From: Kasper VanWijkTo: PSNLIST CC: ted Channel ; Martin L. Smith ; John Taber Sent: Thu, 17 May 2012 4:58 Subject: slinky seismometer follow up Dear PSN-ers: Let me introduce myself - my name is Kasper van Wijk, and I teach=20 geophysics at Boise State University. Shortly after moving to Boise, Mr. Ted Channel, a devoted psn-er, contacted me about collaborating on home-built seismometers. Since hen, we (but mostly he) successfully built lehman's and developed a hort-period vertical sensor as simple as a magnet on a slinky in a lear tube.=20 We have fine-tuned the latter design in our latest version (VIII),=20 as we recognized it has some properties that make this an optimal unit or schools; our primary target. Our sensors pick up any EQ magnitude 6=20 or greater, worldwide. A good spot for the sensor gets this actually=20 down to about Ms 5.6. Clear P-wave arrivals can be used for=20 triangulation to the rough source position, and for any other=20 seismology exercise a K12 school. After years of testing (among things, next to an AS1), I can onfidently state the device is not under-damped by any definition, as ong as the top of the magnets are flush with the top of the coil and amper ring, respectively. I just wanted to clear this up, efore this goes widespread among the experts at the psn! The second oint, however, about the sensor's frequency response is more nteresting: we have decided to produce an interface between sensor nd computer that is more or less flat in response. The result is hat our seismometer, just as the AS1, is largely a ~1s period ody-wave sensor, but I'll gladly post images showing (weak but ignificant) evidence of surface waves. So it is true that we can boost" the low end of the recorded spectrum, to enhance longer-period urface waves. We just have not done this (yet), but our interface is ased on the arduino uno (http://arduino.cc). The code to run it, is reeware, and thus anybody is free to download the code via our ebsite and adjust as he/she sees fit. We linked Martin's google project site for the NERdaq (the interface) nd tried to summarize a lot of information including how to build and et up your slinky seismometer at http://cgiss.boisestate.edu/ssis=20 ut feel free to contact me at kaspervanwijk@.............. if you ave particular questions/comments. Hi there everyone, Here goes ! Someone at Boise State should have a copy of the IASPEI New Manual of= =20 Seismological Observatory Practice. Amongst other information, it gives=20 the characteristics of earthquakes. For PSN members, you can download a=20 Seismogram Training Manual at http://psn.quake.net/info/analysis.pdf dated= =20 1960 giving a lot of the same information. Near Regional Earthquakes < 415= =20 miles away still require a maximum P wave bandwidth approaching 3~4 Hz and= =20 many schools in the USA are likely to observe one !=20 The AS-1 was NOT a seismometer ! It was designed as a 'demonstration=20 instrument ONLY' and could pick up microseisms, a laboratory bench being=20 thumped, a school class jumping up and down, etc. It was basically a 1.5=20 second pendulum, nominally critically damped with oil (non linear) and=20 it's period was electronically extended to ~4.5 seconds. Note that the=20 Boise web site is INCORRECT in claiming a 1 second period. There seemed=20 to be no attempt at a low noise design. There were no instructions for=20 setting up the damping correctly and it only had a working temperature=20 range of about +/-5 F Deg. If you set it up to operate during the day=20 time, it was likely to stop working completely when the room temperature=20 fell overnight. I designed a magnetic damper to fit it, which was linear=20 and not temperature sensitive. John Lahr had several batches made and=20 distributed.=20 The Manual for our school seismometer, which Ted might find interesting= ,=20 is shown at http://www.mindsetsonline.co.uk/images/Seismometer.pdf =20 . I have never used a Slinky Seismometer, but there seem to be some=20 points about the design which might be improved. My previous comments=20 were made after reading the fuzzy description on the Boise web site. Can=20 I suggest that Kasper check that both the text and the photos shown there= =20 are factually correct and up to date, please? I have no way of knowing if= =20 any mistakes have been made, so I will assume that everything on the web=20 site is intended to be correct. =20 I am quite conversant with digital noise reduction and signal=20 recovery in an ADC by over sampling - I first used it way back in 1984.=20 . My first query :- Isn't there ANYONE at Boise who can design a 16=20 bit ADC board and a USB interface ? THAT IS WHAT WE DID ! Can't you get=20 help from either your physics department or your computer department ? =20 . My apologies to Kasper over misreading the Impulse Response of the=20 output filter as the Impulse Response of the seismometer and commenting=20 adversely on the damping. I have now found the separate Boise Poster at=20 http://earth.boisestate.edu/pal/files/2011/12/slinky_poster.pdf=20 I HAVE ALSO FOUND THE DAMPING INSTRUCTIONS FOR YOUR SLINKY :- "For=20 the damper, the magnet should be even with the top of the copper, or =C2=BC= "=20 inside. Properdamp is when the spring will bounce only once, or twice,=20 not repeatedly". Precision ? Er, hmm ? Have you measured the change in=20 the vertical position of the sensor magnet with changes in temperature?