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
Journal Notes:
Monday, February 21, 2011
Tuesday, February 22, 2011
Wednedsday, February 23, 2011
Overnight: PDF, TXT While quiet here in upstate, NY, there has apparently been some geomagetic activity between the 20th and the 22nd to the North as reported by the College, Alaska station: NOAA K Index PDF.
This latest run has been going since about 8 pm local on 2/21/2011. After 1,094 plotted points (9:37am, 2/23), the success rate has once again converged at about 77%, very interesting (no affect on instrument accuracy or use, just a curiosity, no more).
Before I recently realized (courtesy of severly reduced credit lines from Bank of America) our engineering company has some financial problems (apparently this problem is common to millions of US small businesses), I had ordered a new old stock Tektronix AM 5030s Amplifier and the matching A6302 current probe. I was hoping to confirm past measurements of powered coil discharge. In the past I had floated the scope chassis ground to look at the voltage across a series current shunt resistor. I had to remember to disconnect the floating current shunt common before I connected any other grounds, including the GPIB connector to take a screen shot. The AM5030 is a very impressive instrument, even if a bit dated. I think that I have mastered the basic operations and probe calibration and all seems to be working well. The problem is, if correct, I no longer completely understand the powered coil discharge process. Here is a screen shot for one of the powered coil discharge curves PDF. The vertical scale is 500 mA per 10 mV, so at 5 mV/div, it's 250 mA per division. As expected, the coil current flow at the beginning of the powered coil discharge process is about 1.5 A (that is where we have the constant current set point on the Agilent E3615A polarization power supply presently set). So, in the upper left hand side of the page, the curve starts out at +6 divisions, or 1.5 A (notice that common is set in the lower left at +1 vertical division). At FET turn-off there remains a small couple of microsecond wide shallow rectangular pulse.This is where there previously was a very high frequency resonance (at a much higher frequency than could be attributed to powered coil self-capacitance) before we added a small snubber capacitor. The fall is a nice linear negative di/dt ramp, presently falling 90% to 10% in about 430 uS (scope numeric display, lower left). Then we have a slight over shoot slow resonance somewhat damped curve which goes to about -2 mV or about 100 mA negative. The system is mostly damped and does not appear to oscillate further, infact it is only very slightly under-damped. I do not recall if I took further data once we raised the value of the secondary dump resistor from 403 ohms to 1.1 k ohms. The slight surprise is that I no longer see that final shift of change in slope after the FET reverse avalanche mode shuts off and discharge is handed off to the secondary dump R . Ah well, as is the case in most of these types of experiments, typically a new more sophisticated piece of test equipment ends up raising more questions than it answers! In any event the FDM magnetometer continues to perform magnificantly, so all these detailed measurements are just to further understand the instrument, and perhaps later to make minor improvements.
Thursday, February 24, 2011
Friday, February 25, 2011
Today I added pin jacks (the same gold MillMax pin jacks that we use for the relay socket) for the secondary dump resistor. I was able to place a range of Rs in the jacks and monitor the current through the secondary dump resistor using the Tektronix A6302 current probe with the AM 5030 amplifier. The waveforms were displayed on an hp 54645A oscilloscope. The results are quite interesting. Secondary Dump R 100 ohms, 100 mA/div PDF; 200 ohms, 50 mA/div PDF; 301 ohms, 50 mA/div PDF; 402 ohms, 50 mA/div PDF; 1.1 k ohms, 10 mA/div PDF; and 2.2 k ohms, PDF. Notice that the damping of the oscillation following the completion of discharge by the secondary dump resistor decreases with lower valued resistors. Also, as expected, the final fall time increases with smaller valued secondary dump resistors (L/R with the L of the powered coil ~22.1 mH). It turns out that the original selected value of 402 ohms provides the best compromise between oscillation (damping) and fall time. Fall time increased dramatically below about 300 ohms, and after the ~180 uS fall time for 200 ohms, by 100 ohms there was no longer a usable precession signal.
What do these curves mean? How do you read them? Let's take the working curve for the 402 ohm secondary dump resistor PDF. The secondary dump resistor is always in parallel with the powered coil when the relay is in the polarize position. To the left of the first horizontal division (left side of the oscillogram), it can be seen that the current through the resistor is about - 25 mA. This is caused by the polarization voltage (roughly 10 V) across the powered coil. R=E/I or about 10/.025, so about 400 ohms. At the onset of powered coil discharge, there is about 1.5A of current flowing through the powered coil and about 10 V across the powered coil (which until that moment, prevously caused the 1.5A current to flow, typically for about 2 seconds for the polarization pulse). At the moment the FET turns off, removing the +10 V polarization voltage from the powered coil, 1.5A is still flowing. The characteristic of a discharging inductor is that the voltage across it reverses, while the current discharge begins from the last flowing current, here the 1.5 A polarization current. At the first horizontal division, it can be seen that the current flow in the parallel secondary dump resistor very quickly changes polarity and increases to about 2.8 vertical divisions. At 50 mA / vertical division (by the Tek current probe system), that is about 140 mA. For the 400 ohms of the secondary dump R, E=IR or .14*400 is about 56 V, about what we expect as caused by the FET's reverse avalanche mode. The coil is charged to 1/2 L*I^2 (L times I squared) or about 25 milli Joules. By about 2 1/2 horizontal divisions (about 500 uS), there is insufficient energy left in the powered coil to support 55 V to keep the FET conducting in the reverse avalanche mode. Now the circuit revert to a convention exponential L-R discharge. Tau (L/R) is about 55 uS for the 22.1 mh powered coil. The remaining energy is mostly discharged along this LR discharge curve. There is some overshoot which might be due to an RLC oscillation related to the 0.1 uf snubber capacitor (needs further investigation). Note that the resonating capacitor bank is not in parallel with the counter-wound pair until much later when the small signal relay changes state. Also, I believe this oscillation frequency is much lower than can be attributed to coil self capacitance.
