Mark, Today a friend here at JPL brought to my attention your recent letter requesting information on building a seismograph. As it turns out, I built the exact same project from the Scientific American article that you are asking about, and I think I can provide you with some good information. I originally wanted to build this project way back about ten years ago when the article first appeared, but I was deterred by one aspect of the project -- the cost of a recording device. More recently, since acquiring a personal computer (a Macintosh) and some expertise in programming, I realized that I could use the computer as a recorder and solve the problems both of buying a strip-chart recorder and what to do with the reams of paper that would undoubtedly be generated. My implementation of the project consists basically of three parts: the seismograph itself, the analog electronics, and the digital part. The physical seismograph itself is almost exactly like that described in the SA article. The only major difference is that instead of mounting it to a base plate as described, I bolted it directly to the cement floor of my garage using concrete anchors. (I'll describe the effect of the garage's vibration enviroment below.) Also, instead of magnetic damping, I used viscous oil damping. The analog electronics are similar to those described in the article, but considerably enhanced and modified. I am not especially expert in electronics, and at first I built a circuit almost identical to that in the article. Whether it was because of deficiencies in design or in my construction technique, the resultant circuit was EXTREMELY noisy and essentially useless. I enlisted the help of a friend who's good with electronics, and he designed an op-amp circuit, and then built and mounted it in a nice aluminum box for me! If you're interested, I can provide schematics of the design. Finally, the digital part makes use of a commercial analog-to-digital converter that is well-known in the Macintosh world. The "MacRecorder" (sold by Farallon Computing in Berkeley) is intended for sound digitizing, but is easily modified to sample DC analog signals. I did this and then wrote software for the Mac that samples the digitizer output (range 0-255) once per second and displays 15 minutes worth of data as a "strip chart" trace across the screen. The data that is scrolled off the screen remains in memory, and is written out to disk every 24 hours or on demand. Now I assume what you really want to know is how it works! Well I've has it in full operation now for about 4 months, and it works remarkably well. There are sources of environmental noise to be concerned with, but not quite as bad as you might think. Of course the whole instrument must be totally shielded from air currents, to which it is exquisitely sensitive, but once that's taken care of, most common sources of disturbance have much higher frequency than the pass band of the instrument (10-20 seconds). A person standing or walking 4-5 feet away on the concrete slab will cause a full scale deflection, because the floor is tilted by the person's weight. However, the effect falls off rapidly with distance (proportional to 1/r ?). My kids routinely play around the garage some 40 feet away with no perceptible effect. And I've never seen an effect from street traffic, which runs about 100-150 feet or so away from the garage. One thing to watch for though is that large magnetic objects moving in the vicinity can exert a significant pull on the boom. One day I thought I'd picked up a peculiar earthquake that actually turned out to be the rhythmic back-and-forth passage of the lawnmower across the grass near the garage! The worst noise problem I have seems to have to do with slow thermal expansion and contraction of the concrete floor. As it turns out, the only place in the garage where I could locate the instrument was near a south wall, where daily sunshine falls on the slab foundation a couple of feet from the seismograph inside. During the hours when the slab temperature is on the rise (typically from about 10 a.m. to 10 p.m.) the environment is very quiet, and my background noise level is very flat. During the night however, when the slab is presumably cooling, the background becomes quite noisy, with typical amplitudes of 10-20% of full scale and periods of several seconds. I have done experiments that seem to verify that this is actual mechanical noise, and not electromagnetic in origin. To the best of my knowledge right now, the only solution for the problem would be to move it to a more thermally isolated site. From my experience, I would guess that an underground basement isolated from frequent foot traffic might be ideal. Next I'll tell you a bit about the capabilities of the instrument to pick up actual earthquakes. As you're probably aware, this particular design is optimized for picking up teleseisms (distant quakes) as opposed to local events which typically have much more high frequency content (around 1 - 10 Hz). Nevertheless, the instrument has done well at picking up both distant and local events. I can usually detect magnitude 3 events up to a few hundred miles away, and magnitude 4 events anywhere in California/Nevada. Magnitude 5 events are off scale within a few hundred miles, and I have detected them as far away as Central and South America. Magnitude 6 events are usually easy to pick up from across the Pacific, and magnitude 7 or greater (like the recent Iran quake) can pretty much be detected from anywhere in the world. The more distant quakes can show some pretty interesting phenomena, as you can pick out the distinct surface wave arrivals, and the different body wave phases reflecting off and refracting through the Earth's core. In addition to earthquakes, I've also recorded a couple of nuclear weapons tests in Nevada, and some microseism-type signals that seem to correlate with storms at sea and in the local mountains. I had the instrument running this February when the magnitude 5.5 Upland earthquake hit only about 40 km away. The main shock put the seismograph off-scale for over 6 minutes, and I picked up dozens of aftershocks during the following days (it appeared to have detected all events of magnitude > 2.5). It's been quite facinating so far, and my biggest problem has been finding the time to identify all the events I pick up. I've started to get lazy and only work on the especially big or interesting ones. Anyway, I'd be glad to give you more information if you need it. If you like, I could send you some sample earthquake records I've made. Better yet, if you use a Macintosh at all, I could send you a disk with my software and some data on it. Either way, feel free to contact me. I'd like to hear about your experience in comparison with mine. In addition to my Internet address, below are my work and home addresses. I hope to hear from you soon! ---------------------------------- Mark, I was glad to get your reply so quickly! I was hoping my message would get through, since I am not too experienced with all the ins and outs of e-mail. Your strip chart recorder sounds like a good deal. Several years ago I came across a similar thing for free, but it was missing some parts and had neither ink nor paper. I decided at that point to think about going digital :-) . I don't know offhand where you can get more ink, but I'll ask around and see if I can find out. I'll just take the questions from your last message one-by-one, and give you the best answers I can. First, how about cars in the garage? Well, we don't park our cars in the garage, but we do have a trailer parked there. As far