GELLER Labs "Backyard Science"
Thoughts on a proton precession magnetometer design - a Proton Magnetometer Project
The goal of this project is a low cost high performance proton magnetometer (a digital magnetometer) kit 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. There is a regular daily (diurnal) variation in the Earth's magnetic field. During events related to solar activity, there can be sudden changes in the field (such as a sudden impulse) as well as large excursions in the field which can be more than ten times the regular diurnal variation caused by magnetic storms.
(be sure to hit refresh to pick up our latest changes and entries)
Monday, September 28, 2009
Yesterday I rebalanced my "Green" and "Blue" coils for equal inductance. It turns out the steel beam under one of my lab tables was influencing the inductance measurement. That was my problem in getting a reliable and repeatable balance. This time the coils were placed on a nearby wood table top as I removed turns from coil to mach to the other. I have done this balancing several times over the month. The last version turned out to be way off, I needed to remove 24 turns from the green coil. Now both are about 7.6 ohms and 10.2 milliHenry (roughly 447 turns, #24 AWG magnet wire).
Testing the counter-wound coils in the lab for noise rejection (cancellation) is difficult. Wideband measurements seem non-conclusive and subject to capacitive pickup. Today I made a nice (albeit rough) test set up that works very well.
A Stanford SR-510 lock-in amplifier is set to 2295 Hz by an Agilent 33120A signal generator (33120A sync output to 510 sync input). I looked at a test signal using both of my two PVC coils (2" PVC). The test signal was generated by the 33120A signal output (.1V) into an old tektronix 24xx style CRT coil fed by a series resistor. The coil was placed about a foot behind the test coils on the surface of the wood table.
Distance and relative height to the coil is critical, so these are "rough" measurements. The best results were had with both coils sitting side by side (because of the proximity to the table surface). With a zero degree phase reference, the blue coil (the color of some electrical tape) read +5.1 uV and the green coil -5.3 uV. The black lead was connected to the bottom (nearest the table top) for both coils. So the counter-wound effect is good. The difference is more related to proximity the transmitting coil than any difference between the coils (so this is not a number-of-turns balancing technique).
With both bottom "B" leads and top leads "M" (for middle) in parallel and moving the coils a bit for equdistance from the transmit coil, I get 0 uV on the SR510 20 uV scale.
Also with the coils in series, so "M" of the blue coil to black test clip, "B" of the blue coil to "M" of the green coil, and "B" of the green coil to the red test clip (shielded clip cable to BNC), I also got 0 uV showing effective cancelling. Note that the series connection is not top-top of the coils for counter-wound side by side coils, but rather a simple series connection of top to bottom, which makes sense, since for ambient noise pickup hopefully one coil is making a plus signal and the other a negative signal at any point in time (to cancel the external pickup signals with a 180 phase difference by adding or superposition).
Tuesday, September 29, 2009
As I begin to fill in references, Willy Bayot's PPM-MarkIII is a notable project, a commercial quality portable PPM suitable for archeological studies. At $1k+ in partial kit form, it is certainly price competitive with comparable commercial units, but not a low cost amateur science project. Willy also wrote a helpful paper on amateur PPM building. Some of the PPM builders, including Willy, can be found at Carl's Geotech forums in the magnetometer group.
Also, I happened upon a relatively low cost PPM project published in Circuit Cellar Magazine (May 2007, pg. 14) by James Koehler (also I think a member of Willy's MarkII/III team). You can buy a pdf of the Circuit Cellar article for a very reasonable $1.50. James Koehler has also written a very helpful paper, Proton Precession Magnetometers, Rev 2, on amateur PPM building.