=20 A change in the armature position could effect both the damping and the=20 sensor output. A long spring can make quite a good thermometer ! =20 So, does a constructor just have to guess what damping is correct ?=20 There seems to be NO easy way of finely adjusting it, other than taking=20 the column apart repeatedly and repositioning either the copper tube or=20 the damper magnet? If you give an impulse to the seismometer and the=20 signal trace is twenty units high, the trace should then just cross the=20 zero line by 1 unit and tail off without any further oscillation. The=20 damping needs to be ~0.7 critical to give a 'flat with velocity'=20 characteristic and reasonable accuracy.=20 I am really rather puzzled, since Kasper says "I can confidently=20 state the device is not under-damped by any definition...." Maybe, but=20 is it CORRECTLY damped at ~0.7 Critical, please ? And how are operators=20 supposed to set this up reasonably accurately ?=20 =20 I got confused over the Impulse Response, MOSTLY because EVERY ONE ELSE= ,=20 professional and amateur, uses Butterworth Filters! These give minimum phas= e=20 delay and PRESERVE THE SEISMIC WAVEFORM FOR ANALYSIS. Kasper says that the= =20 phase delay is only ~0.4 seconds, but this is a WHOLE CYCLE ERROR at 2.25 H= z!=20 And the delay will NOT be constant with frequency. Steep cut filters like= =20 this seriously distort a complex wave form. A ringing response is definitel= y=20 MOST undesirable.=20 =20 I note that three magnets are used, two for the detector coil and one f= or=20 the damping and that all three have the same orientation, threaded on a mil= d=20 steel eye bolt with steel nuts. Moreover, the eyebolt is directly hooked on= to=20 a steel spring, giving a 'stray magnetic field detector' all of 20" long! W= OW!=20 I suggest that you make a new seismometer using BRASS studding and 4 ri= ng=20 magnets, a N/S pair facing a S/N pair, to give minimum external magnetic fi= eld=20 coupling, like you did on earlier models, which had a brass S hook. Use a= =20 hacksaw to cut a slit in one end of the studding and glue a vertical stainl= ess=20 steel washer into it with 2 part acrylic glue. You may want to drill / file= =20 both sides of the washer to round off the inside edge. Hook this onto the= =20 spring with a blob of BluTack or similar to prevent it sliding noisily=20 around. BluTack is a putty like adhesive which never sets hard. This=20 arrangement should effectively separate the steel spring from the magnetic= =20 fields.=20 It is also likely to increase both the sensitivity and the damping,=20 since no flux will then be lost through the mild steel studding. You might= =20 then be able to 'get away' with just two ring magnets ?=20 =20 The three brass bolt mountings of the coil and the damper tube don't=20 look very secure and there is no automatic alignment. Could you not fit=20 circular Aluminum sheets on an extended coil tube and stick them on with=20 PVC cement / PVC rings ? The outside of the disk could have three short=20 springy strips bent at right angles to align it accurately inside the=20 acrylic tube and the inside edge could have 3 slots to limit any induced=20 currents.=20 Looking at the amplifier circuit, the Slinky sensor coil is included=20 in the feedback loop, allowing force feedback, instead of being simply=20 connected to an earth reference and the +ve input of the opamp. This=20 increases the noise. A 2000 Ohm coil of copper wire makes quite a good=20 thermometer, as well as a pick up for RFI ! =20 =20 The period of a circular section spring vertical pendulum is given=20 by T =3D 2.Pi.SQRT[2.m.l.r^2 / Pi.n.a^4] : a=3Dwire radius, r=3Dradius of t= he=20 spiral, l=3Dlength of the wire, m=3Dmass, n=3Dtorsional modulus.=20 =20 It is possible to compensate for up to about 10x the natural period of= =20 a pendulum electronically, before running into serious noise problems. To= =20 get a fully compensated period of 20 seconds, the spring vertical would=20 need to have a natural period of 2 seconds. Increasing it to 1.5 seconds=20 would allow a compensated period of 15 seconds. Let me know if you want=20 to do this ? Try out the long period extension filter built into AmaSeis? It would be most desirable to detect S waves at their full amplitude.= =20 Schools and amateurs need to measure the distance to the quake using the=20 P to S wave time delay and the tables / graphs.