Digitizers for a Proton Magnetometer

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)

Tuesday, December 1, 2009

One ultra low cost digitizer looks to be a failure. I tried using the board from a Logitech desktop USB microphone as the digitizer, however, it is not stable enough. The problem is probably ADC clock jitter (aperature time variation) which then prevents reliable frequency determination to under 0.1Hz. Also, I did not investigate how many bits the ADC was using (voltage amplitude resolution). Perhaps this observation precludes the use of common sound cards as well. Fortunately low cost digitizers are becomming more common. In fact, some industrial digitizers are so cost effective, there seems to be no point to investigating costly high end sound cards. So, the next project is to identify a suitable cost effective digitizer, perhaps one with at least a couple of digital I/O lines to control the time line as well.

Interesting question now is how many bits of resoution are needed for the system to resolve to below 0.1 Hz? Clearly the 34410A works fine (need to think about how many effective bits it is using, but probably 18 or more). Can such PPM system resolution be obtained with 12 or 14 bits? Or, does it need 16 or more bits?

In reviewing digitizer options, it is interesting to note that most relatively high quality DMMs (Agilent, Keithley, etc) below the $1,100 price range, do not digitize at rates up to the needed ~5 kHz (Nyquist limit for two times the desired ~2,300 Hz PPM signal). Note that there is no advantage in FFT resolution to go to 10 kHz or higher, however without under sampling techniques, the Nyquist criterion demands at least twice the frequency.

Friday, December 4, 2009

The new National Instruments USB-6008 DAQ module arrived this morning. Initial testing with the 12 bit ADC looks very promising! The bit requirement needs more formal analysis at some point, however for now, 0.01 Hz resolution from the Agilent 33120A generator is no problem and 0.01 Hz transitions are repeatable on the nanoTesla plot (a fraction of a gamma). The digital I/O is a little less robust, but should be fine for things like LED indicators. The analog output ironically is supposed to be 5V into 50 ohms, so at first glance looks perfect for driving the cable out to the sequencing FETs. Also, output timing is software, not hardware, however that should be fine for the time line. The critical aperature time is the ADC aperature (jitter in the sample rate), and that looks to be local crystal time base controlled, probably good enough for this application (or a TCXO might be needed, hopefuly not).

Here is some test data, one data point per 30 seconds, Agilent 33120A generator, 1.5 V pk-pk (typical of an amplified precession signal), 2,300.000 Hz sine wave (external frequency reference from a hp 10811A oscillator in a hp 5334A counter). The vertical scale is about 1 nanoTesla or one gamma. Note that typical diurnal variation of the total field (the "F" scalar) is on the order of 10 to 30 nanoTesla. PDF of test data.

The ADC clock on my new unit is off by 0.003% at a 5 kHz sample rate (fs). A constant entered into the program solves that extremely minor problem. There will need to be some temperature sensitivity testing at some point, however the ADC stays in the lab (not out at the sensor or front end amplifier), so Tc might be less of an issue.

The good news is this might work well with a packaged user installation provided by the application builder (so the end user does not need to own LabView).

The USB 6008 is in a tiny plastic package, much smaller than it looks in the photos. It almost takes a jeweler's screw driver to turn the screws in the side terminal strips.

At $169 for a USB-6008, the overall price of the project is a little higher than I had hoped for, but still maybe not so bad for a proton magnetometer observatory. The software (an installable program created with NI Application Builder (not the LabView package) would be a part of the Gellerlabs kit, so maybe not so bad. Also, academic buyers get a National Instruments 10% discount for one unit and 25% if they buy five (don't bother to ask, no discount for amateur scientists and hobbyists).

So, programming continues this evening and maybe back to field testing tomorrow. It's been warm this year, yet evening temperatures are now below freezing, so the Prestone or RainX fluids should be perfect as we approach January and February when the outdoor temperature can fall to -30F here in upstate, NY.

Saturday, December 5, 2009

I noticed that in the sample plots I have been labeling field data as the "F vector". With a fluxgate magnetometer that is sensitive to orientation in the field, the magnitude of the vertical and horizontal field can be seen by so orienting the probe. Similarly, the magnitude of the total field can be seen by orienting the probe along the direction of the "F vector". The proton precession magnetometer here is oriented perpendicular to the direction of the F vector to get the best alignment and highest presession signal. The frequency of the precession signal, however is mostly independent of the orientation. Thus, what is being measured here is only the magnitude of the F Vector, or a scalar. In the future, I will try to remember to display the F Scalar value.