It is interesting to see the different approaches to sensor design. Koehler has trended towards larger volumes and heavier wire (#18 AWG), as I have also noted in an old Presentry company Canadian PPM observatory coil, presumably for higher signal and better S/N. While, the old Wadsworth project (granted a relatively low resolution device) and the Hollo's project used smaller wire (#24AWG) and smaller sample volumes. Koehler refers to the necessity of the resonating capacitance when using "smaller" coils, which might speak to the utility of going to larger diameter wires and larger volumes. It remains unclear to me the advantages or not of counter-wound coils versus same wound side by side coils. Both work, however series connections are different. Like wound side-by-side coils can be thought of as a solenoid folded over in half. Counter wound side-by-side coils can be connected in parallel. It is less clear to me that identical solenoids can be so wired, and therein might be the significant difference.
As attractive as all FET switching looks, the added complexity of isolation components is probably not worth the effort for a low cost amateur science version. I am thinking MOSFET switching for a DPDT relay that switches the coil between power (polarization) and the amplifier. That way, there can be a delay between power down to allow the coil to discharge into a parallel dump resistor before switching the coil over the amplifier. So, the relay can be a small signal type, since there will be no arcing. This is a semi-solid state version of my two relay unit described above. Galvanic isolation is "free" with the DPDT relay, so there should be no need for isolated switching. Modern relays are good for one to ten million cycles which at something like a 10 second repetition rate should be fine for observing magnetic storms (the home or school magnetic observatory version).
Wednesday, September 30, 2009
The Hollos' book arrived this afternoon (see yesterday's entry above). Perfect rainy day afternoon reading. As I said above, criticism in this amateur - semi-professional PPM business should be taken gently. Working PPM instruments, no matter how they are designed are something to be very proud of.
As many of us probably have prototyped, they have an instrumentation amplifier front end. A little odd is the use of an audio transformer to couple two nearly identical IA stages (albeit hats off the relatively low parts count). They use the first IA to be able to "receive" a balanced signal (the two center common counter-wound sensor coils) via a relatively long (50 foot) twisted shielded cable. Then they couple the single ended output of the first IA via a single ended audio transformer to a center tapped secondary to go balanced into a second nearly identical IA.
I am leaning towards the first amplifier(s) out near the counter-wound coils (still not sure if I want to go series connected or parallel connected as discussed above). Then, I like the idea of an audio transformer, but to go from the first amplifier's single ended output using a small transformer to the long twisted shielded run to an IA in the lab to receive the balanced signal from the cable. (And/or should the transformer be on the receiving side in the lab, more to think about.) Hopefully the transformer can directly drive the twisted shielded cable, or active drivers might be needed). It seems to me that it might be good to start with a larger signal to send over the cable? Also, since the counter-wound coils are by definition "isolated" (except for capactive pick up; more on this later) it should be fine to use a single ended amplifier (perhaps a single supply Op Amp(s) or a single supply with an offset ground) at the coils. Another very important aspect to the now audio transformer isolated front end is that it can be run on a separate power source, parhaps batteries near or at the base of a pole supporting the counter-wound coils (or identical coils using the "folded solenoid model" if they are in wired series should be fine too). And, just as in high-end instrumentation and audio applications, the ground-break provided by an audio transformer seems highly desirable for breaking potential ground-loops, perhaps further breaking small ground currents made via stray system capacitances to the sensors and any remote located front end circuitry.
Or, maybe one IA such as the INA166 (one of the IAs we have been using) in the lab is sufficient (without any intervening transformers) since as mentioned above, the sensor coil is by definition a galvanically isolated source. This assumes a twisted shielded cable, probably with the shield grounded only at one side (at the sensor side or the lab side), perhaps at the sensor side for parallel counter-wound coils.
Another point of departure from both the Hollos' and Koehler's designs is with regard to relays. The Hollos' say relays will fail and are not suitable, however, that is how PPMs were originally made in the 60's and 70's. At the low duty cycle use in a PPM, modern relays are fine from 100,000 to 20 million cycles depending on the model. And, while Koehler's various FET boards are most elegant (it is worth the time to go through his various FET board posted schematics with colored pencils), they are relatively complex in parts count, and once he gave up the natural galvanic isolation of relay switching, he added opto-couplers to gain back the lost isolation.