=20 The amplitude response of 20 second waves is down by about a factor=20 of 400 on a 1 second sensor. Being able to locate quakes by their Rayleigh= =20 waves and then search for the P and S waves down in the background noise=20 may triple the number of quakes you can detect. From personal experience,= =20 increasing the compensated period of an AS-1 from 4.5 to 15 seconds made a= =20 simply HUGE difference to the detection rate ! =20 =20 I admit being puzzled as to why, if you needed an AS-1 replacement,=20 that you didn't just find a couple of retired mechanical and electronic=20 engineers to produce a good updated instrument ? It could make a welcome=20 boost to their retirement incomes and they would have a part time=20 occupation, preventing them from getting bored !=20 =20 THE LEHMAN. =20 Ted, you might find it interesting to look at the Lehman that I use=20 at http://www.jclahr.com/science/psn/chapman/2008%20lehman/index.html =20 It has a 0.016" thick SS hard plane rolling on a 1/2" SS bearing for the=20 suspension. I can EASILY set it up for 30 seconds period and I have run=20 it OK at 60 seconds ! A plane rolling on sphere gives a better geometry=20 than your design of a sphere rolling on a plane. =20 I note that you seem to have had difficulty in getting even a 20=20 second period on your older Lehmans ?=20 This can be due to either the top or the bottom suspension, or both. The top suspension, being a wire in your case, needs to be very firmly=20 located / fixed, preferably between the edges of clamping disks, as on my= =20 Lehman. On the 8020, maybe you could fit an L bracket to the top of the=20 vertical column and mount eccentric washers and a winding bolt on it,=20 similar to my system ? The photo of the top wire looks as if it is fouling= =20 on the edge of the dome nut. It needs to be cleanly clamped. I suggest that you consider a 'V' Al strip brace between the vertical= =20 column and the base cross bar to improve the 'side to side' stability and= =20 increase the stable period. One problem with Aluminum is that it is not=20 very rigid and seismometers need to be 'rock solid' ! I note that your NEW type Lehman is 38" long, 24" wide, 26" high and=20 has a 32" boom. This may be starting to get a bit 'willowy' with only a=20 1/2" OD Al boom and a 45mm square Aluminum frame. Is this an attempt to=20 increase the period ? Sorry, but from experience I think that you would=20 very likely get a better performance from a smaller modified U channel=20 version. As I mentioned, I have run my 22" boom SEP prototype at 60=20 seconds period. The practical limit in period is due to natural ground=20 tilt driving the arm off scale. =20 =20 What is the cost of your U Channel frame as opposed to that of the=20 8020 45mm square frame, please ?=20 The DIY 'EASY' LEHMAN that I designed uses 3" x 1" Aluminum U Channel,= =20 but the sides were upward / inward facing (opposite to yours) with 6"=20 corner plates and a trapezoidal end plate, 3" at the top, 8" at the bottom,= =20 7.5" high, were provided. The base bar is 30" long and the two end bars=20 are 18" long. The arm is 56 cm to the CofG, giving a 1.5 second pendulum.= =20 The lower bearing is a 3/8" SS bolt drilled to a cone on the thread end=20 and mounted through the upright with two sets of nuts and washers. A 3/8"= =20 SS ball bearing is stuck in the cone with acrylic adhesive and the=20 position of the bolt is adjusted to give a suitable period. This=20 construction is QUITE ADEQUATE for periods of 30 seconds. If you want=20 longer periods, you need to stabilise the central position of the arm with= =20 a photocell position detector, an amplifier and a feedback coil to eliminat= e=20 the natural long period tilt drifts. This system can also enable you to see= =20 the daily Earth tides ! =20 =20 Back in about 2008, Meredith Lamb, Charles Patton, Brett Nordgren and= =20 I co-operated to write a paper on bearing systems for amateur seismometers.= =20 See http://www.myeclectic.info/SeismoPivots/seismopivots.htm These all=20 worked well It was reported on PSN, but it does not seem to have got onto= =20 the reference list. Apart from various wire and foil systems, the two=20 types which worked OK were a hardened plane rolling on a hardened sphere=20 and hardened crossed rods. We tried the ball + two rods bearing type that= =20 you used, but it did not work at all well. One problem is that a sphere=20 rolling on two rods has a small, but critical, rotation about the point=20 of contact, giving some friction. The 8020 seismometer uses a SS bolt=20 head as the flat surface. Sorry, but austenitic stainless steel is just=20 NOT hard enough ! You DO NEED HARDENED martensitic stainless steel !=20 Also, you don't mention that Stainless Steel Ball Bearings ARE=20 REQUIRED for seismometer suspension systems ! Ordinary steel ball=20 bearings only last a short time before corroding at the contact point.=20 I use a 1/2" OD SS Bearing on the vertical column and a hard 16thou"=20 thick SS flat plate, cut from a Swann-Morton Skin Graft Blade, stuck to=20 the end of the boom with TWO COMPONENT ACRYLIC GLUE. It has a strength=20 of 3,500 psi and when set, IT REMAINS SLIGHTLY FLEXIBLE. It will stick=20 soft metals like Aluminum, Copper, Silver etc just fine. Epoxies don't=20 bond well and they mostly cure brittle hard. There may be some 'break=20 off' DIY SS knife blades available, but most of them are ordinary steel.