INTRO TO ELF/ULF MONITORING TECHNIQUES AT YWK 07-JUL-96 UPDATED 27-JAN-97 William E. Payne, N4YWK LF beacon YWK 184.32 Grid EM74oa 2823 Oak Hills Dr Dallas, GA 30132 n4ywk@amsat.org e-mail Refer to supporting files: ELFDOC.TXT The file you are now reading. ELFMKIV.BMP Schematic of mkIV ELF/ULF reciever. ELFILTER.BMP Schematic of 60 Hz passive notch filter, optional companion mkIV reciever. 3UT-WAVD.BAS Complete ELF/ULF signal processing code. Warty and surly. Works. ELFNOTES.TXT Describes each signal file recorded, notes on equip, config. DTMF.BAS Well documented example of making a MS .WAV file. LOOP200T.BMP Diagram of compact 200 turn octoloop, made of 100-pair phone wire. LOOPTHEO.DOC Article on theory of ULF recieving loops. MS Word document. elf014b.3ut A typical source file from the Radio Shack DVM readings. elf014b.wav A WAV file generated from the above source file using 3ut-wavd.bas. LOOP: I use the largest physical loop possible, in addition to as many turns. This helps reduce interference from small near field disturbances and microphonics. My big loop is 500 ft in circumference, 12 turns, Ae = 22400 m^2, no centertap or shield. I have built a 200 turn octoloop (see LWCA article on "OCTOLOOP" or download 8loop.zip from the LW BBS for construction details) for portable use, but its not nearly as sensitive. OPTIONAL FILTER: I use a centertapped R divider to return the DC bias of the AD620, since my loop has no DC return. If your loop has a centertap, delete these. I also use a 60 Hz passive notch. The notch is a series LC with C split to keep the circuit symmetric about L whose case is grounded. I picked up some hefty 10 Hy chokes at a ham radio rally and put a some 1uF Cs around one, then added smaller Cs to bring Fr down to 60 Hz. THe passive notch allows me to put more gain in the AD620 without clipping the rails with 60 Hz. PREAMP: Very simple, very low noise, no tweeks here; its just a brick doing its job. The AD620 has .009 uV Hz 1/2 noise floor at 1 kHz. More significant, its total voltage noise from 0.1 to 10 Hz is 0.28 uV. This is about 3x what the first figure shows, because of the 1/f, or flicker noise. Most amps have a worse 1/f noise, and they do not quote their 0.1 to 10 Hz total, leading to false confidence in a low ULF noise floor. Rg sets the gain. Use as much gain as possible while reliably keeping peak output below 1v under all conditions. This will depend on your loop, Seebeck potentials, microphonics, temperature drift, local noise, etc. Calculation: Rg=49.4k/(G-1). SIGNAL PROCESSING AMP: This is where I apply DC trim, and filtering. Just do as the spirit moves ya. I like to use leaky integrators for gain stages since they have a 1/f response. One cancels the f response of the loop itself, the other cancels a software f response in my digital processor. The MAX281 gives a sharper rolloff for 60 Hz while remaining pretty flat below Fc, which I set to 10 Hz. LEAKY INTEGRATORS: My first leaky integrator has DC trim. I like to put a 20 turn pot accross the rails and then divide the wiper voltage down if necessary. That way, the pot noise is also attenuated by the same amount. Select the attenuator to allow just enough range to comfortably null any DC offset. Too little and you can't keep the output off the rails, too much and it gets very touchy to adjust the balance. 47k and 1k work nicely for my rig. You may want to add a zeroing pushbutton accross the integrator cap. Remember the DC trim adjusts the speed and direction the integrator charges, not the voltage it rests at! Trim to stabilize the output of the stage, then zero the integrator charge. MAX281: This filter is just like the manufacturers app note. I chose Fc=10Hz and k=1.158 and followed the app note design process for Bessel response. An elliptic response active filter would be nice here with the first zero at 60 Hz, or even a comb notch response. Watch for future developments here. DATA ACQUISITION: The output goes right into a Radio Shack DVM, which limits my sample rate to 1 Hz and my Nyquist limited freq response to 1/2 Hz. Use your favorite A/D. I recommend at least 12 bits of precision, 16 bit A/D is adequate. If you have a sigma-delta converter like Bill DeCarle uses, let me know how it works out! The RS DVM serial port is optically isolated. I plan to cut out the opto's and put in fiber optic transducers. 100 ft of glass fiber will isolate the DVM and loop from the PC even during lightning storms. Don't void your DVM warranty until you get it to work with the RS cable and opto's ! (see below) Be advised the side of the RS DVM serial port opto's which talk to your PC are powered off your PC serial status lines! Your PC must set its status lines accordingly or no joy. Radio Shack tech support will not know nor care why your RS DVM will not talk to your PC. ABOUT RADIO SHACK PRODUCTS AND TECH SUPPORT: I use the Radio Shack !!&^*%#&^%$! Metex software, and they act like its a big !#%**!