I think a nice compromise might be use a hybrid approach so that the power switching and coil energy dump into a resistor is completed before switching the coil from a polarizing mode to the amplifier for observing the precession signal. This is a departure from the Wadsworth switch, and other older PPM designs where one relay broke power (the arcing contacts that the Hollos' book refers) and then connected the amplifier. Even with FET switching with a dump resistor, there will be a momentary large voltage across the FET. Presumably this transient can be dealt with using traditional RC snubber design (as is also appropriate for relay contact circuits, but here the goal is that the coil "flyback" energy is completely dissipated before switching to the amplifier using a signal only relay, as opposed to the traditional power relay). In most of my current scribbles, the coil is on the wiper of DPDT contacts, swinging (after full discharge) between the power circuit and the amplifier input. Since I have been leaning towards relatively low polarization currents (in the 0.5 Amp to 2 Amp range) and small coils (~125 ml bottle in 2" PVC pipes), standard resonating capacitors on the amplifier front end will also probably be needed (I have used them in all of my field tests to date).
hmm, if the first amplifier(s) are out at the coil(s), then I suppose the polarize/measure signal relay needs to be out there too ... ahh, more to think about.
So, just as in the Willy Bayot/Koehler et. al. Mark II/III unit there will need to be some rudimentary time line, albeit somewhat less complicated. Something along the lines of polarize (FET on, signal relay "on" connecting the power circuit), polarize FET off, pause for coil (one or both, not sure yet) millijoules to dump into a small resistor (I have been using 402 ohms in field testing with the dual relay system), then finally switch the coil(s) over to the amplifier when only the proton precession signal is present (signal relay coil "off").
So far I have done about the same bandpass filtering as the Mark II/III approach, but more than the Hollos' approach. Second order filters seem sufficient (I make mine tunable with a trim pot). To date, I have used all capacitive interstage coupling for both high and low pass filtering in addition to one or two second order active filter stages (no transformers to date). More digital filtering before some sort of frequency determining routine appears desirable. Years back (2002/2003) I was using the "Cool Edit" software to remove ambient noise signals by using a sample taken during dead time. It was surprisingly effective for revealing the decaying sinusoid from a noisy sample. Probably this time digital filtering followed by some sort of high resolution algorithm will be sufficient.
Saturday, October 3, 2009
With the first nice day after almost two weeks of rain, I put up a quick stand for proton magnetometer coil testing. It is constructed from four 2 x 2 x 8 foot pieces of pressure treated wood from Lowes. The six foot steel bar in the background (a really useful Lowes product) was used to make about 2 1/2 foot holes. The stand is about five feet high, the frame made from the trimmings. The parts are all held together by 3/8" oak dowels, so no metal was used in the frame. It was quick and dirty construction (as usual, finished at sun down), I didn't glue anything or worry about rough file marks after I cut the end from each dowel with a hack saw. I might throw on a coat of stain / sealer if the weather stays nice for a day or two.
The plan is to place the coils under test in the plastic storage container that drops into the top. Later, if needed, I can affix a flat table to the top frame. The frame is pointed to magnetic north and inclined at about 30 degrees above horizontal for our 69 degree inclination here in upstate, NY. Not sure yet how to run the test cables for best water resistance, yet to also be able to pull the cables back to mow the lawn or rake the leaves. I do not really want to invest yet in water proof connectors, but that would be the best solution.
Sunday, October 4, 2009
I found a reasonably priced 6 pin water resistant connector from the auto racing community that might work well. This should give me the option of connecting the coils directly back to the lab or to an intermediate box at the base of the stand. Also, it should allow for quick removal of the cable to pull it back during leaf raking.