=20 You can also use Tungsten Carbide. SS bearings are supplied by=20 Smallparts, now called www.amazonsupply.com=20 For crossed cylinders, the cheapest solution is to use fully=20 hardened martensitic stainless steel rod. You may have to polish it=20 before and after cutting and hardening it. The SEP uses 1/8" Tungsten=20 Carbide needle rollers, but these are quite expensive. You can also=20 buy 1/8" Tungsten Carbide drills and use the shanks. WC Circuit board=20 drill shanks are also available. Do you have any local machinist=20 companies who use WC milling cutters ? Anyone in the University ? Old /=20 broken cutters are usually just thrown out.=20 =20 On your early Lehman, when the ground moves, the drag force on the=20 damping plate will try to rotate the boom, since it is offset. This can=20 also rotate the sensor coil, giving false responses. The rotation=20 problem has been corrected on the 8020 Lehman. On the SEP, the brass=20 mass is offset and the damping blade lies on the line joining the CofG=20 to the lower bearing, thus dramatically reducing any rotational couple.=20 I also use a V cable made from 7 core SS nylon coated 40 lb fishing=20 trace, which is supplied with crimp tubes. You can buy it in up to 200=20 lb rating. It is very effective at preventing any rotation of the arm.=20 I suggest that you try using a V cable with a side extension to the=20 mass to give a separation of the attachment points of about 6" and also=20 run the top disk of the V, very close to the pillar. . On the Boise web site, you say that Cu and Al are paramagnetic. This=20 is incorrect. Cu is very nearly neutral. Al has a significant diamagnetic= =20 moment, which can cause an Al damping blade to push a Lehman arm to=20 either stop if the blade is not wide enough and it comes close to the=20 edge of a damping magnet. It is difficult to get ~pure Al and it has=20 nearly twice the resistivity of Cu. 'Strong' Aluminum alloys have very=20 much higher resistivities.=20 Sheet Cu makes a much better damping blade ! Instead of the 3/16" Al=20 alloy used on the 8020, I suggest that you try 1/32" thick Copper sheet=20 and also mount BOTH the damper blade and the sensor coil with two bolts?=20 http://www.ksmetals.com stock 0.016" and 0.025" Cu sheet, if you can't=20 buy it locally. You should be able to reduce the recommended clearance=20 between the magnets and the coil from 3/16" to 1/16". This will increase=20 the voltage output of the coil and also increase the damping. Instead of=20 using Perspex sheet for the sides of the coil, have you tried using thin=20 fibreglass sheet / thin circuit board ?=20 Instead of using an electric drill for coil winding, I use a hand=20 drill clamped horizontally on a work bench and count the number of turns=20 of the handle. Count the number of teeth on the two gears to find the=20 turns ratio. A hand drill is MUCH easier to control and it doesn't take=20 long to wind a coil. I use a plastic tube of the smallest diameter=20 heatshrink sleeving, pre-shrunk, to guide the wire. =20 Brass tends to corrode Aluminum in a damp atmosphere. Stainless=20 Steel bolts and studding are not difficult to acquire and are not very=20 expensive. I use all SS nuts, bolts, washers and spring washers, bought=20 from a marine chandler / boat supplier, or McMaster-Carr.=20 I noticed that you are using 0.016" stainless steel wire for the=20 suspension. Have you checked if it is austenitic or martensitic wire ?=20 Austenitic SS is NOT suitable for a wire which flexes. www.daddario.com=20 produce a wide range of steel music wires, usually Nickel plated for=20 rust protection. It is MUCH stronger and works well. What does you local=20 musical instrument shop stock / supply ?=20 The levelling screws are all 1/4" UNC having only 20 tpi. If you used= =20 1/4" UNF set screws, you would get 32 tpi, which would make the levelling= =20 easier to adjust. On the 8020, I suggest that you separate out the level=20 knob from the arm. Shorten the boom a few inches and mount the knob on the= =20 end of the base bar, well clear of the mass. It will give finer period=20 adjustment. I suggest that you fit a wavy spring washer between the nut=20 and the washer on top of the base bar to keep the thread under tension at= =20 all times. The frame will expand and contract with changes in temperature= =20 at a different rate to the ground, so the seismometer needs to be supported= =20 on smooth flat plates so that the ends of the screws can slide a few thou.= =20 . The glass case is fine for the demonstration, but amateur seismometer= =20 cases are usually made from 2" Celotex insulation board. An Al sheet with= =20 heating resistors is fitted inside the top of the case. This maintains=20 a positive temperature gradient in the vertical air column above the=20 ground, inhibiting any convectiion. If you don't do this, when the air=20 temperature falls below the ground temperature in the early hours of the=20 morning, rolls of cold air peel off the inside of the top of the chamber,= =20 fall to the ground and push the arm about. This is seriously noisy !