as I can tell, it shouldn't make too much of a difference if the car is parked nearby, as long as it's not constantly arriving and leaving. Each time the floor tilts (or "untilts") there will be a spike, but that effect dies out in a cycle or two, and is easily distinguished from natural events. On the subject of magnets, you're right that Edmund has a limited selection and is not always cheap. I found a magnet that closely matched the article specs at a local surplus store for about $12. I would suggest you try to find such a surplus store in your area. You might also be able to make a magnet of equivalent strength by using a pair of ceramic speaker magnets (pretty easy to find) separated by a gap. In any case, if you use a somewhat weaker magnet than described, it seems to me that the worst that can happen is you'll need to use a bit more gain in the amplifier. However, from a standpoint of signal-to-noise, you should try to scrounge up the strongest magnet you can. I'll send you a package with the circuit schematics as soon as I get a chance; probably this weekend. (Don't bother with the SASE because I'll probably use a larger envelope to include some sample seismograms and source code as well.) The circuit is not greatly different in concept from that given in the article, but it has some features particularly intended for my digital recording scheme. Since my A-to-D converter expects input voltages between 0 and +5 v., the circuit is designed to elevate the "baseline" zero level to +2.5 v., which is adjustable with a pot. Also, the circuit has two selectable gain ranges, 1000x and 10,000x. I run my setup at a gain of about 2000x. Since my computer is in the house, I also was concerned about possible noise from long cables. My solution was to do all the analog amplification right next to the seismograph, and then run the high voltage ( 2.5 volts :-) ) signal via speaker wire up to the house where all the digitizing takes place. This seems to work OK, since I can't see any noise at the resolution of the 8-bit digitizer when I test it with the pickup coil shorted out. It sounds like your PC card should be able to work pretty much the same. Next, turning to software, my program is all written in C. That's the good news (I think). The bad news is I'm not sure how much use it will be to you considering it is chock full of Macintosh-specific toolbox calls for everything from disk I/O and memory management to screen graphics and windows. Anyway, you're welcome to try to learn what you can from it, and if you like I'll send you a hard copy along with the circuit plans. Just be prepared for lots of weird-looking event-driven programming! If it's more convenient, I could e-mail this and some data files to you. I'm sort of inexperienced with these things, so I don't know for example whether there's a length limit to what you can send that way; one day's worth of data takes 86 kbytes. By the way, my file format is extremely simple; just a few bytes of header information and then one byte per second (value 0-255) for 86,400 seconds (24 hours). You'll need to provide me with some info on the UUENCODE utility since I've never used it before and don't know where to get it. There's an easy alternative to this if you have ftp access to Internet hosts. My Macintosh at work is connected to Internet and can run ftp server software that will allow you to transfer text or binary files directly from my machine to yours. If this is of interest, let me know and I'll give you instructions for reaching my machine. That's about all the info I have for you right now. I'll try to get a package in the mail to you soon including some typical earthquakes (and typical noise, too!). Best wishes, and I hope you have success in getting together the components you'll need. Talk to you later! - Greg Mark, This is just a quick note to say I got your last message, and today I dropped an envelope in the mail containing some answers, along with some circuit info, some seismograms, and instructions on ftp transfer of files. I'll expect to hear from you in a few days. Good luck! - Greg Mark, I just got your message this morning; sorry I missed it yesterday, but I was out of the office most of the day. Anyway, I will leave the ftp server running all weekend, so you should be able to get the files. Just send me a note when you're done. In fact it will be running today as well, so you don't necessarily have to wait until the evening, unless that's more convenient for you. The seismometer amplifier circuit was built on ordinary perforated board with point-to-point wiring, and is housed in an aluminum box about 2 inches on a side. I don't think there's anything *super* difficult about the construction, but then that's easy for me to say, since my friend built it :-) ! As was mentioned in the notes I sent you, I suspect that the large value capacitor used in the circuit is a fairly critical item noisewise, and probably the higher the quality one can afford here, the better. Also, I have found that the output is apparently rather sensitive to fluctuations in supply voltage. I'm just using a pair of 6 volt lantern batteries because I don't own a good power supply, but I wonder whether it might work better with a well-regulated supply. Good luck with the ftp transfer, and with the project in general. I'll be in touch with you later. - Greg Greg, Thanks for leaving the Mac on for the weekend -- I just received the last of the files I told the email server to FTP for me. When I got home yesterday (Sunday) at about noon, it had FTPed everything except four of the data files. It won't let me FTP more than five files in one day. But I quickly sent out another message, and it fetched the rest of the files last night sometime around 9 PM, I think. After taking a look at your C program, I think I may just go ahead and write my own. If I knew much about the Macintosh, it would be one thing, but I've only used them a few times. Your code will definitely help me in a few places, though. I just have one more question :-) to ask you about the seismograph itself. What type of wire did you end up using for the boom support? And how did you attach it at the turnbuckle and at the other ends? I tried using a small size of piano wire, which is nearly impossible to work with! After finally getting some knots tied in it, I found out that the wire breaks very easily around the knots... Is there maybe some sort of clamp that could be used? (String the wire around the object it is to be attached to, then clamp the end of the wire parallel to the main wire). Thanks! Mark, I'm glad you got the files ftp'ed successfully. Just let me know if you need anything else. I was afraid that the source code for my program would be pretty much greek to you, but that's life :-) ! Actually though, I'm just as unfamiliar with the PC/DOS world as you are with the Mac. I've had a number of my PC-literate friends ask me if I had IBM-compatible versions of my seismograph program, and I always gave them a blank stare. Maybe once you get yours working, I'll be able to point them in your direction for advice. One of my more entrepreneurial friends has even suggested that we ought to try to sell the things! On your question about piano wire, I had an experience basically identical to yours. The stuff is very hard to work with, and defies all attempts to bend, solder, wrap, glue, or otherwise secure it. I guess that its redeeming quality is that it doesn't stretch (much) and so it makes for a pretty stable pendulum. After some trial and error (and wasting a few pieces of wire), I got acceptable results by hand-twisting the wire