Sunday, October 11, 2009
The outdoor test stand wired to the lab is now working. First testing is with paralleled counter-wound coils. The ambient noise signals are still relatively high with spurs above and below my working frequency, today around 2292.5 Hz. Single coil testing appears impossible at this site. My test setup noise floor (post amplifier) today was about 1mV and signals ranged from 30 to 60+ mV (measured with LeCroy FFT and hp 3581A wave analyzer, 3 Hz resolution BW) depending the fluids that were being tested. The plan is to take more careful data for a range of fluids and plot the results here. Interestingly, rubber cement which looked so good in the first trial, was marginal today. Maybe it is drying or air has changed its composition (although the plastic bottle can be squeezed and does not look dry). Distilled water is still very attractive as to both amplitude and duration of the signal. Alcohol is looking very promising, with far shorted polarizing times, however the decay time is shorter too. I forgot to try windshield washer fluid. Also, on the to do list is paraffin wax.
For the present tests (amplitude and polarized un-polarize times) I am using a Hewlett Packard 3581 wave analyzer with a 3 Hz resolution bandwidth, zero sweep, and minimum averaging. The Y output is then displayed on the LeCroy scope at 1 second per division with the measurement functions used to characterize the discharge curves (decay from the most polarized state). I almost regretted buying and calibrating the old boat anchor, but today it paid for itself. First I used the FFT routine on the LeCroy (LT344L) to find the frequency, then setting the 3581 to the frequency for today's magnetic field (car in the garage, etc), it worked magnificently to observe the polarize and un-polarize times. I had to readjust the frequency often, so I will let it run over night. If that doesn't stabilize it, it has an external local oscillator input.
In the present test setup there are still batteries out at the test stand. Hopefully remote power, very well filtered at the test stand, can replace the batteries for long term winter R&D and data taking.
Monday, October 12, 2009
The test set was reconfigured today for series operation of the two counter-wound coils. Only one coil was used to hold a 125 ml fluid bottle and only that coil was energized (<1 Amp) for about 3 seconds. The INA166 amplifier was used with both coils across the input (the double relay system described above switched only one of the coils between power (polarization current) and sensing at the amplifier input.
(Note: See the November 25th entry for later fluid testing data)
Several fluids from alcohol and water to acetone were tested for amplitude and decay time (fall time from 80% to 20% was used). The results were plotted as fall time versus amplitude. The raw data is in an Excel sheet (notice the tabs at the bottom of the sheet).
The test setup was dual relay switching as above, an INA 166 amplifier (out next to the test stand) and following stages including a second order filter (Gain ~41,000), twisted shielded to the lab, a common mode filter (choke and capacitor), followed by a PARC preamp set to 100 Hz LP, 10 kHz HP, and Gain 50. The PARC preamp received a single ended signal. The output of the PARC 113 was fed into a Hewlett-Packard 3581A Wave Analyzer. The LO input for the analyzer came from an Agilent 33120A signal generator, the 33120A using a 10811A oven oscillator in a hp 5334A counter as an external 10 MHz frequency reference. The 5334A counter monitored the 3581A tracking oscillator output (2293 Hz). The Y output from the 3581A was input to a LeCroy LT344L scope with measure functions enabled. The noise floor (no low field NMR (LFNMR) precession signal) was around 15 mV to 40 mV on the 3581A display (3 Hz resolution bandwidth) and shown on a broadband scope picture.
Here is an example of the LFNMR signal for distilled water (time domain and fft on the LeCroy) and for acetone (LeCroy display of hp 3581A Y output). There was a small switching transient present with no fluid in the coil.
So far, acetone gives the longest precession signal. Water gives a robust result in both amplitude and decay time and might be easiest to use in warm environments. Rubber cement possibly offers a nice balance in decay time and amplitude. Surprising was the finding that charcoal filtered water from my tap worked fine (not shown in the results).
