Journal notes, New prototype, new minor problems
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)
Friday, April 16, 2010
Still in the process of putting the prototype back in operation. Tacked everthing together to test the new polarization board (the hybrid relay-FET scheme, with a relay socket, re-built in a shielded box). All seems well: PDF, TXT, USGS PDF, sample spectra: PDF.
Adding a common mode filter between the amplifier output and the differential input to the USB 6008 ADC reduced the noise on the precession waveform. This improvment needs to be investigated further, but makes sense, since among other reasons that is where the analog ground meets the computer common, at least indirectly via the USB connection. Also, when all is back in operation, I need to give more thought to the overall approach to grounding and commons.
Verified the switching sequence to confirm that the relay is not switched when the polarize FET is on or when there is still energy stored in the polarize coil of the counter-wound sensor coil pair. All appears okay.
Next step is to mount the modules on a small plywood sheet and to re-cable between the modules. Also, I would like to move the amplifier from batteries over to a small open frame power supply, but one thing at a time.
Saturday, April 17, 2010
Buttoned up the prototype in metal boxes this afternoon and rewired the interconnections. It is a relatively quiet magnetic day, so there is not much scatter in the data, a very good evening for initial testing. Here is the first data: PDF, TXT, USGS PDF, sample spectra PDF.
The prototype is very similar to the package that was out by the sensor all winter. The old UPS batteries are a vestige of the remote operation (once I learned how to properly charge them, the amplifier ran several weeks on this set of 7.5A batteries). I will probably substitute a small open frame linear power supply or a bench supply for continued prototype testing. The same Kepco polarization power supply used earlier is still in operation (not shown in the picture). Here is a rough snap shot of the prototype. The amplifier box has been recycled many times and has many extra holes from past prototypes. Probably the output common mode filter will be merged onto the amplifier board. The separate board is from when it served as the analog line receiver for the analog output from the amplifier in the yard near the sensor stand.
Earlier when there were long control lines to the yard (>50 feet), I used the analog outputs to drive 200 ohm resistors at the FET gates, a relatively low impedance for noise immunity. One recent change is that now the polarization control is run by the USB 6008 digital output lines (~ 8 mils max). The digital outputs drive small switching transistors (2N3904 low side driver for the Relay (~30 milliamps) and a 2N3905 high side driver for the FET 500 ohm gate resistor (for noise immunity). Also, the 5 V power for all digital switching is supplied by the USB 6008 module (the USB 6008 has a 200 mA power supply output connection). Here is a rough pdf of the present control circuit: PDF. As discussed in earlier notes, the relay is closed and allowed to settle before the FET is powered ON. Then the coil is allowed to fully discharge before the relay is opened. hmm, I think I forgot the 402 ohm dump resistor, the coil must be dumping via protection diodes, more work for tomorrow! revised PDF. (see later entries, the system ground is now at the analog and digital power supply ground, there is no longer an earth ground at the sensor counter-wound coils.)
Sunday, April 18, 2010
The overnight data run crashed: PDF. The LabView program (the "vi") was still running, but apparently something got currupted and took out the normal timing sequence. Rebooting the PC and re-starting LabView and the vi brought back normal operation: PDF. I saw this once before since I switched over to the USB 6008 digital outputs. Probably I made a programming error, will recode that section.
This is some of the "tightest" (low scatter) data seen yet. Is it just a super quiet magnetic time, or has the measurement improved? Lots to look at today.
Monday, April 19, 2010
I re-coded the parts of the main LabView vi that provide the USB 6008 digital output timing for the zero-current hybrid FET-relay board. Overnight, the system failed again: PDF. Simply stopping the program without closing LabView and restarting it resumed normal operation. I am monitoring the USB 6008 digital output lines with a Tektronix 2440 digital storage scope to see if the timing goes haywire during the fault. Ah, more minor problems. Is it the old PC (an older P4 from about 2004), the code, the USB 6008? It feels like it might be some sort of memory corruption problem (e.g. over writing code).
Several hours of run time this morning have failed so far to re-create the problem. The run still looks good as of around 3pm: PDF.
Perhaps later I will do an overnight run with one or two of the newer Windows 7 machines for comparison.