ends. The result doesn't look especially pretty, but it works as long as you don't try to bend the wire on too small of a radius. All I can really suggest is to have patience and practice a few times. I also found that it paid to wear gloves (or be careful) while working with the wire, since it's very easy to poke youself with it. Well, best of luck and keep me posted on your progress ! - Greg Greg, Yesterday I received a catalog that I had ordered from the Magnet Sales and Manufacturing Company in Culver City, CA. I found a page with several Alnico Horseshoe magnets, found one that looked about the right size, then called up the company to find out what the price was. They told me $45.39 ! Oh well -- guess it's time to do some more combing of the surplus outfits around here :-) I'm going to get some more piano wire today to experiment with. The type I was using was smaller than 26 gauge -- looked more like 28 or 29. I'm worried that it may not stand up to the strain no matter how I attach it, so I am also going to try and get some that will fit the other oil burner nozzle that I have. It is a larger-than-average nozzle that easily takes 24 gauge wire, so I know that it won't break too easily. Of course, I don't know what effect a larger size of wire will have on the seismograph's operation. I've also run into another snag... I drilled a hole near the lead weight on the boom for the piano wire to go through, but no matter how much I try, it still isn't alligned properly with the verticle knife-edge. Because of this, when I let the boom swing freely, the knife-edge swings around a bit and forms an angle of 20 to 30 degrees from the vertical. How did you get around this problem? Maybe a drill press would work better than my hand drill for getting things lined up the right way. Thanks, Mark, It sounds like a lot of the experiences and problems you've been having are very similar to mine. In most of my past messages, I haven't really gone into extensive detail about the mechanical construction because I didn't want to overwhelm you with a lot of trivia. However, I realize that some of those details might be helpful, so I'll elaborate a bit more. Fairly early in the project, I adopted a philosophy (out of necessity) that nothing much in the Scientific American design was really sacred, as long as the final result worked similarly. As a result, I decided not to spend a lot of time trying to track down odd things like oil burner nozzles and the like (oh great, *now* he tells me :-) ). I felt pretty free to make substitutions and design changes at my own discretion. Instead of the oil burner nozzle, I used some kind of brass bushing with the right hole diameter that I found in a junk bin at the surplus store where I found the magnet. (By the way, if you *really* get desperate in the magnet search, I could go shopping for you at my source, C&H Sales in Pasadena; they don't always have what you need the first time you look, but they have a high rate of turnover in all kinds of electronic and optical junk.) When I went shopping for parts, I couldn't find a local hardware store that carried lead ingots, so as a substitute I used an old discarded disc brake rotor that was lying around the garage. I attached it horizontally to the underside of the boom with a pair of eyebolts, and because its center of gravity is well below the wire attachment point on the boom, it's no problem keeping the boom from rotating about its long axis. Maybe a similar lowering of the c.g. of your weight will cure your rotation problem. That brings me to another modification I made. When I first tried the vertical knife-edge design suggested in the article, the knife edge kept slipping off the bolt head, causing the whole pendulum to collapse in a heap. So instead, I sharpened the boom end to a point (rather than an edge) and made it rest against a cup-shaped surface to keep it from slipping down. The cup surface was obtained by super-gluing an old plano-concave lens to the flat bolthead surface. As you can see, while my seismograph is closely patterned after the one described in the article, I took a number of liberties to get it to work for me. After all, the only thing you're *really* after is a pendulum that has the correct period and the right amount of damping. Beyond that, the details are fairly irrelevant. So I wouldn't lose too much sleep over exactly what gauge of wire to use, etc. since any problem usually has many solutions. My approach was just to try different things until something finally worked. The result isn't always elegant looking (I get lots of puzzled expressions over that brake rotor!) but it does the job. Even though I've given you more details here than in my previous messages, I've probably still left out some things you might be wondering about, so feel free to quiz me about anything else. I wished that I had had someone to find out the ways *not* to do things before I started on mine. Good luck and I'll look forward to your next message. - Greg Greg, I finally found a magnet! :-) I bought it from... would you believe: the piano repair store! The guy who sold me the piano wire uses a couple of big magnets mounted on broom sticks to "vacuum" the floor of his workroom. I was mentioning that I needed one for the project (he was asking me what the piano wire was for) and ended up buying one from him for $10. The one I bought measures about 6" by 7", with a gap of only 11/16"". It is also quite heavy. I plan to change the design by placing the magnet stationary and mounting the coil to the boom. I have also finished construction of the coil form. I decided to use a coil form similar to the one described in the article, but thinner, with 4" diameter plastic sides to compensate for the smaller magnet gap (the coil can swing into the gap all the way, though, as it is quite deep). The finished coil form now slides into the magnet with a clearance of about 1/16" on each side :-) I better make sure it gets alligned properly before leaving it in operation... Regarding the use of different weights at the end of the boom, I ended up using a fairly symmetrical lead weight similar to the one in the article, except that mine has a hexagonal shape. Since it is about 3" thick, my hand drill didn't like drilling through it too well -- and ended up taking quite a long time to get through the weight without binding. I'm afraid lowering the center of gravity just may not be practical in my case. I'll keep investigating other options, I guess. Next thing to do is wind the coil (what fun! ...?) ; I need to get out my cheap-pocket-calculator-turns-counter to help me keep track of the number of turns... Thanks very much for your helpful suggestions -- talk to you later, Mark, It sounds like you're making good progress. I got your mail just fine via the new mailer; it got to me in about 3 hours. Your plan for reversing the coil and magnet sounds fine. I almost did the same thing. I would just be careful that the leads coming from the coil don't interfere with the motion of the pendulum. I don't expect it should cause any difficulties though. By the way, I devised a fairly painless way of winding the coil. I mounted the coil form in the chuck of an electric drill that was held to the table in a vise. I controlled the speed of the electric drill with a foot pedal-operated rheostat (borrowed from my wife's sewing machine). It worked great and I got the whole coil wound very neatly in less than an hour. I didn't actually count the turns, but just figured out how many feet of wire would give me the desired number. Good luck and keep me posted! - Greg Mark, Sorry it has taken me a