Even with balanced canceling coils, nearby power line spurs pic1, pic2, pic3, pic4, are potentially problematic, although perhaps far enough away for my field. The 2293 Hz signal in pic2 is the LFNMR signal. These are all the same fft taken near a precession maximum at 2293 Hz with different labels on the nearby peaks using the cursors. Notice that 2339 Hz - 2219 Hz is 120 Hz as expected. Since such spurs are systematic and line syncd, it should be possible to remove them by some sort of fft/Ifft or equivalent process. For this technique, digitization would start with a line syncd trigger.
Thursday, October 15, 2009
Using an Agilent 34410A as a digitizer, the output of the PARC 113 preamp (Gain 20) was digitized for signal processing in LabView 2009. The test stand setup was as described above with rain x De-Icer windshield wiper fluid in the sensor (series connected counter-wound coils). (The rain x De-Icer is the only windshield washer fluid I have found that does not smear the windows on the highway below -10F.) The signal was 2048 points long and the sample rate was about 5012 Hz. An equi-ripple bandpass filter with 256 taps was used (passband 2288 to 2300 Hz, stop band 2219 Hz to 2339 Hz). Frequency was determined using a Buneman Frequency Estimator, a built in LabView Signal processing function. Here is an example of the test setup output display (the filtered waveform is plotted against number of sample, the interval was about .4 second). The nearby 60 harmonics are actually 2220 Hz (37th), 2280 Hz (38th), and 2340 Hz (39th). As expected from early tests described above, the car at roughly 30 feet from the coil sensors (in garage) to roughly 70 feet (in driveway) slews the field in the backyard. These Excel sheets (in garage, out of garage) show the proximity of our local field signal to the 60 Hz harmonic spurs.
So, for now, fortunately at my location, I have sufficient spacing to work within the harmonics. Notch filters are not an ideal solution since if the field were to pass through a notch filter, the instrument would not indicate the field. Probably what is needed is active cancellation based on data taken without a precession signal. That way only the actual 60 Hz harmonic spur is subtracted out, leaving the ability to record and measure an actual precession signal as it passes through (or sits over) a power line harmonic. Probably some sort of adaptive filter would be helpful, although, the frequencies of the nearby spurs are well known, so perhaps fixed bandpass filters dedicated to nearby spur measurement (spurs measured during a dwell time) and cancellation would work.
Friday, October 16, 2009
The test stand / very rough early prototype / proof of principle setup is working well. Here is a pdf of some data points taken 30 seconds apart with a 2 second polarization cycle. I was testing other fluids (acetone and charcoal lighter fluid), however most of the data was taken with RainX DeIcer fluid (the best amplitude / decay time combination found so far). The short spikes and shifts are largely due to passing cars or cars moving in and out of driveways. Some of the spikes are from "runt" pulses. I am still on the dual relay circuit, so it could simply be bad contact closures on relatively old relays. I need to buy some new RainX (that bottle was quite old) and perhaps it is worth trying some antifreeze too. As discussed above, I plan to try an FET and no-current relay switching scheme, but one thing at a time. Also, today's timing was Rube-Goldberg at best, still the dual relay contact timing to dump coil energy (only one coil is powered), with a roughly 30 second software cycle. Within the 30 second software cycle (via a LabView 2009 vi) a hp 5359A Time synthesizer (events counts on a 1 kHz clock) provided the polarization pulse delay and width timing.
Saturday, October 17, 2009
Small improvements, one step at a time. Today I ran a 10 conductor (#22) shielded cable out to the test stand. Now I can run the polarization current from a small supply in the lab. I am using an old but trusty Kepco linear power supply running at about 6V and using sense leads from the test stand, to hold a relatively steady 6 Volts out there (it runs just over 8V in the lab during a polarization cycle). The polarization current is around 900 mA (under 1A) and currently running 2 seconds on, every 30 seconds. Here is a pdf of about 3 hours of sample data taken this afternoon.
