Build a Geomagnetic Observatory ! 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 universities and 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, Links and References (very early stages)
Journal Notes:
Monday, April 11, 2011
Overnight: PDF, TXT, Polarization time controller PDF.
It appears curves of Tau 2 (the NMR parameter related to the decay of the precession signal) versus fluid temperature might be one good way to compare NMR fluids for use in proton magntometers. The FDM PPM is able to directly measure Tau 2 as one time constant (A*1/e) of the free induction decay (FID) signal. Also, we can periodically take direct measurements of the fluid sample in the 125 mL Nalgene bottles using a fast responding immersion thermocouple probe (checked at ice bath and boiling water).
As an aside, while I now question some of our early fluid testing at fixed polarization times, no doubt, one of the most valuable lessons from that excerise was that almost every fluid tried both made a precession waveform as well as having measurable Earth's field NMR (EFNMR) characteristics. My initial thoughts that fluid type or purity might be critical to a successful proton magnetometer experiment were found to be without merit.
Today, I tried switching back to an old bottle of 70% isopropyl alcohol used earlier in the experiment. Although I now believe our original fluid data at a fixed polarization time is of limited usefulness, that original data did suggest that alcohol would have a lower Tau 2 than the Preston DeIcer. Here is a very rough first comparison (based only on a few measured points) of Tau 2 versus temperature for alcohol (70%) and Prestone DeIcer PDF (revised 4/13). Recall that 70 % alcohol (freezes probably above -10 F) is not an option for our upstate NY winters where the outside air temperatures can fall to -30 F. However, as we noted last week, at extreme summer temperatures, a relatively high NBLNA gain is used (over 500,000) with the Prestone DeIcer sample to keep the polarization times under about 5 seconds. In some locations, it might be desirable (this is far from clear) to change out fluids from winter to summer.
After placing the alcohol sample in the FDM magentometer, I lowered the NBLNA gain quite a bit to give about a one second polariation time for a filtered precession waveform peak envelope voltage of about 2 V at about 77 F (there was an abnormally warm afternoon today with further heating by direct sunlight). The PEV servo was engaged, and the adjustment was made with very fast cycling (about a 10 second measurement cycle). At the original gain (551,000) recently set in anticipation of summer temperatures with the Preston DeIcer fluid, the system was slightly less stable with alcohol, as indicated by wider correction swings on the polarization chart. With the gain significantly reduced, here is a first look at the polarization time controller curves PDF. (the step in fluid temperature near the end was from adjusting parameters).
So, is a shorter Tau 2 better? Not necessarily, since a smaller Tau 2 value means a faster decaying precession waveform and less time to digitize at relatively high amplitudes. On the other hand shorter Tau 2 generally means shorter Tau 1, which means for a given polarization time, a higher percentage of the protons become spin aligned. Also, FDM is so very powerful as a frequency estimator, that decay time is now less of a concern for the frequency estimator.
Rather than just saying that some fluids are "better" than others for use in proton precession magnetometers, I hope to develop such fluid experiments as plotting Tau 2 versus fluid temperature, to offer more quantifiable differences between fluid types. None of the this work has much direct impact on the most important aspect of the FDM magnetometer, the magnetograms PDF generated of the F scalar magetic field. Nonetheless, for some students and experimenters, this project will likely be more of a PPM laboratory development system than a geomagnetic observatory. And, for the observatory only folks, we should have a finely tuned system before long!
Tuesday, April 12, 2011 Kp 5 - Minor Magnetic Storm! (higher mid-latitudes)
Overnight: PDF, TXT, NRCan OTT PDF, NOAA Space Weather Now PDF, NOAA Wing Kp index PDF, USGS PDF. Looking at the USGS plots, this event appears to be more significant at northern latitudes (although upstate, NY, US, is at the southern edge of such a definition). The Ottowa magnetogram shows a significant swing in the Earth's F scalar field. There is a Canadian Space Weather warning page PDF, useful for events such as today's geomagentic storm where there are less affects to the south. Nothing official has called a storm yet, however this is certainly a storm like behavior! NOAA K indices show a global Kp 5 event PDF. Auroral activity has extended south towards the northern US, too bad it is day time PDF, PDF. Overview of the Event 11:12 local (1512 UTC) PDF. An ISES (The International Space Environment Service) link has been added to our references and links page. The ISES page provides links to the global geomagnetic warning centers. Noon: There was a very fast ~15 nT rise in the field just before noon PDF, this minor geomagnetic storm is not over yet. 1:50 pm PDF. 4:15 pm PDF., 10:35 pm PDF, interesting fast sinewave like disturbance about 9 pm; it was an active geomagnetic day.