few days to get back to you. It sounds like things are going well on your end. In answer to your question about the circuit, yes, I do have terminals 5 and 6 jumpered together. As I understand it, this would only be changed if there was a need to change the +2.5 v offset level to something else. I have not ever removed the jumper and have been able to take care of all necessary level adjustments using the offset potentiometer. By the way, starting tomorrow I will be away on vacation for 2 weeks, so I won't be reading or responding to mail during that time. I'll be back in touch again on Monday the 20th. Talk to you later! - Greg Thanks, I had a good vacation, and I'm now trying to get back into my routine again. It sounds like things are going well with you; good luck with the start of school. I don't have too much news to report. I was looking forward to going over two weeks' worth of seismograph data when I got home, only to learn that the power to our house went out momentarily just a couple days after we left. I guess Murphy doesn't go on vacation! I have one bit of trivia to ask you about. If you got this message successfully, then you will have noticed that it is from a new Internet address. I am experimenting with a new system that will allow me to be notified of arriving mail directly at my desk, without having to log onto a remote host somewhere. I would like to know if this message arrives intact to you. Let me know if it includes any extraneous line breaks or other quirks. Also, I'd like to verify that you are able to reply to me at this new address. First try the address gregl@freia.jpl.nasa.gov, and if that isn't recognizable, you can try the physical address gregl@128.149.70.6 . Thanks, and I'll talk to you soon. - Greg --- Mark, Thanks for replying so quickly and serving as a "guinea pig" for my e-mail test. Your reply came through clean and only took about 45 minutes to arrive. In answer to your questions, my oil damping paddle is oriented parallel to the boom so that it moves broadside through the oil for maximum drag. Again, whatever you get to work for you is best. I covered my seismograph with just a rather crude plywood enclosure that covers it on two sides. The other two sides and top are shielded by the garage walls and an overhanging bench countertop. It works, but it's ugly and I think the Scientific American article suggestion to make a see-through plastic enclosure might be kind of neat (if not too expensive!). Thanks for the concern about Magellan. My colleagues here who are connected with the project are pretty optimistic that they'll get through this "childhood illness" stage and get things smoothed out. After all, you may remember that the Voyagers went through some nasty problems before the mission folks figured out all their quirks. Stay tuned! I'll talk with you later. - Greg --- Mark, It's good to hear from you again. I know how the school year can swallow one up! I particularly understand because I just started a new job this fall as a college physics professor. I'm still working part time at JPL (this e-mail address is still good) and as a result I've been *very* busy the last couple of months. I'm glad to hear that the project is progressing. It sounds like you'll be up and running pretty soon. Yes, I did pick up the Mono Lake earthquake; it was very large and registered full-scale on my instrument. Let me know if you're interested in a xerox copy of the record and I can mail it to you. I haven't made any changes to my setup lately, but I'm going to get a chance to apply some of my experience in my new job. The school has a seismology lab that we would like to convert to digital recording, so I'll be doing some of the same things there that I've been doing at home as a hobby! I'll be anxiously awaiting word on how things work out for you. Good luck and enjoy the upcoming holidays! - Greg --- Mark: Sorry for the delay in replying! I got both of your messages, but I've been a bit swamped in work lately, and have been slow in replying to my mail. Sounds like good news with your setup. Let me know when the bits start flowing! My setup has been running fairly smoothly. Yes, I picked up the Carson City area events (at least the bigger ones) as well as a number of interesting more distant events. I independently discovered the seismology newsgroup you referred to recently; it looks like a good source of info on California events. I can't remember now if I told you before, but for worldwide events, the USGS maintains a free BBS with current epicenters going back a few weeks. The number (toll free :-) ) is (800)358-2663 . I happened to notice your name on a list of respondents to a newsgroup poll on whether to form a geology forum. It's a small world, isn't it! The only problems of note that I've had recently have to do with electronics. I now know why my electronics textbook rates electrolytic capacitors as having *terrible* characteristics. Over the year my setup has been operating, the large-value electrolytics we used in the construction of the power supply and filter parts of the circuit have gradually degraded. By this, I mean that they randomly "leak over" doing horrible violence to the output of the electronics. At it's worst point, this leakage rendered useless up to 10-20% of a day's data. I am currently planning to search for some replacement components, although most alternatives to high-value electrolytics that I'm aware of tend to be rather *expensive*. I'll let you know what I find out/decide. That's all for now. Thanks for the update. - Greg Greg, My seismograph is finally up and running! I have written some preliminary C software which displays the trace across the screen with one second sampling intervals and saves data to the disk. The remaining task, of course, is to calibrate and adjust everything properly. The only hitch is that my large magnet has more of an effect on the boom than anticipated. Because of this, I've been forced to keep the magnet as far away as possible, and the sense coil barely comes into the magnet's gap in the rest position. What would happen if I put the magnet closer, then adjusted the position of the knife-edge pivot point so that the boom came to rest in the center? I haven't done this yet because I think it would cause other problems and mess up the instruments response. I should give you some numbers on my adjustments so far. The natural swing period of the boom was adjusted to 13 seconds and the boom was centered approximately, then the damping was put in place. I've adjusted the damping as per the instructions so that the boom overshoots 2mm after being displaced 1cm. (One of my main concerns about the magnet, by the way, is that it could create a sort of "unsymmetrical damping," by attracting the boom in one direction only). I adjusted the amplifier offset control for a center voltage of ~1.25V, since my ADC has a 0-2.5V input. I've experimented with different gain settings; right now the gain switch is in the 10,000X position with the attenuation control adjusted to give a gain of a few thousand. I would greatly appreciate hearing of your experiences in the calibration and adjustment phase of the project. In particular, how did you adjust your damping? Also, what is the purpose of the "bias" knob on the amplifier and where do you have it set? Finally, how often, in your experience, have you been able to detect an "obvious" earthquake, on average? Thanks for your help and advice! I'm looking forward to being able to concentrate 100% on the software aspects of this rather than fooling around with the hardware :-) Mark: Wonderful news about your instrument! You