, Dallas Semiconductor,

[3] Data sheet for DS2490, Dallas Semiconductor,

[4] Data sheet for DS2409, Dallas Semiconductor,

[5] "Advanced 1-Wire Network Driver", Dallas Semiconductor application note 244, May 2003;

[6] HMC2003 datasheet, Honeywell Solid State Electronics Center; .

[7] "Set/Reset function for magnetic sensors", Honeywell application note AN213, ; see also higher voltage H-bridge circuit at

[8] USGS Geomagnetic Field Calculator at

Local copies of datasheets for quick access

Generic Maxim/Dallas datasheet URL: Honeywell magnetic sensor datasheet URL: Application notes: Other information:

Double Chooz notes and internal presentations:

Redundant information:

Please contact me (Glenn) if you have any questions. (Unless you don't know who I am or how to contact me, in which case you shouldn't.)