From: Bob HolmstromTo: psnlist Sent: Wed, 3 Oct 2012 23:47 Subject: Re: Digest from 10/02/2012 00:00:07 I have seen silicon wafers etched thin enough that they could be rolled= up like cigarette paper - perhaps silicon in that form could be used as a = flexure. Yes, metal construction is the norm, but it seems to be in the sp= irit of this list to consider alternative methods and materials. Hi Bob,=20 I don't have a figure for the tensile strength of Silicon, but I doubt = that any sheet thin enough to bend easily would be able to support a pendul= um weighing several kilograms. Sheet silicon is also liable to brittle frac= ture under shock loads. =20 The book by R. V. Jones - "Instruments and Experiences" is an organized col= lection of many of his articles on instrumentation. Each section has overv= iew material that I have not seen published elsewhere. Unfortunately it is= not easy to find considering that it was published in 1988. bookfinder.co= m has a few copies available from $170 to $850. Worldcat.com shows 12 libr= aries in the UK that hold a copy. A similar number of libraries have Bob M= atthys' book. Thanks. Er... But 'Worldcat.com' is about boats - catamarans ! Could yo= u check your lists for the World Library Web Site reference, please ? I mentioned Aluminium Silicon Bronze because Bob's experiments seem to show= that it comes back to the same size after temperature cycling than most ot= her materials suitable for pendulum bobs.=20 Sorry, I was thinking about metal rods suitable for pendulum arms !=20 Invar is not a panacea as a pendulum rod material. It requires elaborate t= hermal/shock treatment to achieve good performance, and if mechanically sho= cked or machined, needs to be retreated. Invar is also infamous for unpred= ictable micro jumps in dimension. See Dieter Riefler's book for details. = Zerodur does not come in thin rod form of sufficient length for 1 second pe= ndulums. Carbon rods have very good thermal performance and machine abilit= y, but the binder epoxy is humidity sensitive (e.g. the rod gains and losse= s weight which effects the pendulum period) - attempts to slow the process = by coating the rod with something just slows down the process - no coating = is perfect. Fused quartz is a very good material, note tubing is far stron= ger than rod, but end fittings must be attached with adhesive, or with pins= in cross drilled holes.=20 How about using Super Invar, which has 5% Cobalt added ? It has greatly= improved properties. I appreciate that the epoxy binder in carbon fibre rods will have a sma= ll water absorption, but this should be baked out for any vacuum applicatio= n. If you want to use the rod in damp air, it can be coated with Epoxy varn= ish and then rolled in the thicker Aluminium foil used for cooking Turkeys,= to give a permanently impervious coating - this will not significantly eff= ect the linear expansion coefficient and the coating can be expected to be = 'perfect'. It should be painted after assembly to prevent corrosion.=20 In my last Email, I mentioned the use of two part Acrylic glues for sti= cking glass, quartz, carbon fiber and most metals. It has a tensile strengt= h of about 3,500 psi and it cures very slightly flexible, unlike Epoxies wh= ich mostly cure brittle hard. Eg. Holdtite ST3295 has the consistency of Go= lden Syrup and it can be used for about 5 minutes after being mixed. It set= s in under 10 minutes at room temperature and cures completely in 24 hours.= Several other brand names are available. It has an extremely high resistan= ce to shock loads. I suggest that you avoid ST3298 type Acrylic adhesive, w= hich is a thick paste. =20 Yes, I am very familiar with the Littlemore clock. I believe that Teddy Hal= l's first published presentation on the clock was at a NAWCC Symposium in C= leveland, USA. The clock is currently in storage at the NAWCC Museum - it a= rrived in not very good shape. Mice had moved in to the computer and had m= ade a real mess. Documentation of the software that controlled the clock s= eems missing. If anyone knows someone that worked on the project for Teddy= - I would love to get the contact information. The hard drive from the co= mputer is currently in the process of trying to extract any information tha= t may still be on it. Slow process, because of the components that made up= the computer. =20 Which make and model of computer was it, please ? Is it just the progr= am documentation which is missing, or the software as well ? Could you adve= rtise for an identical old computer and then plug in the old hard drive ?= =20 There is some controversy about the performance of the clock. Teddy used a= quartz controlled oscillator to measure the amplitude of the pendulum and = adjusted the drive force to compensate. Some people are concerned that the= quartz oscillator drove the performance of the clock. It is not a simple = issue - Tom Van Baak covers the problem very well in his article > http://l= eapsecond.com/hsn2006/hybrid-pendulum-1.pdf=20 The problems may be fully described, but what it does lack are mathemat= ical formulae / estimates of the accuracy / errors.