Wednesday, April 13, 2011 - Very Active
Overnight: PDF, TXT NRCan OTT PDF. The geomagnetic field, at least at the higher latitudes, continued to be very active overnight and through this morning. Afternoon: The geomagnetic field continues to be unsettled PDF. Two day view PDF.
Thursday, April 14, 2011
Overnight: PDF, TXT, morning PDF, NRCan PDF. The morning down-turn is normal (diurnal cycle), however the field is slightly unsettled.
It seemed that the relationship between Tau 2 and polarization time was an exponential fit. We have been scatter plotting Tau 2 vs. polarization time while in the PEV servo mode. PDF It is increasing looking like the fit might be a quadratic equation. Outside air temperatures are supposed to fall to about 26 F overnight. It will be interesting to see more of the lower part of the curve for our present NMR working fluid, 70% alcohol. Polarization time controller chart PDF.
Friday, April 15, 2011 - System Failure
Overnight: PDF, TXT. First failure ever of a switch control board overnight. This one had seen a lot of abuse during testing and ran just over four months. With nearly 10,000 relays in stock, no experimenter will ever be without a relay. At least for this first version, the simplicity and outstanding isolation of the small telecom relay in the zero-current switching FET-relay hybrid circuit justifies a change out once in a while. Will do a relay autopsy later. While we missed the low temperature overnight for Tau 2 work, it was interesting to see the the zero polarization noise level for the peak envelope voltage was about 12 or 13 mV for hours (compare to the normal operational 2 V servo'd peak amplitude) with no significant changes from EMI/RFI noise pickup PDF.
Evening: This first failure of the switch control board was more than just the relay, I am still trying to understand what happend. So far voltage and current waveforms have yet to reveal the problem. After the small signal relay was replaced, the system ran okay in terms of magnetic measurements, however the PEV servo was exhibiting very wide swings caused by large shot to shot variations in the precession signal and the average figure of merit was a little higher than normal. I double checked with another new relay, just incase I got the odd failed new part, no change. Finally, only after looking at waveforms all afternoon did not reveal the problem, I put in a new switch control board to rule out a failure somewhere else in the system, and all is fine. So, there is some subtle failure on the switch control board from this morning's failure, beyond the relay. The good news is that the relay autopsy showed no sign of contact arcing at all, so there have been no timing failures. That is, the relay always switched during zero coil current, as designed. Also, we are now seeing less scatter in the Tau 2 data, and I forgot to put in the FET snubber capacitor. Perhaps that capacitor is not a good idea, or it might need a different value, or the lower scatter in Tau 2 might just be the present short term EMI/RFI environmental conditions. After we get some run time on the new board 'as is', I might add the snubber ... or, perhaps that very high frequency, very short duration burst of pulses (I think it was a couple of microseconds) is simply not an issue and the snubber capacitor on the drain does more harm than good. Most important is to find what else failed on the switch control board! I am usually pretty fast and thorough at diagnosis and trouble shooting, not so much today! The shot-to-shot Tau 2 measurement scatter is at the lowest levels I've ever seen PDF. The scatter in the PEV servo is up a little, however, all that means is that PEV servo is used to determine the curve fit for Tau 2 to polarization time (for a given NBLNA gain) and then, if the Tau 2 measurement remains this incredibly consistent, the system will very likely run better in the Tau 2 feed forward mode. This was the orginal intention, I was surprised when the PEV servo mode was outperforming the Tau 2 feed forward mode. If all goes well, we should have a few more cool overnight periods at least down to about 35 F, so that we can acquire enough data over a large enough temperature range to derive the curve fit for the 70 percent alcohol, our present working NMR fluid. Perhaps today's failure was a good thing, providing us with a bit of serendipity that will lead to better system!