=20 The difficulties of temperature control increase with the physical=20 size of a seismometer. The Sprengnether horizontal had a 14" boom and=20 was limited to a period of 15 seconds by tilt instabilities for reset=20 intervals of several months. The SEP was designed with a 22" boom and=20 the drift is checked weekly.=20 =20 I am puzzled that you record data with AmaSeis and then analyse it=20 with Winquake ? AmaSeis has a perfectly good set of filters, INCLUDING =20 a digital period extension filter that could be used to extend the period= =20 of the Slinky ! What is the reasoning, please ?=20 Ted, feel free to contact me if you are interested in a discussing=20 any of these topics ? I also run a small machine shop and have quite a lot= =20 of experience of making things.=20 =20 Regards, =20 Chris Chapman (Physics) = From: Kasper VanWijk <kaspervanwijk@..............>
To: PSNLIST <PSNLIST@..............>
CC: ted Channel <tchannel@............>; Martin L. Smith <martin@b= lindgoat.org>; John Taber <taber@........>
Sent: Thu, 17 May 2012 4:58
Subject: slinky seismometer follow up
Dear PSN-ers: Let me introduce myself - my name is Kasper van Wijk, and= I teachgeophysics at Boise State University. Shortly after moving to Boise, Mr. Ted Channel, a devoted psn-er,contacted me a= bout collaborating on home-built seismometers. Since then, we (but mostly he) successfully built lehman's and developed a short-period vertical sensor as simple as a magnet on a slinky in a clear tube.=20 We have fine-tuned the latter design in our latest version (VIII),as we recogn= ized it has some properties that make this an optimal unit for schools; our primary target. Our sensors pick up any EQ magnitude 6or greater,= worldwide. A good spot for the sensor gets this actuallydown to about Ms 5.6. Cle= ar P-wave arrivals can be used fortriangulation to the rough source position, and = for any otherseismology exercise a K12 school. After years of testing (among things, next to an AS1), I can confidently state the device is not under-damped by any definition, as long as the top of the magnets are flush with the top of the coil and damper ring, respectively. I just wanted to clear this up, before this goes widespread among the experts at the psn! The second point, however, about the sensor's frequency response is more interesting: we have decided to produce an interface between sensor and computer that is more or less flat in response. The result is that our seismometer, just as the AS1, is largely a ~1s period body-wave sensor, but I'll gladly post images showing (weak but significant) evidence of surface waves. So it is true that we can "boost" the low end of the recorded spectrum, to enhance longer-period surface waves. We just have not done this (yet), but our interface is based on the arduino uno (http://arduino.cc). The code = to run it, is freeware, and thus anybody is free to download the code via our website and adjust as he/she sees fit. We linked Martin's google project site for the NERdaq (the interface) and tried to summarize a lot of information including how to build and set up your slinky seismometer at http://cgiss.boisestate.edu/ssis<= /FONT>=20 but feel free to contact me at kaspervanwijk@.............. if you have particular questions/comments. Hi there everyone,Here goes ! <= TT> Someone at Boise State should have a copy of the IASP= EI New Manual ofSeismological Observatory Practice. Amongst other information, it = givesth= e characteristics of earthquakes. For PSN members, you can download aSeismogram Tr= aining Manual at http://= psn.quake.net/info/analysis.pdf dated1960 giving a lot of the same information= .. Near Regional Earthquakes < 415miles away still require a maximum P wave bandwidth approaching 3~4 Hz andmany = schools in the USA are likely to observe one !=20= The AS-1 was NOT a seismometer ! It was designe= d as a 'demonstrationinstrument ONLY' and could pick up microseisms, a laboratory bench bei= ngthumped, a school class jumping up and down, etc. It was basically = a 1.5second pend= ulum, nominally critically damped with oil (non linear) andit's period was electronically extended to ~4.5 seconds. Note that the <= PRE style=3D"FONT-SIZE: 9pt">Boise web site is INCORRECT in claiming a 1 second period. There seemed
to be no attempt ata low noise design. There were no instructions for setting up the damping correctly and it only had a working temperaturerange of about +/-5 F Deg. = If you set it up to operate during the daytime, it was likely to stop= working completely when the room temperature =fell overnight. I design= ed a magnetic damper to fit it,= which was linearand not temperature sensitive. John Lahr had several batches made and=distributed.The Manu= al for our school seismometer, which Ted might find interesting,is shown at http://www.mindsetsonli= ne.co.uk/images/Seismometer.pdf.