=20 What puzzles me is why pairs of large area Silicon photocells are not u= sed to monitor the swing angle ? This is dead easy to do and you should be = able to get an accuracy of better than 0.1 micron ! I can get a detection s= ensitivity of about 14 nano metres with an optical seismometer detector usi= ng VTD34 photocells.=20 A 1 second pendulum back and forth more than 31 million times per year over= much larger amplitudes than a seismometer. Wear, particulates, variations= in spring constant due to temperature, etc are all problems to be solved. = The spring I measured was a standard Syncronome spring - it is not unusual= ly thick.=20 I appreciate the difference between the Synchronome suspension, both ty= pes of 'normal' pendulum clock suspensions and a clamped flex foil - Cardan= type. What I am querying is just how do you determine the precise pendulum= length with the Synchronome suspension and does this vary with the swing a= ngle ? =20 The Short and Fedchenko suspensions have a feature that might be worth expl= oring for seismometers. The springs are not flat stock clamped between "che= eks" as are most pendulum suspensions instead the springs are ground from t= hicker stock - i.e. they have thick ends. Stuart Smith's book on Flexures = has as section on the issues that arise at the clamping points. The Fedche= nko clock is the only production that incorporates a working circular error= compensation mechanism. Most other attempts have failed because the extra= mechanism is very difficult to temperature compensate. What 'other' methods have been tried to compensate for 'circular errors= ', please ? I would be very interested to hear of anyone trying out a clock suspens= ion using the rolling crossed foil system that I suggested. This uses two p= arallel circular rods with 2 'S' shaped and 2 'mirror S' shaped foil strips= wrapped around them. The foils are stuck to the top and bottom of the rods= with two part acrylic adhesive. Since the rolling foils give equal torques= in opposite directions, there is NO NET torque. Since the foil bending rad= ius is large, the intrinsic noise is very low - and there AREN'T ANY clamp = points to give problems !! You do need to use a long sleeve dust cover. The= performance can be expected to be superior to clamped crossed foils.=20 Another type of suspension which would definitely be worth trying out i= s rolling crossed cylinders, preferably using circular artificial sapphire = rods. =20 Again - the goal here is cross fertilization - perhaps something from the w= orld of horology might be of use to seismologists. =20 Bob Regards,=20 Chris Chapman =20 =20 =20 =20 =20 =20 =20 =20 =20 =20 From: Bob Holmstrom <hol= mstro@..........>
To: psnlist <psnlist@..............>
Sent: Wed, 3 Oct 2012 23:47
Subject: Re: Digest from 10/02/2012 00:00:07
= I have seen silicon wafers etched thin enough that they could be rolled up= like cigarette paper - perhaps silicon in that form could be used as a fle= xure. Yes, metal construction is the norm, but it seems to be in the = spirit of this list to consider alternative methods and materials.
Hi Bob,
I don't have a figure for the tensile strength of Silicon, bu= t I doubt that any sheet thin enough to bend easily would be able to suppor= t a pendulum weighing several kilograms. Sheet silicon is also liable to br= ittle fracture under shock loads.
The book by R. V. Jones - "Instruments and Experiences" is an organize= d collection of many of his articles on instrumentation. Each section= has overview material that I have not seen published elsewhere. Unfo= rtunately it is not easy to find considering that it was published in 1988.= bookfinder.com<= /a> has a few copies available from $170 to $850. Worldcat.com shows 12 libraries in = the UK that hold a copy. A similar number of libraries have Bob Matth= ys' book.
Thanks. Er... But 'Worldcat.com' is abo= ut boats - catamarans ! Could you check your lists for the World Library We= b Site reference, please ?
I mentioned Aluminium Silicon Bronze because Bob's experiments seem to= show that it comes back to the same size after temperature cycling than mo= st other materials suitable for pendulum bobs.
Sorry, I was thinking about metal rods = suitable for pendulum arms !
Invar is not a panacea as a pendulum rod material. It requires e= laborate thermal/shock treatment to achieve good performance, and if mechan= ically shocked or machined, needs to be retreated. Invar is also infa= mous for unpredictable micro jumps in dimension. See Dieter Riefler's= book for details. Zerodur does not come in thin rod form of sufficie= nt length for 1 second pendulums. Carbon rods have very good thermal = performance and machine ability, but the binder epoxy is humidity sensitive= (e.g. the rod gains and losses weight which effects the pendulum period) -= attempts to slow the process by coating the rod with something just slows = down the process - no coating is perfect. Fused quartz is a very good= material, note tubing is far stronger than rod, but end fittings must be a= ttached with adhesive, or with pins in cross drilled holes.