didn't have to wait as long as I did for my first "significant" event. Although I seem to pick up an average of about one teleseism (distant quake) per week or so, really remarkable ones like the Bering Sea event are somewhat rarer. Local earthquakes in the magnitude range of 2.5 - 3.5 seem to come in at about a couple-per-week rate, although it's highly variable (sometimes several in one day, other times nothing for weeks). I happened to be watching the computer when this one came in, so it was kind of exciting. My recording sounds basically similar to yours; the P-wave was about 20% of full scale, the S-wave was quite sharp and ~30% of full scale (actually I think it was a superposition of direct S and PcS), and the surface waves were at times over half full scale. I'll try to answer your questions roughly in order. I don't have your problem with the magnet, since I have the magnet on the boom and the coil on the floor. As a result, the boom doesn't get pulled by the magnet, but on the other hand, I have to be careful about moving magnetic objects (like wrenches or lawnmowers) near the boom. The only thing you'll gain by putting the magnet closer is a higher raw voltage from the coil. If the gain and noise you're currently getting from the electronics are OK, then there's not much reason to change. (Your gain setting sounds about the same or slightly higher than mine.) If however you do want to boost the pre-amplified signal, you can try moving the magnet closer. I'd just be a little worried about the asymmetric restoring force problem you alluded to. (In theory, this would just mean your pendulum's not a "perfect" harmonic oscillator, but the motions we're interested in are so small that I doubt there would be any detectable effect.) Another point to keep in mind when positioning the magnet is this: To get maximum output, you don't necessarily want the coil where the magnetic field is maximum, but rather where it is *changing* most rapidly. For this reason, putting the coil just outside the gap might not be such a bad place. I have put my coil in a position just about 1/4 to 1/3 inside the gap for just this reason. On damping adjustments, make sure you do them *with* the pickup coil connected to its normal load. I found out by experience that a significant fraction of the damping comes from electromagnetic braking between the magnet and coil. If you adjust the damping with the coil open-circuited, the boom will be over-damped when you hook it up. The best method for fine tuning the damping is to use the "floor deflection" method described in Scientific American, once everything is running. On the "bias" adjustment, I have some notes on that at home; I'll send those along to you in a separate letter. Next question: telling LR from LQ. The answer is that it can be tough. In general, the Rayleigh wave (LR) travels slightly slower than the Love wave (LQ), so this is one clue. Consult a standard textbook on geophysics or seismology for a graph of distance vs. arrival time for all the various phases, including surface waves. Another clue is the sense of motion. Rayleigh waves involve elliptical ground motions in the vertical plane parallel to the direction of wave propagation. Love waves involve purely horizontal motions perpendicular to the direction of propagation. With three perpendicular-axis instruments, you can separate out the components. With only one component as we have, sometimes you will be lucky (or unlucky ? :-) ) enough that because of the direction of travel, one of the two modes will be invisible to your instrument, thus allowing you to identify the other one unambiguously. There is a fundamental property of surface waves though, that makes them quite hard to "pick" precisely; they are *dispersive*. By this I mean that their speed of propagation depends on wavelength. This is *not* the case for body waves. So, what starts out as a sharp "pulse" near the epicenter gets spread out after a few thousand km, with different frequency components arriving at different times. I find that it is actually quite rare that I can cleanly pick out a sharp LR or LQ arrival; they usually just sort of "mush" in gradually. Your background noise doesn't sound too bad. Have you noticed any time-of-day or temperature dependent effects of the kind that have bothered me? Your "spike" noise problem sounds very familiar, and I may be able to offer some advice. I noticed one-sided noise spikes (that seem identical to yours) that cropped up occasionally but didn't cause me much concern. Over time however, they got more and more frequent and bothersome. I eventually tracked the problem down to the electrolytic capacitors that appear in parallel with the power supply. As they age, they apparently tend to "short over" momentarily from time to time. This doesn't effect their usefulness in most traditional circuit applications like radios, but for us it's kind of disastrous. I yanked out the two offending (10 microfarad I think) capacitors, and the problem went away. Those particular capacitors weren't critical; they were just there for noise suppression, although they were doing just the opposite! :-) As a result, I havent even bothered to replace them yet, although if I do, I'll replace them with something of higher quality. I'm more concerned about the one remaining electrolytic in the circuit, the big 470 microfarad one. It *is* critical, and mine seems to be currently in the early stages of incipient failure. I guess I should start snooping around the electronics stores for a suitable replacement when it finally "goes". I hope this helps you with your problem. Let me know. That's all I have for now. I'll look forward to hearing more progress from you soon. Later! Mark: I have some answers to your questions about the amplifier circuit in your last couple of letters. First, about the "bias" adjustment. Here is what my notes say: The "bias" pot adjustment depends upon the supply voltage. It cancels offset variations with attenuation adjustment. Set the switches to "null" and "10k", and attenuation to 1.0. Then drop the attenuation to 0.0 and note the output offset voltage (with a voltmeter). Then return the attenuation to 1.0 and adjust the "bias" pot until the same voltage reading is obtained. Then proceed to set the offset pot as desired. I hope this tells you what you wanted to know. Your other question had to do with capacitors. I think I can give you an answer by reproducing here part of a table of capacitor characteristics that I found in an electronics text called "The Art of Electronics" by Horowitz and Hill (Cambridge Univ. Press). Here's what Horowitz and Hill have to say (in part) about capacitors: ------------------------------------------------------------- Type Capacitance Accuracy Temperature Leakage range Stability ------------------------------------------------------------- ceramic 10pF-1uF poor poor moderate polyester .001-50uF good poor good (Mylar) polystyrene 10pF-2.7uF excellent good excellent polycarbonate 100pF-30uF excellent excellent good polypropylene 100pF-50uF excellent good excellent Teflon 1000pF-2uF excellent best best tantalum 0.1-500uF poor poor --- electrolytic 0.1uF-1.6F terrible ghastly awful ------------------------------------------------------------- I have selected out just the parts of their table that were appropriate to our application. Let me know if you are able to find some substitutes for the "awful" electrolytic and tantalum components we're using now. I'm not sure where to buy some of those other kinds, although I assume you could get them by mail order or something.. Talk with you later! -Greg