I have never used a Slinky Seismometer, but = there seem to be somepoints about the design which might be improved. My previous commentswere made after reading the fuzzy description on the Boise web site. CanI suggest that Kasper check that both the te= xt and the photos shown thereare factually corr= ect and up to date, please? I have no way of knowing ifany mistakes have been made, so I will assume that everything on the website is intended to = be correct.I am quite conversant with digital no= ise reduction and signalrecovery in an ADC by over sampling - I first used it way back in 1984..My f= irst query :- Isn't there ANYONE at Boise who can design a 16=bit ADC = board and a USB interface ? THAT IS WHAT WE DID ! Can't you gethelp from either your physics department or your computer department ? = PRE>.My apologies to Kasper over misreadi= ng the Impulse Response of the<= FONT size=3D2>output filter as the Impulse Response of the seismometer and = commentingadvers= ely on the damping. I have now found the separate Boi= se Poster athtt= p://earth.boisestate.edu/pal/files/2011/12/slinky_poster.pdfI HAVE ALSO FOUND THE DAMPING INSTRUCTIONS FOR YOUR SLINKY :- "Forthe damper, the magnet should be even with the top of the copper, or =C2=BC"<= PRE style=3D"FONT-SIZE: 9pt">inside. Properdamp is when the spring will bounce only onc= e, or twice,not repeatedly". Precision ? Er, hmm ? Have= you measured the change inthe vertical position of the sensor magnet= with changes in temperature? A change in the armature position could effect both the dampin= g and the = sensor output. A long spring can make quite a good thermomet= er !So, does= a constructor just have to guess what damping is correct ?There seem= s to be NO easy way of finely adjusting it, other tha= n taking
the column apart repeatedly and reposit= ioning either the copper tube orthe damper magnet? If you give an impulse to = the seismometer and thesignal trace is <= FONT size=3D2>twenty units high, the trace should then just cross the<= PRE style=3D"FONT-SIZE: 9pt">zero line by 1 unit and tailoff without any furt= her oscillation. The damping needs to be ~0.7 c= ritical to give a 'flat with ve= locity'character= istic and reasonable accuracy.I am really rather puzzled, since K= asper says "I can confidentlystate the device i= s not under-damped by any definition...." Maybe, but is it CORRECTLYdamped at ~0.7 Critical, please ? And how are operators = supposed to set this up reasonably accurately? = FONT>I got con= fused over the Impulse Response, MOSTLY because EVERY ONE ELSE,professional and amateur, us= es Butterworth Filters! These give minimum phasedelay and PRESERVE THE SEISMIC WAVEFORM FOR= ANALYSIS. Kasper says that the= phase delay is only ~0.4 seconds, but this is a WHOLE CYCLE = ERROR at 2.25 Hz!And the delay will NOT be constant with frequency. S= teep cut filters likethis seriously distort a c= omplex wave form. A ringing response is definitely = FONT>MOST undesirable. <= /FONT>I note that three magnets are used, two for the = detector coil and one forthe damping and that all three have the same orientation, threaded= on a mildsteel = eye bolt with steel nuts. Moreover, the eyebolt is di= rectly hooked ontoa steel spring, giving a 'str= ay magnetic field detector' all of 20" long! WOW!I suggest that you make a new seismo= meter using BRASS studding and 4 ringmagnets, a N/S pair facing a S/N pair, to give minimum= external magnetic fieldcoupling, like you did on earlier models, which had a brass S hook.= Use ahacksaw to cut a sl= it in one end of the studding and glue a vertical sta= inless steel washer into it with 2 part acrylic glue. You may want to = drill / fileboth= sides of the washer to round off the inside edge. Hook this onto the<= PRE style=3D"FONT-SIZE: 9pt">spring wi= th a blob of BluTack or similar to prevent it sliding noisily =around. = BluTack is a putty like adhesive which never sets hard. Thisarrange= ment should effectively separat= e the steel spring from the magneticfields. It is also likely to increase both the sensitivity and the damping,=since no = flux will then be lost through the mil= d steel studding. You mightthen be able to 'get= away' with just two ring magnets ? The three brass bolt mountings of the coil and t= he damper tube don'tlook very secure and there is no automatic alignment. Could you not fi= tcircular Alumin= um sheets on an extended coil tube and stick them on withPVC cement / PVC rings ? The outside of th= e disk could have three short <= /FONT> springy strips ben= t at right angles to align it accurately inside theacrylic tube and the inside= edge could have 3 slots to limit any inducedcurrents.Looking at the amplifier circuit, the Slinky sens= or coil is includedin the feedback loop, allo= wing force feedback, instead of being simplyconnected to an earth ref= erence and the +ve input of the opamp. This = PRE>increases the noise. A 200= 0 Ohm coil of copper wire makes= quite a goodthe= rmometer, as well as a pick up for RFI !<= FONT size=3D2> The period of a circular section spring vertical pendulum= is givenby T =3D 2.Pi.SQRT[2.m.l.r^2 / Pi.n.a^4] : a=3Dwire radius, = r=3Dradius of thespiral, l=3Dlength of the wire, m=3Dmass, n=3Dtorsional modulus. = PRE>It is possible to compensate for up t= o about 10x the natural period ofa pendulum electronically, before ru= nning into serious noise problems. Toget a fully compensated period of 20 seconds, the spring vertical would=need to = have a natural period of 2 seco= nds. Increasing it to 1.5 secondswould allow a compensated period of = 15 seconds. Let me know if you wantto do this ? Try out the long peri= od extension filter built into AmaSeis?It would be most desirable to detect S waves at t= heir full amplitude.Schools and amateurs need to measure the distance= to the quake using the =P to S wave time delay a= nd the tables / graphs.The amplitude response of 20 second waves = is down by about a factorof 400 on a 1 second sensor. Being able to locate quakes by their Rayleighwaves and then search for the P and S waves down in the background noisemay triple the= number of quakes you can detect. From personal experience,<= FONT size=3D2>increasing the compensated period of an= AS-1 from 4.5 to 15 seconds made asimply HUG= E difference to the detection r= ate !