Also, note that while it is difficult to achieve absolute accuracy with a fluxgate magnetometer, the precession magnetometer by itself cannot record the component parts of the F scalar, eg. the standard H, Z, and E measurements. My understanding is that a precession magnetometer could make such measurements if certain components of the total field are nulled out by coils, but that is way beyond the scope of this project, which is to accurately measure and record the total field scaler of the F vector.

Sunday, December 6, 2009

The test stand is back up and running with the National Instruments USB-6008 DAQ module running in the lab. The front end amplifier out at the test stand is the ADI SSM-2019 version, and switching is the FET - Relay hybrid scheme. The analog signal from the outdoors test stand is terminated in 100 ohms and a common-mode filter and directly connected to the USB-6008 (the PAR 113 pre-amp is not being used). Polarization current is 1.5 A for 2 seconds about every 30 seconds. The working fluid is RainX low temperature winshield washer fluid. The 6008 is taking 8192 samples at a sample rate of about 6 kHz for a digitization window of about 1.4 seconds. The software FIR filter on the PC is now running 512 taps.

The present system apears to be resolving to 0.1 nT or better. Here is a pdf of field data from this afternoon and evening. As before, spikes are generally caused by passing cars, trucks, and school buses. Step changes are nearby parked cars that come and go out to over 200 feet depending on the size and ferrous content. Moving the nearest vehicle at about 35 feet from the sensor causes a shift (offset) in excess of 15 nanoTesla.

Outside temperatures are now down to 20F at night, so far with no adverse effect cause by detuning of the active second order bandpass filter on the SSM-2019 amplifier board. The tempco of that section needs to be reviewed before we got a final design of the experiment, however experimentally, it seems to be within operating limits over a 50 F range so far from about 20 F to 70 F.

I tried to set the 6008 to the +0.1 to -0.1 scale, however with the "express" VIs it is difficult to confirm what range the programmable gain amplifier is set to. I need to reprogram the 6008 operation in lower level VIs to get better direct control over the 6008 configuration. But, for now the express vis were fine for proof of principle testing. It does look like the 12 bit ADC is going to work out fine.

The NI USB-6008 has also taken over all of the time line operation by use of its two analog outputs. That sounds counter-intuitive, but at 5V with 50 ohm source impedance, they are perfect to drive the long lines out to the gate resistors on the FETs (relay and polarization control) at the test stand.

Monday, December 7, 2009

There is no 6008 +0.1 to -0.1 scale, the lowest scale turns out to be +1V / -1V, or 12 bits across 2V rail to rail. So, at about 80 mV pk-pk the prior testing was only using 8 or 9 effective bits. This evening I added back into the system the PAR 113 preamp in the lab with gain set to 10. Now, at least at the early portion of the decay envelope, there are 11 to 12 effective bits. One interesting aspect of this digitization is that as the waveform decays, the waveform spans a smaller range of bits.

Using more of the available 12 bits yields a minimum resolution of under 0.02 nanoTesla. Here is a PDF of about a half hour of data using the extra gain 10.

Will run overnight in this mode and post the results.

Earlier data from last night showed very large excursions as a garbage truck stopped at various distances along the road.

I acquired some 16 bit digitizers (not here yet) that have a 10 MHz external clock input. They might be useful to see the effect of a "perfect clock" from a GPS disciplined oscillator versus clock jitter in the digitization aperature times.

Thursday, December 10, 2009

It turned out that it was easier to use an X-Y graph in LabView than the waveform chart to add a time stamp to the x axis. At the same time, I noticed that when using only symbols to mark each data point, outlying points caused by passing vehicles are less disruptive to the plot .

Here is an overnight run from last night. Compare our magnetometer plot with the USGS overnight raw data. Note that our plot is about 12 hours and the USGS Geomag Real Time Station Display strip chart is a 24 hour plot.

Tuesday, December 15, 2009

A window was added before the FFT. Here is an overnight run with the NI USB-6008 from last night. Compare our magnetometer plot with the USGS overnight raw data.