Mark: Sorry you haven't heard from me for a while. It turns out I got all 5 of your messages just fine. I noted that you mentioned you wouldn't be around until after April 2, so I procrastinated about replying to you. Then after a while with other things coming up, it just slipped my mind. Sorry. Anyway, I think you can go ahead a keep using the JPL address; it's where I have everything forwarded anyway. I got all the stuff you sent. I should be able to compile a version of the UUENCODE utility that I can use without much difficulty if I can just find a little spare time to do it :-) . So keep me posted, and let me know about any news, or if you've got any questions I can answer. Later! -Greg Greg, Thanks for the message. As long as no mail is getting trashed, I'm happy :-) I was getting a bit worried for a while. Just to keep you up to date on my progress: I replaced the two 4.5 uF capacitors in the amplifier circuit with 2.2 uF film types. I had a bit of trouble getting them to fit (they are quite bulky), so I put one of them inside and one outside the case :-) I haven't had any of the "ordinary" spike type noise since then. But, keep in mind that there were many periods of days or even weeks before this when I didn't get any noise either. The spectacular leakage events (presumably in the 470 uF cap) still occur once in a while, with their characteristic noise burst followed by an RC discharge pattern for a couple of minutes. I think you have a printout of one of those. The software development is, well... slow. My C experience is minimal, so it takes some time. I hope to have a reasonable package put together by summertime, and will add extra features and refine things over summer vacation. If you get UUxxCODE working and would like copies of any of my data files, let me know. The newer ones have the proper 4 byte long duration value at a file offset of 4 bytes, as per your format. The time and date fields are zeros for now, though. I haven't gotten around to implementing that yet... BTW, I also adjusted the bias setting and it works great. At the same time, I worked on fine tuning the damping, but am still unsure about it. The background noise when I made my adjustments was high enough that it wouldn't allow me to be sure on the ratios of the different peaks. When the noise amplitude is a significant percentage of the peak amplitude caused by walking up to the instrument, the required ratios of 6:1 to 10:1 are hard to see. I'll try it again sometime when the noise level is a bit lower. Hi Mark: I hadn't heard from you in a while, and I thought I'd just drop you a line to see how things are going. My excuse for not being in touch for a while was that I was buried in school work (the semester's over for me now :-)). So if you're currently buried in work, I understand; no need to reply to me right away. Anyway, how's the seismograph working? I've picked up a number of significant events including the big Costa Rica quake, however the last month or so has been pretty quiet. One amusing result I got was a series of peculiar-looking "quakes" that turned out to be a spider spinning a web on my pendulum! Just one more of the pitfalls of garage seismology! I can't remember if I ever acknowledged receiving your snail-mail package, but I did and found your results most interesting (and familiar!). Good luck and I'll be in touch with you soon. - Greg Greg, Thanks for the message; I've been quite busy with school also. Graduation is coming up in about two weeks and until then I may be rather swamped with work. I hope to work on the seismograph over the holiday weekend, though (see below)/. Yes, the last month has been fairly quiet. I'm actually rather happy about that, though, because my instrument has been off line now for almost one month :-( I built the amp circuit in too small a case, and it started acting up more and more, probably due to shorts of the internal wiring... I finally decided to take the guts out and rebuild them into a much larger box. So the circuit has been lying on my worktable for the last couple weeks while I've been away on various school trips :-) Software development is progressing... sort-of. Just a few major bugs to fix, and I'll have a fully functional recording/display program without many bells and whistles but with timestamps on-screen. I found out, BTW, that some of my noise was the result of not allowing enough conversion time in my ADC. A few NOP instructions remedied that problem! I also have spiders in the seismograph closet, by the way; I had better clean them out once every few weeks... I missed the big Costa Rica quake, but in another two weeks I expect to have the instrument running 24hrs/day with no disruptions from human noise, and I hope to pick up some more interesting events then. This last Sunday I attended the once-every-three-years USGS open house in Menlo Park. Lots of neat stuff there; I'll discuss that more later as I'm running out of time right now. Mark: Sounds like great progress! Did you happen to pick up the big quakes last weekend from the Gulf of California and Argentina? They came in quite well on my instrument, although I never heard a word about either one on the news. I suppose that given all the tourist traffic that will be heading down to Baja shortly, the Mexican gov't isn't too anxious to publicize big earthquakes :-) Talk to you later! Greg, My neighbor site (and Internet mail forwarder), pacbell.com, is sick and is currently not accepting outbound mail. This message and the one I wrote last Thursday (the 23rd) will be delayed for a while. Today I finished rewiring the seismograph amplifier, and it seems to test out ok. I'll know for sure on Tuesday when I fire it up with the rest of the system at school. I found the probable cause of the erratic operation: a .01 uF capacitor with axial leads had been broken -- one end just snapped off. At the USGS open house a week ago, as I mentioned in my last message, there was a variety of seismograph equipment on display. One table had some examples of older instruments, including a Wood-Anderson (I think that's the name) torsional seismometer. I was quite interesting to see the older mechanical devices on display. Their "modern" exhibit included an operational 3-axis (vertical, N-S, E-W) long period instrument, a bank of ten or so recording drums from remote stations in their extensive short period network, and a workstation computer linked into their data collection systems. On a CA map they put dots at the ~100 locations of the short period stations, which relay their data by VHF radio (100 mw solar powered NBFM transmitters with FM subcarrier modulation), phone, and other means. They are all monitored automatically by the USGS computer systems, and USGS personnel are notified automatically by pocket pager whenever a quake in a critical area occurs. The mechanical recording drums are now the backup system, in case of computer failure. Mark: Murphy's Law strikes again! I was out of the state on some personal business when the quake hit, so I can't give you any first hand accounts of what it was like. I couldn't believe my ears when I turned on the car radio Friday morning and heard the news. Kinda makes me wonder what will happen when I leave for vacation later this week :-) Even though we were quite close to the epicenter, there was no discernible damage at our house. In fact, as far as our house is concerned, the '87 Whittier quake was significantly worse. Murphy didn't get it *all* his way; the power stayed on during the whole thing, and I got complete records of the main shock and aftershocks. The seismograph trace was clipped