^%$# proprietary deal to talk to the DVM. Your best ally at Tandy is a technically competent or at least cooperative store manager. If yours isn't, then find another store. I used to believe Tandy policy was to prey on ignorance, but I am now convinced the Tandy policy makers themselves PROMOTE technical ignorance. If you have trouble talking to your RS DVM using your own software and computer, drop me a line because you sure won't get any help in Fort Worthless. In Fort Worth, you can't get an answer because nobody understands your question, and the top tier tech help acts like you are trying to steal the design of a secret nuclear device. You won't be allowed to speak directly to an engineer, and your question may get garbled by the go-between. These people are really there to help Joe Sixpack hook up his VCR, nothing more. Not only that, but the RS software is buggy. The RS apps will lose your data if you trust the software. When you end a session and close the file, DO NOT EXIT!! The file is not really closed! Your disk will contain hashed files! Open another dummy file and close it before exiting the RS software. Opening the dummy file seems to force the PC to actually close the previous file. Only the dummy file will be hashed. Many hours of my work went into the bit bucket before I learned this. If I were starting over, I would go to a company catering to the professional engineer and serious hobbyist, such as Fluke, Keithley, Data Translation, or anybody who makes a clone of the venerable DAS-08 PC data acquisition card. These companies have worthy tech support, but their documentation is deep enough you won't need to call them anyway. SIGNAL PROCESSING IN SOFTWARE: Now write some signal processing software that reads the receiver output voltage samples, either directly from the DVM or from the Radio Shack log files. You need to take the DC components out, and scale the data, then write a Microsoft .WAV file of the samples. The MS .WAV file is one of a set of RIFF standards by MS. Detailed example with generous comments is in DTMF.BAS, which translates a number to a string of touch-tone digits and writes them into a .WAV file. The tricky part is scaling and removing the DC drift components. It's not as simple as it first seems. I finally just wrote a differential of the samples, that is, the straight numerical difference between each sample and the previous sample. That way, the DC drift is removed and the deemphasis of one of the hardware integrators is cancelled. THe total frequency response from magnetic flux to the .wav file can be made flat from 0.001 Hz to 10 Hz. 3ut-wav.bas and 3ut-wavd.bas read RS DVM logs and make .wav files. These are my developmental programs, they are ugly and user surly. But they work. I did not write them to be marketed, but your'e welcome to distribute them freely. (and they don't hash your disk file structure!) Now the really slick way to do this is to take the samples and write a file of 16 bit signed integers during data acquisition, then you can just write the 44 bit MS .wav header on the front and be done. Also it takes about 1/10 of the disk space RS takes. Of course you will have to bypass the Temple of Ignorance in Fort Worthless. When I have a free weekend some afternoon, i'll finish this, too. RESULTS: PLEASE!! keep notes of your signal processor config, and your loop N and Ae. This is simply the number of turns and the effeective aperture. (turns x area) You will need a faster A/D than the RS DVM to see the Earth's Schumann resonance. I see a persistent narrowband 1/3 Hz signal and lots of burst events. Try to correlate solar and seismic events to your traces. Perhaps meteor or spacecraft launch and reentry plumes may be seen, or large scale automobile traffic patterns. Researchers in Texas are transmitting 0.1 Hz square waves into the earth's crust at very high power. If you can see them from 1000 mi away, they might try to hire you. 1996 UPDATE: Not much new work here due to illness and death in the family, but a few improvements. The passive notch filter works great, schematic included here. In October 1996 on a business trip to Colorado I introduced John Doran, N0ZCG, to the joys of VLF/ULF using the AD620 preamp. John had already been working with a homemade WWVB rig and was interested in LF. We scramble wound 196 turns of wire around the legs of a small table, taped it up, and hooked it into the receiver. On a couple of local expeditions, we heard distant autos passing on the road, the ticking of a watch, Omega, and sferics. John was so tickled I still haven't got the preamp back, may just make another one. I also met the experimental oil seekers and some of their professional consultants in Colorado. Their gear works pretty much like ours, no magick or secret technology, just more money. They wrestle the same problems and limitations we have to deal with. We talked technology til the Mongolian BBQ proprietor had to throw us out before closing. ULF transmitter in the works here, using 2000 watt surplus commercial amps when I get them installed. Gotta get 220 volt power run, build the exciter, and an antenna and coupler. Frequency TBD, may be near 1530 Hz. If ULF don't work, could sell the amps to somebody who likes to blow up speakers.