Late evening/early morning: The Tau 2 data (green curve) is better than ever before Controller PDF, magnetograms PDF, PDF. With enough data in the PEV Servo mode to provide the Tau 2 - polarization curve fit, we can most likely achieve better peak envelope control in the the Tau 2 feed forward mode.
Saturday, April 16, 2011
Overnight: PDF, TXT. Despite still being very broken (?), the system continued to produce a perfectly acceptable magnetogram overnight. The step change just after 12 am is from a departed vehicle that parked too close to the counter-wound coil pair.
I ran overnight in the peak envelope (PEV) servo mode hoping to collect enough Tau 2 and polarization time data to produce a curve fit for the Tau 2 feed forward mode. After some weeks of operating in the PEV servo mode, we now know what to expect for short term variation in the polarization time record (a short tracking error in holding a desired peak envelope voltage of 2 V). Well, the overnight record shows a failed system PDF. Note that we did change over to the Tau 2 feed forward mode for a bit near the end of this record, and as expected, the peak envelope voltage settled down somewhat. There could be more than one thing going on here. This was the coldest we have run the very old bottle of 70 % alcohol, and the Tau 2 - polarization time record shows that the system hit some kind of "wall" below a Tau 2 of 0.4 seconds PDF. Whether that is the fluid near freezing or some other failure at the sensor related to freezing temperatures, I do not yet know.
So, what happened? yesterday I focused on the switch control module, since the original symptom was that the relay had stopped operating. Here is a polarization controller chart 4/14 11:55 pm PDF, from before the total system failure of 4/15 after midnight. We had already switched over to taking some measurements with 70% alcohol as the working NMR fluid. Note that the darker blue curve shows that in the PEV servo mode, the system was generally holding a constant peak envelope voltage (of the precession waveform) to about 2 V +/- 0.1 V or better (by varying or servo action of the polarization time, the red curve).
The main failure: I was hoping to get a first look at tau 2 versus polarization time for the 70 % alcohol down to about 25 F in the overnight period of 4/14 to 4/15. However, in the morning of 4/15, the system was totally down (no precession signal). Here is the polarization controller curve PDF that I found. Although it is very quiet with the switch control cover on the SWCTRL module, I could hear that the relay was not operating. Also, the FDM magnetometer control page showed no precession signal PDF. Here were the last few plotted records:
date time F Scalar Prec. Frq FDM Ampl FDM Fig. Merit NB S/N Pk EnvV Pol.t Tau 2 Sens T Fluid T 4/15/2011 12:38:08 AM 53692.43 2285.263 2.195 0.000000020 22.2 2.081 0.666 0.403 34.59 38.18 4/15/2011 12:40:07 AM 53691.96 2285.243 2.142 0.000000600 36.8 2.006 0.657 0.402 34.51 38.09 4/15/2011 12:42:05 AM 53691.89 2285.240 2.162 0.000000300 37.7 2.014 0.647 0.404 34.39 38.13 4/15/2011 12:44:03 AM 53691.54 2285.225 2.162 0.000001000 32.9 2.029 0.639 0.390 34.28 37.45 4/15/2011 12:46:01 AM 53691.23 2285.212 2.126 0.000001000 36.3 1.960 0.626 0.401 34.14 37.39
While we do not save the non-plotted data file, since the polarization controller curve uses that data, we have it from the excel spread sheet (post processing data):
4/15/2011 12:42:05 AM 53691.89 2285.24 2.162 3.E-07 37.7 2.014 0.647 0.40 34.39 38.13
4/15/2011 12:44:03 AM 53691.54 2285.225 2.162 1.E-06 32.9 2.029 0.639 0.39 34.28 37.45
4/15/2011 12:46:01 AM 53691.23 2285.212 2.126 1.E-06 36.3 1.960 0.626 0.40 34.14 37.39
4/15/2011 12:47:59 AM 18 0 0.000 0.E+00 0 0.013 0.626 0.56 34.04 44.92
4/15/2011 12:48:12 AM 18 0 0.000 0.E+00 0 0.011 0.814 0.63 34.02 55.08
4/15/2011 12:48:24 AM 18 0 0.000 0.E+00 0 0.012 1.193 0.74 34.03 67.17
after the 12:46 plotted point, the system went into the auto-retry mode at point #2957, and continued to try to take a measurement about every 12 to 15 seconds (15 second auto-retry rate with the polarization time maxed out at 3.5 seconds) until point #5079 when the system was turned off. The PEV servo had extended the polarization time to the maximum (a temporary software safety limit) value of 3.5 seconds, so the auto-retry rate had extended to about 15 seconds very early on by point #2962. Despite the relatively fast cycling with a polarization current of 1.5 A for 3.5 seconds, the sensor air temperature continued to fall to 32 F overnight, so there was no significant overheating, at least not at the inside of the PVC coil form near the top of the 125 mL fluid bottle.