This PPM run was made with RainX DeIcer windshield washer fluid. Here is a snapshot of a test setup to explore precession decay of various fluids using a Tek 2440 to display the amplitude of the precession decay at 2293 Hz as measured with a hp 3581A Wave Analyzer. The waveform shown on the 2440 CRT is the average responding rms amplitude of the AC voltage at 2293 Hz, a display of one half of the envelope of the exponentially decaying sine wave. The 3581A local oscillator (LO) frequency of 1.02293 MHz was provided by an Agilent 33120A locked to a hp 10811 oven oscillator in a hp 5334A counter. Note that the 3581A test setup is not part of the working magnetometer prototype, just one of the diagnostic tools at hand for R&D.
The Kepco PCX 15 1.5 MAT in a CA3 housing is providing the polarization current (0.9A at 6V 2 seconds on, every 30 seconds). Hats-off to Kepco for providing older equipment operator manuals free of charge, a very helpful service to small businesses and amateur scientists alike.
Sunday, October 18, 2009
The test stand ran for just under four hours this evening at just under 1A, 2 seconds on every 30 seconds. The chart was set to a vertical resolution of 1 gamma (1 NanoTesla). Some of the spikes were likely caused by passing cars. The rectangular change is most likely a neighbor's car in or out of the driveway.
While I remain convinced there is nothing wrong with full or partial relay switching, the test stand relays are very old and have been through many experiments over the years (some not so kind to the contacts). Some of the noise in the chart is likely due to erratic contact operation.
Probably the next step is to move towards FET switching of the polarization current (still a single coil of the counter-wound pair) with an energy dump into a resistor, followed by no-current switching into the amplifier front end. At some point I need to look at amplifier biasing to avoid a transient as the coil is switched in. Such techniques are well known in charge injection situations, such as when switching inputs to analog integrators.
Thursday, October 22, 2009
I tried out an FET / relay hybrid prototype / proof of principle today. To extend the life of the contacts, the relay only changes state when there is no polarization current. The FETs in the rough prototype are controlled by two hp 5359A time synthesizers running on a 1 kHz clock (events input) giving 1 mS timing resolution. The relay FET was set to operate 50 mS before the polarization FET was turned on and then to wait 50 mS until after the polarization FET was turned off. A scope screen print shows the timing sequence with a trace showing the decay envelope of the proton precession signal. The digitization sequence was delayed by about 400 mS from the end of the polarization current and lasts for about 400 mS. While there are no longer any "runt" pulses, there is still some variation of the leading edge of the precession signal. This needs further investigation and might still be related to amplifier biasing. In the mean time, the 400 mS delay places digitiztion into a cycle-cycle stable prortion of the precession signal which lasts well past the the digitization window. The plot of field versus time (one point every 30 seconds, 2 second polarization time, ~1A polarization current) appears to be somewhat less noisy now, perhaps old relay contacts did contribute to measurement scatter in the first setup. The multistage amplifier / first filter unit (represented by the amplifier symbol) is the INA166 amplifier discussed above. With the decrease in noise, the PARC 113 amplifier in the lab (still being used as a single ended input) was reduced from gain 20 to gain 10.
Data was taken tonight with RainX (-25F), RainX 2-in-1 (-25F), and Prestone DeIcer (-35F) (all low temperature windshield washer fluids). While I need to plot the results, all three were very comparable.
Saturday, October 24, 2009
Those inexpensive Lowes plastic storage containers turn out to be other than water resistant. The nice tight fitting edges are misleading as the depressed area around the two handle pivots make two good holes for water to drain from the depressed handle area into the bin. I should have noticed them. The bottom sections, on the other hand, are quite water tight. My INA166 amplifier and the FET/relay/terminal board sat in a few inches of water over night. The FET/relay/terminal board might be reused, the amplifier is probably gone. I washed them in distilled water and baked both, but there was just too much corrsion from electrolysis caused by the power supply rails. Looks like I need to build another one.