Interesting, around 4 to 5 pm a spur became the dominant signal, but not sure if this is an onset of the "problem". With the new more solid ground/common connections of the new prototype in the lab, I cast off the earth ground out at the sensor. All seems okay. I need to study the data file and look more carefully at the spectra. There was likely a ground loop, as well as ground currents between here and the sensor (>50 feet). So far the digital timing looks okay. Note that a relatively long settling time is provied after the relay closes (before the polarization current is applied), however, at coil discharge, I only wait 10 mS to open the relay. This is so the precession signal does not couple into the 402 ohm dump resistor for more time than needed.
I should mention for those of you who might be new to the project notes, that there have been months of reliable data, so the problems are only of the moment and surely will be solved. Data is presently taken about every two minutes (seemingly fine for many types of geomagnetic studies, especially given the exceptional single-shot high resolution capability of FDM method). The "auto-retry" system keeps taking at a far higher rate, presently about once per 15 seconds if a data point does have a low enough FDM figure of merit (FOM). Looking at text data of unsuccessful measurements, it looks like occassionally noise spurs are rising to an amplitude level that qualifies them to be considered as a valid measurement. With the present algorithm, such a high noise spur, regardless of the source, if measured with a sufficiently low FOM, gets plotted. That might be some of what is happening in the new failure mode (new since the electronics was moved into the lab from the sensor, the prototype was rebuilt, and the polarization board was shifted to control by the USB digital outputs as opposed to the analog outputs). Possibly, under the present conditions, the spectra is becomming very noisy at times, with one of the noise spurs falling into an acceptable measurment criteria. I have no idea at present why restarting the program could recover operation and/or affect the spectra of a given measurement that way. This present problem was not observed until the prototype was recently rebuilt...
I might try an isolation transformer between the final common mode filter output and the differential input into the USB 6008 tomorrow. So far, the system is still taking good data ( PDF ) except for a couple of relatively short periods this afternoon and this evening. I expect to find currupt data in the morning, we shall see (still on the old WinXP P4 from 2004, but that one ran fine for many months). So far, there have not been any observed glitches in the timing, my orginal guess as to what is occassionaly going very wrong.
Tuesday, April 20, 2010
The problem of the present prototype is certainly software and/or the PC. The overnight run failed as expected: PDF, TXT. Immediately after making the screen print and saving the text file, I stopped the LabView vi (but not the LabView 2009 program). The two timing signals, relay and polarize looked okay on the digital scope. When the vi was restarted, all was fine: PDF (there was a short period of low signal strength, a different and minor issue solved by changing a threshold amplitude). While there is possibly more improvement to be had in how the system is grounded and how stages are coupled together, that is not the present problem. The last prototype ran months without this problem, but such is the business of system development, just one more problem to solve.
Wednesday, April 21, 2010
When I moved the USB 6008 USB connection over to another computer, the system stopped working. I took that as an indication that the commons and grounds are still not right. Today I replaced the common mode filter between the final amplifier output and the USB 6008 input with a simple audio transformer circuit, as a "ground break". A 100 ohm resistor feeds the 100 ohm primary (matches the line driver output in the amplifier module) and a center-tapped 600 ohm secondary is loaded by two 402 ohm resistors to the center tap and connected via a three wire connect to a USB 6008 differential input. The USB 6008 input z is around 144 kilo ohms per channel. The sensor is floating in the yard (no earth ground). The digital circuits and the USB 6008 presently make earth ground in the lab at the AC connection of the polarization power supply. All is well so far: PDF . Sample spectra: s1, s2, s3, s4, s5, s6. If it continues to run okay, I should let it run overnight on this machine. Perhaps some coupling via the USB connection was causing a glitch in the USB 6008? So, lets see if runs okay over night on the desktop Win 7 machine. Funny to be watching this closely after months of reliable runs, oh well. Still looks good: PDF. We will see what it looks like in the morning.
Project Documentation (very early stages)
QUESTIONS/COMMENTS/notice of typos, etc. send email to joegeller @ gellerlabs dot com
COPYRIGHT © 2009, 2010 JOSEPH M. GELLER, All rights reserved.