what we did: After changing out the relay, the polarization time controller showed poor PEV servo operation. I tried another new relay, just incase there was a bad part, there was no change. After some hours of looking at voltage and current waveforms for the SWCTRL module failed to show any problems, I swapped out the SWCTRL board. There appeared to be some improvement, however the overnight PEV servo operation failed. On the other hand, there was anomalous behavior of the system near a Tau 2 of 0.4, which ironically is exactly where the failure occurred on 4/15 just after midnight.
As best as I can tell, such as by viewing the powered coil voltage during a polarization cycle (9.44 V 1.5 A, 6.26 ohms, sensor 43.7 F, fluid, 45.4 F), there is no intermittent in the powered coil or cable to it. [I have a previous reading of 6.21 ohms at about 33 F from months ago, so the powered coil seems okay.] It is very windy and I should do a resistance check of the non-powered coil of the counter-wound pair to rule out an intermittent in that circuit. An intermittent there could cause a varying feedback value for the PEV servo. On the other hand, now that I switched over the Tau 2 feedforward mode this morning, the peak envelope values have settled down to near what they were PDF (see after about point 501 where I switched over to the Tau 2 feed forward mode). This renewed stability in the peak envelope values would seem to suggest all cabling, coils, and the NBLNA module are okay.
Ironically, our relatively new Tau 2 measurement system is working better than ever before! The only other change that might affect it is that I forgot to install the snubber capacitor at the power FET drain. That had been installed to suppress a very short (2 us) very high frequency series of pulses at powered coil shut down. It might have added to a low frequency component of the precession waveform envelope, and might do more harm than good ... still to be determined.
Where to go from here? I am thinking the best course of action for now might be to go back to some "known" conditions with the Prestone DeIcer fluid. Here are the last records of 70 % alcohol from today PDF, TXT before switching back to Prestone DeIcer. Note that for all my concerns, the magnetogram is fine. Also the precession waveform filtered envelope graph shows our new found improved Tau 2 curve! (with some variation shot-to-shot) The non-powered coil looks okay at about 6.18 ohms, non-powered coil about 6.15 ohms (Agilent 34410A, 2 wire, nulled for test wires to cable connections) with no sign of intermittent connection with very strong wind gusts today.
Evening: We had a power outage, however I installed an older NBLNA board with a relatively low fixed gain (TBD), possibly not center frequency aligned. I set the desired peak envelope voltage low at about .85 V. The system is running fine with a mode (most common value) of figure of merit (FOM) at 5e-7, a familiar value from past month's runs. It is possible that operation at relatively high gains (e.g. 500,000 +) is less desirable, especially in terms of scatter in the Tau 2 measurement. There is a 1 V USB 6008 range, so there is really no need for a 2 V signal as we have been using in recent weeks. It probably makes sense to further distribute the system gain into the USB 6008 module. Many small gains in this type of instrument is always desirable over fewer high gain stages. After taking some data as is, we will switch back to the newer NBLNA and study operation at lower gain (e.g. 250,000 to 300,000). It is possible that I lost track of NBLNA gain settings and that some of the recent very high scatter plots were simply the expected result of operation at very high NBLNA gain settings.
All indications for low gain are looking promising. I suppose it was not until we had the Tau 2 and polarization performance charts that we could begin to optimize the system. Here is the magnetogram PDF, which despite the stormy weather and high winds, is incredibly quiet. The FDM FOM and FDM amplitude also look very good PDF. The detail on the precession waveform after apodization might be another indication of the advantages of running at lower NBLNA gain. The old NBLNA board installed this afternoon needs to be characterized, before returning the working version.