The single ended input prototype appears to be too noisy to get a good frequency measurement. It might also be that the INA166 prototype is working better because it has an additional ferrite core common mode filter on the input for RF rejection. The precession signal is clearly present with the single ended model, however even with the 256 tap digital filter, the performance is poor. This is surprising, I would not have expected such a difference in performance (unless I made some sort of error with shielding).
Regardless, the prototype parts have served their purpose. It is clear now that a kit for a backyard proton precession magnetometer is very doable. I need to start thinking about how to do a cost effective ADC board for the lab side to replace the Agilent 34410A as the digitizer. Also, I wonder if there is an easy way to input waveform data from a custom digitizer into LabView or if the software should be done on a micro (probably a micro demo board) or on a PC using custom software or a LabView alternative.
Friday, November 6, 2009
Work has slowed down a bit as I tackle some other jobs. I was thinking about the front end amplifier this evening and took another look at the Analog Devices website. I am not sure how I missed it before; they seem to have an ideal front end amplifier, the SSM2019. I will try it out, probably in the next front end prototype, and post the results here. It is very good to find a suitable low noise front end amplifier in a DIP package! I prefer, where practical to stay with through hole construction. It so much easier for most amateur scientists and older hobbyists. Also, I am looking into front end common mode filter designs to minimize RF pickup. This is a different problem than the common choke at the end of the longer lead back to the lab bench, more later.
Monday, November 23, 2009
Back to work at the test setup now using the new ADI SSM2019 amplifier front end and taking some data with Prestone deicer today. More data to follow on the 2019 amplifier strip, initial results are excellent. Probably next is to test out some relays that look promising for the production experiment, which probably means building a new FET - Relay hybrid prototype.
Do not be put off by the stack of commercial gear in the test stand, eventually the project should end up on only two or three pc cards. Note the 3581A, 33120A, and 2440 are just diagnostic tools used to look at the precession decay envelope during R&D and fluid evaluation. These instruments are not used to determine the magnetic field in nanoTeslas. Also, the 34410A digitizer will eventually be replaced by an ADC and the two time synthesizers by a simple micro.
The LT1357 cable driver at the outdoor test stand now has a 100 ohm series resistor that drives a 100 terminating resistor in the lab at the end of the twisted shielded cable.
Probably once the test setup is little more stable, I will more carfully retest all of the fluids discussed above and publish those results. Then, it's back to designing the experiment so there can be PC cards someday.
Tuesday, November 24, 2009
The new relay FET hybrid board is in service. This is the technique described earlier, where the relay switches one of the counter-wound coils between polarizing current and the amplifier, but only switches at zero coil current. That is, with timing, first, the relay switches over to the FET power circuit, second, after a relay contact swing delay, the FET power circuit is energized, then the FET power circuit is de-energized and the energy bleeds off through a 402 ohm dump resistor, after which (at zero coil current) the relay switches over to the center-tapped coil configuration with the counter-wound coil and the digitizer takes data for around 400 milliseconds near the peak proton precession signal.
Note, that the small signal relay cannot directly switch the polarizing current. Presumably fouled timing that causes the relay to switch with coil current on would damage or destroy the relay contacts.
Well, the new small signal relay is fantastic. Most of the odd variation in the precession envelope rise time is gone and the variation of peak envelope voltage is a tiny fraction of what it was with the previous prototype FET-Relay board.
I dropped the local 8 V regulator and the coil FET circuit runs on rail voltage now, typically 12 V. The coil current is clamping around 1.2 A, but that might be as simple as a current limit set inside the polarizing power supply (need to study the old Kepco manual).
Once I get a few hours on the new FET-Relay board, I am anxious to repeat the fluid testing data with a far more stable system.
QUESTIONS/COMMENTS/notice of typos, etc. send email to joegeller @ gellerlabs dot com
COPYRIGHT © 2009 JOSEPH M. GELLER, All rights reserved.