QUESTIONS/COMMENTS/notice of typos, etc. send email to joegeller @ gellerlabs dot com

Geller Labs	Geller Labs
Products	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) Tuesday, December 1, 2009 One ultra low cost digitizer looks to be a failure. I tried using the board from a Logitech desktop USB microphone as the digitizer, however, it is not stable enough. The problem is probably ADC clock jitter (aperature time variation) which then prevents reliable frequency determination to under 0.1Hz. Also, I did not investigate how many bits the ADC was using (voltage amplitude resolution). Perhaps this observation precludes the use of common sound cards as well. Fortunately low cost digitizers are becomming more common. In fact, some industrial digitizers are so cost effective, there seems to be no point to investigating costly high end sound cards. So, the next project is to identify a suitable cost effective digitizer, perhaps one with at least a couple of digital I/O lines to control the time line as well. Interesting question now is how many bits of resoution are needed for the system to resolve to below 0.1 Hz? Clearly the 34410A works fine (need to think about how many effective bits it is using, but probably 18 or more). Can such PPM system resolution be obtained with 12 or 14 bits? Or, does it need 16 or more bits? In reviewing digitizer options, it is interesting to note that most relatively high quality DMMs (Agilent, Keithley, etc) below the $1,100 price range, do not digitize at rates up to the needed ~5 kHz (Nyquist limit for two times the desired ~2,300 Hz PPM signal). Note that there is no advantage in FFT resolution to go to 10 kHz or higher, however without under sampling techniques, the Nyquist criterion demands at least twice the frequency. Friday, December 4, 2009 The new National Instruments USB-6008 DAQ module arrived this morning. Initial testing with the 12 bit ADC looks very promising! The bit requirement needs more formal analysis at some point, however for now, 0.01 Hz resolution from the Agilent 33120A generator is no problem and 0.01 Hz transitions are repeatable on the nanoTesla plot (a fraction of a gamma). The digital I/O is a little less robust, but should be fine for things like LED indicators. The analog output ironically is supposed to be 5V into 50 ohms, so at first glance looks perfect for driving the cable out to the sequencing FETs. Also, output timing is software, not hardware, however that should be fine for the time line. The critical aperature time is the ADC aperature (jitter in the sample rate), and that looks to be local crystal time base controlled, probably good enough for this application (or a TCXO might be needed, hopefuly not). Here is some test data, one data point per 30 seconds, Agilent 33120A generator, 1.5 V pk-pk (typical of an amplified precession signal), 2,300.000 Hz sine wave (external frequency reference from a hp 10811A oscillator in a hp 5334A counter). The vertical scale is about 1 nanoTesla or one gamma. Note that typical diurnal variation of the total field (the "F" scalar) is on the order of 10 to 30 nanoTesla. PDF of test data. The ADC clock on my new unit is off by 0.003% at a 5 kHz sample rate (fs). A constant entered into the program solves that extremely minor problem. There will need to be some temperature sensitivity testing at some point, however the ADC stays in the lab (not out at the sensor or front end amplifier), so Tc might be less of an issue. The good news is this might work well with a packaged user installation provided by the application builder (so the end user does not need to own LabView). The USB 6008 is in a tiny plastic package, much smaller than it looks in the photos. It almost takes a jeweler's screw driver to turn the screws in the side terminal strips. At $169 for a USB-6008, the overall price of the project is a little higher than I had hoped for, but still maybe not so bad for a proton magnetometer observatory. The software (an installable program created with NI Application Builder (not the LabView package) would be a part of the Gellerlabs kit, so maybe not so bad. Also, academic buyers get a National Instruments 10% discount for one unit and 25% if they buy five (don't bother to ask, no discount for amateur scientists and hobbyists). So, programming continues this evening and maybe back to field testing tomorrow. It's been warm this year, yet evening temperatures are now below freezing, so the Prestone or RainX fluids should be perfect as we approach January and February when the outdoor temperature can fall to -30F here in upstate, NY. Saturday, December 5, 2009 I noticed that in the sample plots I have been labeling field data as the "F vector". With a fluxgate magnetometer that is sensitive to orientation in the field, the magnitude of the vertical and horizontal field can be seen by so orienting the probe. Similarly, the magnitude of the total field can be seen by orienting the probe along the direction of the "F vector". The proton precession magnetometer here is oriented perpendicular to the direction of the