off-scale for about *eight minutes* following the main shock. Speaking of aftershocks, there have been amazingly few of any size. The sequence strikes me as having died down unusually fast for an event of this size. Anyway, I'll be glad to send you uuencoded files of the records as soon as I get things sorted out. I've only been home since last night though, so I'm still a bit disorganized. Since nominally we're supposed to leave for Mexico on Wednesday evening, I'm not sure how much quake-tracking I'm going to have a chance to do before then. All this means that you shouldn't be too surprised if you don't hear much from me before the second half of July. But never fear, I'll get the info to you eventually :-) Meanwhile, take it easy and have a good holiday weekend! - Greg Hi Mark: I just got your mailing. Thanks a lot! I haven't got a chance to uudecode them yet, but I didn't want to wait to acknowledge getting them. I'll probably have a look at them this evening. You just beat me to the punch...I was going to send you copies of my data from both the Sierra Madre and Oregon quakes this week. The latter was very strong on my instrument too. As a matter of fact, even though it sounds strange, it was off-scale for a longer time that the Sierra Madre quake, which was only 9 km away! You can expect to get the files in the next couple days. I will go ahead and send you files that have been first compacted with Stuffit (version 1.5.1) and then uuencoded. If you have any trouble reading them, just let me know, and we'll try something else. I assume your PC Unstuffit program knows what to do with the two "forks" of the Macintosh file format. The "data fork" contains the data you are interested in; the "resource fork" contains nothing but icon information and other Mac-specific stuff of no interest to you. I hope this allows you to extract what you want. Keep me posted on your address, and on your progress with the upgrades. I've occasionally toyed around with the idea of event-detection algorithms myself, but I haven't really done anything. Is there a way that you might be able to use some kind of "circular buffer", that retains data for some fixed length of time and continually overwrites itself (like a snake swallowing its tail)? Just an idea... I'll be in touch with you soon, and get those files to you. Bye for now. - Greg Hi Mark: Assuming that your mail will wait for you while you're on vacation, I'm sending along (in two separate messages) the records from the Sierra Madre and Oregon quakes, as promised. I have Stuffit'ed and then uuencoded each file. By the way, with regard to my comments last time on Mac file formats, I think I've figured out what's what. I believe that when you uudecode each file, you'll actually get out *three* files: the first (and longest) is the data fork ; that's what you want. The other two pieces are the resource fork and "finder info", which I believe you can throw away. Again, if you have problems, let me know. Also, I got the file you sent me decoded, and it looks really neat! Besides your nice recording of the quake, I'm pretty envious of your nice noise-free conditions, as compared with mine! This might give me some extra motivation to try and find a better location for my instrument (yeah, I've been saying that for a while... :-) ) Anyway, enjoy your vacation, and I'll be in touch later. - Greg Greg, I finished identifying most of the events detected by my instrument during July. I've been editing the seismicity reports for northern/southern CA and the NEIC QED listings to show just the events detected, then saving them to filenames similar to my data files. It's interesting how many events can be identified by looking for prospective quakes, then going back over the data and looking at the right time of day. I've pegged at least two or three quakes this way which I didn't notice when I first scanned through the data files visually. One of them was a M5.7 quake in the Kuril Islands at 1215 UTC on July 13 (5:15 AM on the 13th local time). It was just barely discernable above the [fairly low] noise. My noise floor has been really low over this summer -- lower than during last school year. If you ever have the chance to move your instrument to a quiet site, by all means do so. It really helps! As for my site, it will become very noisy about the first of September :-) My idea for event detection is quite simple. Briefly: 1) Create an array of x bytes, say, 20. 2) For each byte sampled from the seismometer, take the difference between it and the last byte sampled, and store this in the array. 3) When the end of the array is reached, simply wrap around to the beginning. 4) After each of these operations (at each sample), sum the contents of the array, and check against a preset threshold value. 5) If the sum is large enough, an event is declared. This gives shorter-period wave components a built in advantage, since higher frequency waves will yield larger fluctuations per unit time than long-period waves of equal amplitude. A good thing, since all events (local or teleseisms) begin with short period components. The threshold for ending the event should be different and lower (in other words, add hysteresis) -- as close to the background noise level as possible. That's all for now. I can't wait to go collect data from the past couple of weeks -- the location off the California/Oregon coast has been very active, it seems! Mark: I'm glad to hear you were able to figure out the problem. If you want any more data, just let me know. Those northern CA quakes of last week came in really *big*! As an aside, got curious last night wondering what our records would *sound* like if they were played as audio (speeded up by a few thousand times :-) ). So I figured out a way of sending the earthquakes through my Mac's sound synthesizer, and the results were kind of fun. Local earthquakes (even big ones) aren't all that impressive; mostly sounding like crackles, pops, and snaps (Rice Krispies, anyone?). My floor noise sounds remarkably like ocean surf, but the most interesting are the large distant quakes. They sound sort of metallic, like a cross between banging on a tin roof and crashing cymbals. Not too "scientific" but sort of fun, huh? :-) With regard to your question on header format, I just write out the different fields one at a time as binary int values. I'm not sure I could say that your way or my way is more "correct"; it's just whatever you settle on as far as I can see. Below is a sample chunk of code that shows how I write my files. Despite the weird Mac-specific system calls, I think you'll be able to figure out what's happening... --- #define ONEDAY 86400L DayTime(&year,&mon,&day,&hr,&min,&sec) ; SetFPos(fileRef,1,0L) ; bcount = (long) sizeof(short) ; zone = save_zone ; try1 = FSWrite(fileRef,&bcount,&year) ; /* writing header...*/ try1 = FSWrite(fileRef,&bcount,&mon) ; try1 = FSWrite(fileRef,&bcount,&day) ; try1 = FSWrite(fileRef,&bcount,&zone) ; bcount = (long) sizeof(long) ; len = ONEDAY ; try1 = FSWrite(fileRef,&bcount,&len) ; /* length of data stream...*/ bcount = ONEDAY * (long) sizeof(Uchar) ; try1 = FSWrite(fileRef,&bcount,(*data_hndl)) ; /* the data itself ! */ --- I didn't realize I had never before mentioned my "step offset" problem in the records. It invariably draws questions and comments from people whenever I show it to them for the first time. I *don't know* what the cause of it is, but it definitely has *nothing* to do with real motions of the ground. The steps occur when all is quiet and nobody is anywhere near the garage. The cause seems to be electronic, and I tend to suspect either a capacitor problem (similar to those we discussed earlier) or a slightly flaky potentiometer. I have a slight preference for the former, since the steps have been less frequent and smaller since I removed those really *bad* electrolytics from the power supply circuit. So maybe the remaining (indispensible!) 