Journal notes, Analog Board first run PCB
GELLER Labs "Backyard Science"
Thoughts on a proton precession magnetometer design - a Proton Magnetometer Project. Build an Earth's field magnetometer.
The FDM MAGNETOMETER1 project is a low cost high performance proton magnetometer (a digital magnetometer) kit under development for amateur scientists to be able to accurately measure and monitor changes in the Earth's total magnetic F field and to observe geomagnetic storms. Magnetic storms can cause large excursions in the field and are of concern to interests ranging from electrical power grids, radio communications, and satellite operations, to aurora watchers and amateur radio operators.
1 Filter Diagonalization Method "FDM" (harmonic inversion), see Jan 21 and Jan 23 entries, based on: Vladimir A. Mandelshtam, Howard S. Taylor, Harmonic inversion of time signals and its applications, Journal of Chemical Physics (1997), Volume 107, Issue 17, 1997, Pages 6756-6769
(be sure to hit refresh to pick up our latest changes and entries)
Project Documentation (very early stages)
Journal Notes:
Saturday, July 17, 2010
Overnight: PDF, TXT, NRCan OTT PDF, sample spectra, log spectra. Ran overnight on the new analog printed circuit board. All looks okay.
Picture of the newly assembled and tested Analog NBLNA PCB (Ver. 0.8). I already see quite a few changes! For example, the output DC trim is not needed. Taking out those parts will let us move more of the filter components farther from the input as well as to completely change the layout of the second order active filter section. Also, we are dropping the loading Rs on the output center tapped transformer coil. The input series Rs of the input common mode filter have been replaced by a lead bent for one complete loop around a ferrite bead. These parts (R1 and R2) are mounted below the ground plane on the back side of the board. The LEDs to see proper V+ and V- connections are high brightness types running on only microamps, so they will not present a significant load. Note the input protection diodes also have one ferrite bead each. The guard traces as they are now on the component side, turned out to be counter-productive. Instead, clearing a break on the back ground plane side between the front end common mode filter and the second order filter stage below it probably makes more sense. Blank board area around the input common mode filter and first very high gain SSM2019 stage (x1000) is now believed to be preferable on the component side to strips of ground traces. So, we are making quite a few changes, oh well, I guess that is what prototypes are for!
I am making some performance measurements today to characterize the board.
To characterize the prototype narrow band low noise amplifier (NBLNA), we attached two 50 ohm resistors to common from the plus and minus inputs. Then a hp 3325B signal generator was connected between the plus input and common (to simulate the precession signal from a powered coil with the ppm fluid sample).
Gain Measurement: At center frequency, as tuned by the trimmer of the second order active filter, (which only needs to be in range, we tested at 2283 Hz and 2290 Hz), for a 1 mV RMS input and a series KAY 837 attenuator set 47.5 dB, giving an input test signal of about 4.2 uV, the output was about 2 V RMS as measured with an Agilent 34410A DMM. So the amplitude of the input signal was -60 dBV, minus the 47.5 dB attenutation of the Kay attenuator, or about -107.5 dBV. The amplituded of the output signal was about 2 V RMS, or +6 dBV, so the total gain is 113.5 dB or about 474,000.
Effective Noise Bandwidth: (Note that the EFNBW is not the same as the -3 dB bandwidth.) We use a LabView program to do a piecewise integration over a desired frequency range using the hp 3325B and the Agilent 34410A DMM. The EFNBW was measured to be about 176 Hz PDF.
Input referred noise: With no signal at the input (the two 50 ohm resistors to common were still connected), the output was about 12.7 mV as measured with the 34410A DMM. The 34410A AC bandwidth is on the order of 300 kHz, however with a narrow band board under test, the effective noise bandwidth of the board is used (176 Hz). (12.7 mV/474,000) / (sqrt (176 Hz)) yields an input referred noise of about 2 nV/rt Hz. Although excellent as is, this measurement was a bit higher than expected and should be repeated with the NBLNA board in a shielded box (it was out in the open on the table). However using FDM, with a noise floor of tens of mV and a precession signal between 1 V and 2 V peak, it is of little consequence if the number is actually somewhat lower. Afternote: Accounting for the uncorrelated Johnson noise from the 50 resistors at the input, the input referred noise is less than 2 nV/rt Hz, this needs further study when the machine is down, more to follow. See July 28 entry for an updated input referred noise measurment of 1.7 nV/rt Hz using the JCan noise measurement technique.
Afternote: While the handwired prototype was correctly wired, there was an error on the protoype NBLNA PCB. The new numbers are: Gain: 588,000, effNBW: 181 Hz, noise floor: 1.8 nV/rt Hz. See our August 31 journal entry.
Our received precession signal generally falls between about 1 V and 2 V peak at the NBLNA output. Therefore our precession signal is on the order of 2 uV to 4 uV peak. (Our sample coil is 600 turns of #18 wire, e.g. as was described May 19, 2010 , we are still running with about a 125 mL (4 Oz) fluid of Prestone De-Icer windshield washer fluid, preferred because our winter temperatures can approach -20 degress F here in upstate, NY.)