Sunday, April 17, 2011
Overnight: PDF, TXT. Polarization Controller PDF
(See afternote below) I happend to be looking at the magnetogram when another high wind event began. About 6 am, I was alerted to the high wind by rattling windows; a local weather page PDF (scroll down for their wind graph) confirmed a large step change in the wind speed. About that time I noticed a minor disturbance in the magnetogram. I also noticed a worsening of the polarization time controller chart, which otherwise had run very well overnight. It is possible that the increased activity both on the magnetogram and the polarization chart (independent of the field measurement) were influenced by the high winds. For example, during some of our most confusing indications on the polarization time controller chart of yesterday, we were experiencing continuous wind gusts of 30 to 40 mph. The system has run through the most active thunderstorms with no affect, but maybe there is one minor weather influence, as wind affecting only the polarization time controller. This affect, if the case, might be either due to agitation of the NMR fluid, and/or movement of the relatively light construction sensor stand. There is a known magnetic gradient at the stand extending across the counter-wound coil pair, at least in part do a nearby parked vehicle which causes a known 18 nT depression in the field value (F scalar) at the stand. A disturbance of the polarization controller activity with high wind is of little consequence, as long as the system continues to generate a useable precession signal. Wind affects on the magnetogram, if the case, would be far more disturbing. High winds are very unusual here, however if the case (not clear at all, still to be determined), builders in high wind areas might use 2" x 4" lumber (wood) or the international equivalent, instead of our recommended 2" x 2" to build the sensor stand. Fortunately, the FDM FOM amplitude shows exceptional performance overnight, with a FDM most common figure of merit (FOM) value (the statistical mode) of 2e-7, FDM FOM amplitude PDF, Tau 2 vs. polarization time PDF. While we need to regroup and sort out the performance data of recent days, the system is clearly okay now (not broken). Afternote: There does not appear to be any connecton between high winds and the magnetogram. It is a trend of recent weeks that unsettled geomagnetic conditions cause some scatter of the magnetic record, believed not to be related to vibration at the sensor stand. On some days there is some correlation to sunrise.
In any event, now I need to remove the temporary NBLNA board (an older board used for a test) and measure its center frequency and gain. Also, I need to check the resonating capacitor bank resonant frequency with the sensor conter-wound coil pair. Plots before shutdown PDF, TXT, NRCan OTT PDF, Polarization Controller PDF. First measurements of the old NBLNA board that we ran on overnight. I thought that I had properly peaked the frequency adjustment by watching the waveform envelope display. Apparently not, the center frequency was set to 2,094 Hz with the gain at center f at about 600,000. However, we operated overnight with Larmor frequencies ranging from about 2283.7 Hz to 2285.4 Hz, where the gain was down to about 150,000 (with about a 60 degree phase shift), so we operated at a relatively low gain, which explains the long polarization times, at times approaching 3.5 seconds.
The capacitor bank on the SWCTRL board was resonated to about 2285 Hz using a signal generator and an AC voltmeter (need 4 1/2 to 5 1/2 digits for this job). Our present resonating capacitor bank value is 113 nF (was at 2,330 from the quick change out the other day). I need to revisit the question of the snubber capacitor (it is presently out, but will probably go back into the working board. The new NBLNA board was tuned up (probably not needed) to a center frequency of 2285 Hz. Since we are leaning towards using the USB 6008 1 V scale, the gain was set to 200,000 (so, some of the gain effectively moves into the USB 6008).
okay, back up and running. I would like to see less scatter in the peak envelope control, however with a FOM statistical mode of 1e-7 and lots of values in the -8 range, I am not complaining. PDF, polarization controller PDF, FDM FOM and FDM amplitude PDF . I will try to let it run undisturbed for a while and see what happens.
Project Documentation, Links and References (very early stages)
QUESTIONS/COMMENTS/notice of typos, etc. send email to joegeller @ gellerlabs dot com
COPYRIGHT © 2009, 2010, 2011 JOSEPH M. GELLER, All rights reserved.