470 uF capacitor is the culprit. Some day I may get brave enough to tear up the box and replace it ! :-) That's all I have for now. Keep me posted, and let me know when you want some more data. And yes, I'm thinking more a more seriously all the time about trying to fing a more quiet site... - Greg Mark: I saw your posting. I have plans to build the same seismograph when I get time. I would very much like to hear your experiences if you get it built. My feelings about the cultural noise you described is that its too high a frequency to disturb your instrument, especially the people walking by. Jets and large vehicles also have specta that are much different than teleseisms, which are what you are trying to record. Even if some of the energy is within your passband, the resulting event will look different from an earthquake. Its important that you get good coupling with the ground and don't set the instrument on loose dirt. This will help your signal to clutter ratio more than anything. I've collected quite a few articles on this subject if you're really interested, I'll try and send them to you. Also, once upon a time I was a geophysicist and actually interpreted similar records. It was a long time ago, but I'll try and answer any questions you might have. Keep in touch. From: Richard Stead Subject: Re: Building a Simple Seismograph -- Need Help Yes, this instrument will record waves from earthquakes even on the other side of the globe, if they're large enough (the instrument may not be sensitive enough to catch a quake like the recent Iran quake, which was only 7.3. I say "only" because there are probably 20-30 quakes a year larger than that, just not located where they can cause such damage.) It's period, 12-18 seconds, is designed to be most sensitive to the largest waves from distant quakes, the surface waves. These waves can last over an hour at that range of periods from a very large event, however you will not have sufficient information to get the location or size of the event. This instrument will not be very sensitive to people walking (1 sec period) or traffic (typically 0.1 to 1 second). Jets flying overhead are high frequency by seismic standards, most of the energy is at 60 Hz and above, unless there is a sonic boom. Things the instrument will be very sensitive to are "wind noise" which is wind-to-ground coupling enhanced by the presence of trees, etc. to sway. It may also be sensitive to wave motion against the shore if you are anywhere near the coast (this is called "microseism" and can be recorded by our best seismometers, even in the middle of the continent). Microseism is strongest at 7 and 15 sec. Other considerations are heavy manufacturing or rail lines. Given that this instrument will not be very sensitive by modern seismometric standards, it should be satisfactory to install the instrument at your home. It would be better to place it on a solid base - a garage or basement floor would be fine - than on soil, even firm soil. Seismic installations in your area are normally done by digging a "vault", pouring some concrete and mounting the instrument to the concrete. One way to see that the distances you are concerned about are not important is to examine the wavelengths of the energy. SEismic waves at 12-18 seconds would be traveling at roughly 3 km/second. This means the wavelength is 36-54 kilometers, so you could practically install the instrument directly on the road surface and it wouldn't make much difference, except the high frequency jarring there might damage it. You would have to get several kilometers away before you would notice a difference at those periods. Of course, the instrument does respond weakly to shorter periods, and a large enough signal at short periods will record. The instrument you describe is nicknamed a "garden gate" seismometer. The idea is to provide mechanical gain by lowering the restoring force on the pendulum as much as possible. So the arm is nearly horizontal, tilted just slightly toward the weight. The restoring force for the system is gravity, but at that angle the force of gravity acting along the arm of the pendulum is very small. This is the easiest seismometer to build and get good sensitivity. The most modern seismometers are "force balance" seismometers in which a feedback circuit is used to keep a mass stationary relative to a coil. The current in the feedback circuit is measured to calculate the force necessary to move the mass in a rest frame such that it follows the coil which is moving in the rest frame with the vibration of the earth. This force is directly related to how far the point on the earth moves during the vibration. There are several other designs that have been used in seismology. Good luck with your project. Richard Stead Center for Seismic Studies, VA From: Richard Stead Subject: Re: seismographs Yes, you should cover it, wind and direct sun will have a strong effect on it. Temperature fluctuations will also have an effect, more so on the instrument directly than indirectly via the patio. This is another reason why seismic stations are usually installed in underground vaults - the temperature is more stable. I'll represent your instrument by this little figure. As the temperature changes, the lengths of | A and C will change, and the tension on B will |\ change. This changes the fundamental response A| \B of the instrument. | \ +----O | C It is also true that the concrete slab could tilt slightly from heating by the sun, but this is probably a much smaller effect. If would depend on the size of the concrete slab, how deep it's buried, what other loads are on it, etc. It's very difficult to predict. Nevertheless, such slabs are a favorite spot for seismologists to install temporary instruments (although usually in a garage or basement to protect them from wind and sun and to stabilize temperature a little). For demonstration purposes, we have a seismometer installed in the parking garage under our building here in the Rosslyn section of Arlington, VA. This is a 17 story building surrounded by other high rises, and the DC Metro goes underneath it. We still record most of the major quakes worldwide quite clearly. The instrument is of a different design, but would still be affected by temperature variations and tilts. One other thing your instrument will pick up quite well - lightning, even far away. Both the thunder and electrical activity will affect the instrument. --Richard Stead From: Jim Scheimer Subject: Re: Building a Simple Seismograph -- Need Help I had a friend in Livermore who constructed one of these... He put it on his hearth because it gave him better coupling. His major problem was finding a drum recorder which he got from Livermore Lab surplus. I though it was agreat device and I am a seismologist by trade. Don't worry too much about the noise sources you mentioned. The seismometer in question is a critically damped long period instrument. The noise sources you talk about are all in the 1 to tens of Hz. frequency ranges. The instrument should not be too adversely affected by sources so far out of its peak response range. If I can remember the fellow's name I'll send it to you (he retired several years ago, he used to work at the visitor's center at LLNL.) Drop me a line if you have more questions, I still am involved in instrument design.