Sunday, July 18, 2010
Overnight: PDF, TXT, NRCan OTT PDF. Ran overnight on the new analog board. While there were slow changes in the field over hours, there were several remarkably quiet periods (at the 0.01 nT level of resolution) overnight.
7/18/2010 6:48:54 AM 53776.62 2288.907 1.903 0.000000800 40.4
7/18/2010 6:50:54 AM 53776.76 2288.913 1.953 0.000000400 38.8
7/18/2010 6:52:54 AM 53776.90 2288.919 1.890 0.000000300 35.8
7/18/2010 6:54:54 AM 53776.95 2288.921 1.962 0.000000009 28.7
7/18/2010 6:56:54 AM 53776.95 2288.921 2.025 0.000000090 40.8
7/18/2010 6:58:54 AM 53776.95 2288.921 1.912 0.000001000 38.3
7/18/2010 7:00:54 AM 53777.07 2288.926 1.399 0.000002000 24.7
It is unclear if the field was unusually quiet over those periods of minutes, or if our instrument noise floor is lower now with the new analog board. At present, while we have relatively high end frequency and voltage calibration capabilities, we have no way to system test (absolute magnetic field calibration) at the 0.01 nT level, other than to observe the field as presented at the counter-wound sensor coils. How could one get a relatively low gradient calibration field comparable to Earth's field, and not be affected by changes in the ambient field at the 0.01 nT level over a relatively large test volume (to surround the two counter-wound coils)? While high accuracy low gradient fields at the Tesla or 10 Tesla level are not uncommon in various scientific endeavors, it is not clear to me that a 50 micro Tesla (50,000 nT) test or calibration field stable and accurate to 0.01nT is even possible?
Monday, July 19, 2010
Overnight: PDF, TXT, NRCan OTT PDF.
There continue to be some incredibly quiet geomagnetic periods over minutes at the 0.01 nT level PDF, TXT.
7/19/2010 11:07:57 PM 53772.35 2288.725 1.863 0.000000040 24.4 7/19/2010 11:09:57 PM 53772.18 2288.718 1.845 0.000000200 38.5 7/19/2010 11:12:06 PM 53772.28 2288.722 1.809 0.000000030 28.4 7/19/2010 11:14:06 PM 53772.82 2288.745 1.849 0.000000400 25.0 7/19/2010 11:16:06 PM 53772.82 2288.745 1.899 0.000000400 34.3 7/19/2010 11:18:33 PM 53772.89 2288.748 1.890 0.000002000 41.3 7/19/2010 11:20:34 PM 53772.89 2288.748 1.917 0.000000300 12.7 7/19/2010 11:22:43 PM 53772.89 2288.748 1.363 0.000000700 21.3 7/19/2010 11:24:43 PM 53772.86 2288.747 1.080 0.000000900 27.7
I wrote a draft of the first article for the Backyard Magnetic Observatory Project today. The article needs a few more revisions, a couple of pictures and drawings, then I'll send it in for publication and see what happens! The present plan is to spread out the construction articles over several issues, each article focusing on a specific aspect of the project.
Tuesday, July 20 , 2010
Overnight: PDF, TXT, several day view PDF.
Wednesday, July 21, 2010
Overnight: PDF, TXT, NRCan OTT PDF, USGS PDF.
Thursday, July 22, 2010
Overnight: PDF, TXT, several day view PDF.
Friday, July 23, 2010
Overnight: PDF, TXT, several day view PDF, NRCan OTT PDF, USGS PDF. spikes and small offsets are mostly vehicle traffic (garbage trucks, buses, etc.), however the field did become more noisy this morning (not atypical).
The field settled down tonight PDF and I got a change to re-do an experiment related to the -18 nT offset caused by a car in the garage about 33' from the backyard sensor (18 nT is added in the algorithm, since the car is almost always present). I moved the car out to end of the driveway (about a 16 nT change) PDF. The question was, does the car at 33' cause a significant gradient (inhomogenity) in the field across the sensor volume (the 125 mL fluid sample bottle) such that when it is substantially removed, would the precession signal last longer? Using the hp 3581C wave analyzer (essentially a frequency selective AC voltmeter) read out into a hp 54503A digital scope (afternote: the hp 54503 only has a single shot memory depth of 1024 points, so fall time resolution was on the order of 5 ms at best), I took 12 shots with car in the garage, CIG1, CIG2, CIG3, and 12 shots with the car out near the end of the driveway, ED1, ED2, ED3. With the car in the garage, the average precession time was about 2.0 seconds with one standard deviation of about 0.1 second, and with the car out at the end of the driveway, the average was 2.1 seconds with a one standard deviation again of about 0.1 seconds. So, the answer continues to appear to be that the 18 nT depressed field caused by the car in the garage does not correspond to a destructive gradient across the sample volume that reduces the timeduration of the precession signal.
Saturday, July 24, 2010
Several day view: PDF
Sunday, July 25, 2010
Overnight: PDF, TXT, several day view PDF. The geomagnetic field cycled thorugh a relatively normal diurnal variation and remained relatively quiet today PDF.
Project Documentation (very early stages)
QUESTIONS/COMMENTS/notice of typos, etc. send email to joegeller @ gellerlabs dot com
COPYRIGHT © 2009, 2010 JOSEPH M. GELLER, All rights reserved.