When the FET turns off and the remaing inductor powered coil current entirely hands off to the secondary dump resistor, the current flow is about 140 mA (for a 400 ohms secondary dump resistor). From previous measurements with the Group3 longitudinal Hall probe JPG, we know that the B field center coil is about 134.8 Guass for the 1.5 A polarization current PDF, See also the Jan. 27 journal entry. Therefore at hand-off to the secondary dump resistor at about 140 mA, the B field is about 12 or 13 Gauss.
One of our readers told us that a Russian paper reports that the fall time (rate of fall) of a powered coil waveform through a lower field (just above and then comparable the Earth's field) is most critical for Earth's field NMR (EFNMR) based proton precession magnetometers (PPMs). Afternote: See our Monday, May 16, 2011 journal notes on the Varian Fast Discharge.Today's measurements appear to support that theory. For all of these measurments, keep in mind that the bulk of the fall time was at a constant di/dt (determined by the avalanche mode (~55V) of the power FET). The secondary dump resistor curves are "flat" for most of the powered coil discharge time. This is because for most of the powered coil discharge time, the stored energy is being discharged by the avalanche mode of the FET and the coil volage is relatively constant (Vcoil=Ldi(t)/dt + i(t)R). The slight slope is due to a relatively small voltage drop that can be attrbuted to the R in the circuit. As the powered coil current falls, so does the "iR" voltage drop attributable to the series R.
The secondary dump resistor only comes into play when so much energy has been discharged from the powered coil, that the FET reverse avalanche mode can no longer be supported and FET conduction ends, shifting the current discharge for the relatively small amount of remaining stored energy over to the secondary dump resistor which is electrically in parallel with the powered coil when the SWCTRL relay is in the polarize position (and not in the receive, or amplifier position). So, we return to 402 ohms as the recommended value for the secondary dump resistor. [Here is the overall discharge curve of the powered coil (Tek A6302 on the wired to the coil) (250 mA / vertical division) PDF. It is unclear to me why the final discharge is not steeper corresponding to the hand-off to the secondary dump resistor. There could be a measurement issue, or there could be an equivalent RLC circuit with the 0.1 uf snubber capacitor that is rounding out the very bottom of this waveform at the SWCTRL end of the long cable (this part makes no sense yet, needs further investigation).]
Therefore, it would appear that the di/dt of the bulk discharge (in our case for a 55V avalanche mode, about 400 microseconds) is less important compared to the rate of fall needed to remove the last bit of stored energy when the polarization falls through the last few Gauss or so. It would appear that this is where the ~100 uS discharge time "requirement" comes from, although, the maximum usable discharge time is probably closer to 200 uS, perhaps dependent on the uniformity or configuration of the polarization field.
For our present configuration with a powered inductor inductance of about 22.1 mH, to achieve an end discharge curve fall time on the order of 100 uS, we need a secondary dump resistance over about 350 ohms. At present, to minimize post discharge oscillation, the resistance should be under about 1.2 k ohms. Another consideration might be B field at the moment of hand-off from the avalanche mode to the secondary dump resistor. The current at hand-off is about 55 V/R. Therefore a lower valued R, peforms hand-off sooner, at a higher field. For example, for a 400 ohm secondary dump R, hand-off happens around 140 mA or about 12 Gauss. For a 1.1 k ohms, hand-off occurs at about 50 mA or at under 5 Gauss.
Now, if the success rate falls below our 77% value of late (at 1.1 k ohm secondary dump R, now back to 402 ohms), we will know that final fall time is more important than damping. (Ran overnight with 0.1 uF ceramic snubber capacitor and a 402 ohm secondary dump resistor).
Saturday, February 26, 2011
Overnight: PDF, TXT , success rate with the secondary dump R reduced back to the original value of 402 ohms is about 79%.
I noticed that the overshoot might be caused, as suspected, by the 0.1 uF (100 nF) snubber capacitor. For a 400 ohms secondary dump R, the energy left in the powered inductor is about 217 milli Joules when the resistor takes over. At that moment, with 55 V across the snubber capacitor, the stored energy in the snubber capacitor (0.1 uF, 55 V) is about 151 mJ. If the overshoot, or underdamped oscillation (with the secondary dump R > ~1 k ohm) is caused by the snubber C, changing it to about 33 nF will significantly change both the frequency and damping; more later.
More interesting results. The 0.1 uF ceramic snubber capacitor across the FET was replaced with a 33 nF film capacitor (310 V). Here are the new secondary dump resistor waveforms: 402 ohms PDF; 1.1 k ohm 10mA / div PDF; 2.2 kohm, 5 mA / div PDF. The underdamped oscillation frequency tripled, confirming that the RLC circuit is the powered coil inductance (22.1 mH) and the snubber capacitor (100 nF to now 33 nF). Here are the new main coil current waveforms (250 mA / div) PDF, and a detailed view of the very fast spike at the start of powered coil discharge PDF (at FET turn-off, long before the secondary dump R comes into play). With the snubber capacitor reduced to a 33 nF film capacitor (slightly higher ESR), the fast spike is larger and more pronounced than before. The dip and underdamped MHz oscillations at FET turn-off might be associated with the FET entering the reverse avalanche mode at the onset of the powered inductor discharge.
With the new snubber capacitor and a 2.2 k ohm secondary dump resistor, the precession waveform amplitude has increased by about 15%. We will run in this configuration for a while.
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.