F vector to get the best alignment and highest presession signal. The frequency of the precession signal, however is mostly independent of the orientation. Thus, what is being measured here is only the magnitude of the F Vector, or a scalar. In the future, I will try to remember to display the F Scalar value. Also, note that while it is difficult to achieve absolute accuracy with a fluxgate magnetometer, the precession magnetometer by itself cannot record the component parts of the F scalar, eg. the standard H, Z, and E measurements. My understanding is that a precession magnetometer could make such measurements if certain components of the total field are nulled out by coils, but that is way beyond the scope of this project, which is to accurately measure and record the total field scaler of the F vector. Sunday, December 6, 2009 The test stand is back up and running with the National Instruments USB-6008 DAQ module running in the lab. The front end amplifier out at the test stand is the ADI SSM-2019 version, and switching is the FET - Relay hybrid scheme. The analog signal from the outdoors test stand is terminated in 100 ohms and a common-mode filter and directly connected to the USB-6008 (the PAR 113 pre-amp is not being used). Polarization current is 1.5 A for 2 seconds about every 30 seconds. The working fluid is RainX low temperature winshield washer fluid. The 6008 is taking 8192 samples at a sample rate of about 6 kHz for a digitization window of about 1.4 seconds. The software FIR filter on the PC is now running 512 taps. The present system apears to be resolving to 0.1 nT or better. Here is a pdf of field data from this afternoon and evening. As before, spikes are generally caused by passing cars, trucks, and school buses. Step changes are nearby parked cars that come and go out to over 200 feet depending on the size and ferrous content. Moving the nearest vehicle at about 35 feet from the sensor causes a shift (offset) in excess of 15 nanoTesla. Outside temperatures are now down to 20F at night, so far with no adverse effect cause by detuning of the active second order bandpass filter on the SSM-2019 amplifier board. The tempco of that section needs to be reviewed before we got a final design of the experiment, however experimentally, it seems to be within operating limits over a 50 F range so far from about 20 F to 70 F. I tried to set the 6008 to the +0.1 to -0.1 scale, however with the "express" VIs it is difficult to confirm what range the programmable gain amplifier is set to. I need to reprogram the 6008 operation in lower level VIs to get better direct control over the 6008 configuration. But, for now the express vis were fine for proof of principle testing. It does look like the 12 bit ADC is going to work out fine. The NI USB-6008 has also taken over all of the time line operation by use of its two analog outputs. That sounds counter-intuitive, but at 5V with 50 ohm source impedance, they are perfect to drive the long lines out to the gate resistors on the FETs (relay and polarization control) at the test stand. Monday, December 7, 2009 There is no 6008 +0.1 to -0.1 scale, the lowest scale turns out to be +1V / -1V, or 12 bits across 2V rail to rail. So, at about 80 mV pk-pk the prior testing was only using 8 or 9 effective bits. This evening I added back into the system the PAR 113 preamp in the lab with gain set to 10. Now, at least at the early portion of the decay envelope, there are 11 to 12 effective bits. One interesting aspect of this digitization is that as the waveform decays, the waveform spans a smaller range of bits. Using more of the available 12 bits yields a minimum resolution of under 0.02 nanoTesla. Here is a PDF of about a half hour of data using the extra gain 10. Will run overnight in this mode and post the results. Earlier data from last night showed very large excursions as a garbage truck stopped at various distances along the road. I acquired some 16 bit digitizers (not here yet) that have a 10 MHz external clock input. They might be useful to see the effect of a "perfect clock" from a GPS disciplined oscillator versus clock jitter in the digitization aperature times. Thursday, December 10, 2009 It turned out that it was easier to use an X-Y graph in LabView than the waveform chart to add a time stamp to the x axis. At the same time, I noticed that when using only symbols to mark each data point, outlying points caused by passing vehicles are less disruptive to the plot . Here is an overnight run from last night. Compare our magnetometer plot with the USGS overnight raw data. Note that our plot is about 12 hours and the USGS Geomag Real Time Station Display strip chart is a 24 hour plot. Tuesday, December 15, 2009 A window was added before the FFT. Here is an overnight run with the NI USB-6008 from last night. Compare our magnetometer plot with the USGS overnight raw data. QUESTIONS/COMMENTS/notice of typos, etc. send email to joegeller @ gellerlabs dot com COPYRIGHT © 2009 JOSEPH M. GELLER, All rights reserved.
Manuals
Tech Notes
About Geller Labs
Contacts
Links
Ordering