How about using Super Invar, which has = 5% Cobalt added ? It has greatly improved properties.
I appreciate that the epoxy binder in carbon fibre rods = will have a small water absorption, but this should be baked out for any va= cuum application. If you want to use the rod in damp air, it can be coated = with Epoxy varnish and then rolled in the thicker Aluminium foil used for c= ooking Turkeys, to give a permanently impervious coating - this will not si= gnificantly effect the linear expansion coefficient and the coating can be = expected to be 'perfect'. It should be painted after assembly to prevent co= rrosion.
In my last Email, I mentioned the use of two part Acryli= c glues for sticking glass, quartz, carbon fiber and most metals. It has a = tensile strength of about 3,500 psi and it cures very slightly flexible, un= like Epoxies which mostly cure brittle hard. Eg. Holdtite ST3295 has the co= nsistency of Golden Syrup and it can be used for about 5 minutes after bein= g mixed. It sets in under 10 minutes at room temperature and cures complete= ly in 24 hours. Several other brand names are available. It has an extremel= y high resistance to shock loads. I suggest that you avoid ST3298 type Acry= lic adhesive, which is a thick paste.
Yes, I am very familiar with the Littlemore clock. I believe that Tedd= y Hall's first published presentation on the clock was at a NAWCC Symposium= in Cleveland, USA. The clock is currently in storage at the NAWCC Museum -= it arrived in not very good shape. Mice had moved in to the computer= and had made a real mess. Documentation of the software that control= led the clock seems missing. If anyone knows someone that worked on t= he project for Teddy - I would love to get the contact information. T= he hard drive from the computer is currently in the process of trying to ex= tract any information that may still be on it. Slow process, because = of the components that made up the computer.
Which make and model of computer was it= , please ? Is it just the program documentation which is missing, or = the software as well ? Could you advertise for an identical old computer an= d then plug in the old hard drive ?
There is some controversy about the performance of the clock. Teddy u= sed a quartz controlled oscillator to measure the amplitude of the pendulum= and adjusted the drive force to compensate. Some people are concerne= d that the quartz oscillator drove the performance of the clock. It i= s not a simple issue - Tom Van Baak covers the problem very well in his art= icle > http://leapsecond.com/hsn2006/hybrid-pendulum-1.pdf <= /a>
The problems may be fully described, bu= t what it does lack are mathematical formulae / estimates of the accuracy /= errors.
What puzzles me is why pairs of large area Silicon photo= cells are not used to monitor the swing angle ? This is dead easy to do and= you should be able to get an accuracy of better than 0.1 micron ! I can ge= t a detection sensitivity of about 14 nano metres with an optical seismomet= er detector using VTD34 photocells.
A 1 second pendulum back and forth more than 31 million times per year= over much larger amplitudes than a seismometer. Wear, particulates, = variations in spring constant due to temperature, etc are all problems to b= e solved. The spring I measured was a standard Syncronome spring - it= is not unusually thick.
I appreciate the difference between the= Synchronome suspension, both types of 'normal' pendulum clock suspensions = and a clamped flex foil - Cardan type. What I am querying is just how do yo= u determine the precise pendulum length with the Synchronome suspension and= does this vary with the swing angle ?
The Short and Fedchenko suspensions have a feature that might be worth= exploring for seismometers. The springs are not flat stock clamped between= "cheeks" as are most pendulum suspensions instead the springs are ground f= rom thicker stock - i.e. they have thick ends. Stuart Smith's book on= Flexures has as section on the issues that arise at the clamping points. &= nbsp;The Fedchenko clock is the only production that incorporates a working= circular error compensation mechanism. Most other attempts have fail= ed because the extra mechanism is very difficult to temperature compensate.=
What 'other' methods have been tried to compensate for 'ci= rcular errors', please ?
I would be very interested to hear of anyone trying out = a clock suspension using the rolling crossed foil system that I suggested. = This uses two parallel circular rods with 2 'S' shaped and 2 'mirror S' sha= ped foil strips wrapped around them. The foils are stuck to the top and bot= tom of the rods with two part acrylic adhesive. Since the rolling foils giv= e equal torques in opposite directions, there is NO NET torque. Since the f= oil bending radius is large, the intrinsic noise is very low - and there AR= EN'T ANY clamp points to give problems !! You do need to use a long sleeve = dust cover. The performance can be expected to be superior to clamped cross= ed foils.
Another type of suspension which would definitely be wor= th trying out is rolling crossed cylinders, preferably using circular artif= icial sapphire rods.
Again - the goal here is cross fertilization - perhaps something from = the world of horology might be of use to seismologists.
Bob
Regards,
Chris Chapman