&n= bsp;I admit being pu= zzled as to why, if you needed an AS-1 replacement,that you didn't ju= st find a couple of retired mechanical and electronicengineers to pro= duce a good updated instrument ? It could make a= welcomeboost to= their retirement incomes and they would have a part timeoccupation, preventing them from g= etting bored != FONT>THE LEHMAN.= FONT>=Ted, you might find it int= eresting to look at the Lehman that I useat http://www.jclahr.com= /science/psn/chapman/2008%20lehman/index.htmlIt has a 0.016" thick SS hard plane roll= ing on a 1/2" SS bearing for thesuspension. I can EASILY set it up fo= r 30 seconds period and I have runit OK at 60 seconds ! A plane rolli= ng on sphere gives a better geometrythan your design of a sphere roll= ing on a plane. = FONT>I note th= at you seem to have had difficulty in getting even a 20second period = on your older Lehmans ?This can be due to either the top or= the bottom suspension, or both.= The top suspension, being a wire in your case, needs to be v= ery firmlylocate= d / fixed, preferably between the edges of clamping d= isks, as on myLe= hman. On the 8020, maybe you could fit an L bracket t= o the top of the= vertical column and mount eccentric washers and a winding bolt on it,similar to my system ? The phot= o of the top wire looks as if it is foulingon the edge ofthe dome nut. It needs to be cleanly clamped. I suggest that you consider a 'V= ' Al strip brace between the verticalcolumn and the base cross bar to improve the 'side to = side' stability andincrease the stable period. One problem with Alu= minum is that it is notvery rigid and seismometers need to be 'rock solid' !I note that your NEW type Le= hman is 38" long, 24" wide, 26" high and
has a 32" boom. This may be s= tarting to get a bit 'willowy' with only a1/2" OD Al boomand a 45mm square Aluminum frame. Is this = an attempt to inc= rease the period ? Sorry, but f= rom experience I think that you wouldvery likely get a better performance from a smaller modified U = channelversion. = As I mentioned, I have run my 2= 2" boom SEP prototype at 60seconds period. The practical limit in per= iod is due to natural ground tilt driving the arm off scale. What is the cost of your U Channel frame as opposed to= that of the8020= 45mm square frame, please ? = FONT>The DIY 'EASY' = LEHMAN that I designed uses 3" x 1" Aluminum <= FONT size=3D2>U Channel,but the sides were upward / inward facing (op= posite to yours) with 6"corner plates and a tra= pezoidal end plate, 3" at the top, 8" at the bottom,7.5" high, were provided. The base bar is 30" lo= ng and the two end barsar= e 18" long. The arm is 56 cm to the CofG, giving a = FONT>1.5 second = pendulum.The lower bearing is a 3/8" SS bolt drilled to acone on the thread end and mounted through the upright with= two sets of nuts and washers. A 3/8"SS ball be= aring is stuck in the cone with acrylic adhesive and = the
position of the bolt is adjusted to give a <= /FONT>suitable period. Thisconstruction is QUITE ADEQUATE for periods of 30 seconds. = If you wantlonger periods, you need to stabilise the central position of the arm with <= /FONT>a photocell positi= on detector, an amplifier and a feedback coil to eliminatethe natural long period tilt drifts. This system ca= n also enable you to seet= he daily Earth tides !Back in about 2008, Meredith Lamb= , Charles Patton, Brett Nordgren andI co-operated to write a paper on= bearing systems for amateur seismometers.See http://www.myeclectic.info/SeismoPivots/seismopivots.htm These allworked well It was reported on PSN, = but it does not seem to have got ontothe reference list. Apart from various wire and foil systems, the tw= otypes which wor= ked OK were a hardened plane = FONT>rolling on a hardened sphereand hardened crossed = rods. We tried the ball + two rods bearing type thatyou used, but it did not work at all well. One = problem is that a sphererolling on two rods has a small, but critical, rotation about the pointof contact, giving some fricti= on. The 8020 seismometer uses a SS bolthead as the flat surface. Sorr= y, but austenitic stainless steel is justNOT hard enough ! You DO NEED HARDENED martensitic= stainless steel !Also, you don't mention that Stainless Steel Ball Bearings AREREQUIRED for seismometer suspension systems ! <= FONT size=3D2>Ordinary steel ballbearings only = last a short time before corroding at the contact poi= nt.I use a <= /FONT>1/2" OD SS Bearing on the vertical column and a hard 1= 6thou"thick SS <= /FONT>flat plate, cut from a Swann-Mor= ton Skin Graft Blade, stuck to =the end of the boom with TWO COMPONENT ACRYLIC G= LUE. It has a strengthof 3,500 psi and when set, IT REMAINS SLIGHTLY = FLEXIBLE. It will sticksoft metals like Aluminum, Copper, Silver etc = just fine. Epoxies don'tbond well and they most= ly cure brittle hard. There may be some 'breakoff' DIY SS knife blades available, but most of them are ordinary steel.You can also use Tungsten Carbide. SS = bearings are supplied bySmallparts, now called = www.amaz= onsupply.com
For crossed cylinders, the cheapest so= lution is to use fully
hardened martensitic stainless steel rod. You may have t= o polish it before and after cutting andhardening it. The SEP uses 1/8" Tungsten C= arbide needle roller= s, but these are quite expensive. You <= FONT size=3D2>can alsobuy 1/8" Tungsten Carbide drills and use the shanks. WC Circuit boarddri= ll shanks are also available. Do you have any local m= achinistcompanie= s who use WC milling cutters ? = Anyone in the University ? Old /broken cutters are usually just throw= n out.&nb= sp;On your early Leh= man, when the ground moves, the drag force on thedamping plate will try to rotate the boom, since it is offset. This can=alsorotate the sensor coil, giving false responses. The rotation problem has been corrected on the 8020 Lehman. On the SEP, the brassmass is = offset and the damping blade li= es on the line joining the CofG= to the lower bea= ring, thus dramatically reducing any rotational coupl= e.I also use= a V cable made from 7 core SS nylon coated 40 lb fishingtrace, which= is supplied with crimp tubes. You can buy it in up to 200lb rating. = It is very effective at preventing any rotation of the arm. I suggest that you try using a V= cable with a side extension to the
mass to give a separation of the a= ttachment points of about 6" and alsorun the top disk of the V, very = close to the pillar..On the = Boise web site, you say that Cu and Al are paramagnetic. This=is incorrect. Cu is very nearl= y neutral. Al has a significant diamagneticmoment, which can cause an Al damping blade to p= ush a Lehman arm toeither stop if the blade is not wide enough and = it comes close to theedge of a damping magnet. It is difficult to get= ~pure Al and it hasnearly twice the resistivity of Cu. 'Strong' Alu= minum alloys have verymuch higher resistivities= ..Sheet Cu ma= kes a much better damping blade ! Instead of the 3/16" Alalloy used on the 8020, I suggest = that you try 1/32" thick Copper sheetand also mount BOTH the damper blade and the sensor co= il with two bolts?http://www.ksmetals.com stock 0.= 016" and 0.025" Cu sheet, if you can'tbuy it locally. You should be able to reduce the reco= mmended clearancebetween the magnets and the coil from 3/16" to 1/16". This will increase <= /FONT>the voltage output= of the coil and also increase the damping. Instead ofusing Perspex sheet for the sides of = the coil, have you tried using thinfibreglass sheet / thin circuit board ?Instead of using an electric = drill for coil winding, I use a hand drill clamped horizontally on a work bench and count th= e number of turnsof the handle. Count the number of teeth on the two gears to find theturns ratio. A hand d= rill is MUCH easier to control and it doesn't takelong to wind a coil. I use a plastic tub= e of the smallest diameterheatshrink sleeving, pre-shrunk, to guide the wire.Brass tends to corrode Aluminum i= n a damp atmosphere. Stainless<= FONT size=3D2>Steel bolts and studding are not difficult to acquire and are= not veryexpensi= ve. I use all SS nuts, bolts, washers and spring washers, boughtfrom = a marine chandler / boat supplier, or McMaster-Carr.I noticed that you are using 0.016"= stainless steel wire for thesuspension. Have you checked if it is austenitic or martensiti= c wire ? Austenit= ic SS is NOT suitable for a wire which flexes. www.daddario.comproduce a wide range of steel music wires, usually Nickel plated for = rust protection. It is MUCH stronger and works well. What does you loc= almusical instru= ment shop stock / supply ?The levelling screws are all 1/4" UNC having only 20 tpi. If = you used1/4" UNF= set screws, you would get 32 tpi, which would make the levelling = PRE>easier to adjust. On the 8= 020, I suggest that you separate out the levelknob from the arm. Shorten the boom a few inc= hes and mount the knob on theend of the base bar, well clear of the mass. It will give fine= r period adjustme= nt. I suggest that you fit a wavy spring washer between the nutand the washer on top of the= base bar to keep the thread under tension atall times. The frame will expand and contract = with changes in temperatureat a different rate to the ground, so the seismometer needs to b= e supported on sm= ooth flat plates so that the ends of the screws can slide a few thou..=The glass case is fi= ne for the demonstration, but amateur seismometercases are usually made from 2" Celotex i= nsulation board. An Al sheet withheating resistors is fitted inside t= he top of the case. This maintainsa positive temperature gradient in the vertical air column above theground, inhibiting any convectiion. If you don't do this, when the airtemperature falls below the ground= temperature in the early hours of the<= PRE style=3D"FONT-SIZE: 9pt">morning, rolls of cold air peel off the inside of the top of t= he chamber,fall = to the ground and push the arm = about. This is seriously n= oisy !The di= fficulties of temperature control increase with the physical<= PRE style=3D"FONT-SIZE: 9pt">size of a seismometer. The Spre= ngnether horizontal had a 14" boom andwas limited to a period of 15 seconds by tilt instabilities for resetintervals of several months. The SEP was designed with a 22" boom andthe drift is checked weekly= ..=I am puzzled that you reco= rd data with AmaSeis and then analyse itwith Winquake ? AmaSeis has a perfectly good set of= filters, INCLUDINGa digital period extension filter that could be = used to extend the periodof the Slinky ! What is the reasoning, pleas= e ?Ted, feel= free to contact me if you are interested in a discussingany of these topics ? I also run a= small machine shop and have quite a lotof experience of making thing= s.Regards,Chris Chapman (Physics) [ Top ] [ Back ] [ Home Page ]