The Micrometer

A two-monthly review of the engineering of time

Issue 10   March - April 2018

A rotary broaching tool

In the last issue I presented my design for a rotary broaching tool*.  One correspondent wrote: "Your device for making square holes is interesting, but the ones I need to make regularly are about 1.2 mm square, so some miniaturisation would be required...".


*The design is still available in this issue as a download from the Archive page.


     To me this was a challenge, especially as I have rarely found a large lathe attachment a bar to making small components.  To go some way towards satisfying myself that small square holes could be cut, I chose to make a 1.5 mm square broaching tool, which is the smallest size listed by the Somma Tool Company Inc. of Waterbury, CT.  To give the bit increased strength I also made it from the shank of a blunt high-speed steel 2.2 mm diameter drill, snapping off the drill at the root of the flutes.  The drill shank was set into a silver steel holder using Loctite, with the broken flute end outermost to ensure that the hardest part of the drill was being used for cutting.  Taking care not to overheat the tip, the cutter was ground in the Quorn tool and cutter grinder with a slightly reduced back taper (1½° cf. 2°) to give an overall length of 3.5 mm to the tapering square.

     The first photograph shows a trial 1.5 mm square hole being formed in a 3/16 in. outside diameter length of bronze rod (inset, right).  A pilot hole of 1.6 mm diameter was drilled to a depth of 4 mm before presenting the rotary broach to the hole.  After 'drilling' the square hole, the end was chamfered with a 90 degree countersink bit to leave the overall depth of the square hole at 3 mm.  A decorative taper was turned to (perhaps) represent a large pocket watch key.

     Encouraged by this square hole I then made a 1.2 mm a/f broach tool and cut a trial square hole in a length of 4 mm diameter brass (inset, middle and left).  In brass and at this small size, I found it was necessary to press the tool bit tip into the brass slightly before starting the lathe spindle rotating; if the tool was presented with the spindle rotating the square tool bit initially tended to ream the hole as it accelerated up to speed rather than rotating with it (inset, middle).  I also found a single cut to the full depth best (i.e. not withdrawing the tool bit); if withdrawn and re-entered, without accurate realignment there was a tendency to start a second square hole.

     The second photograph shows a selection of the cutting tools I have made to date, the 1.2 and 1.5 mm square tools being compared with a larger ¼ inch across flats square tool (inset).


     After making the two small tools mentioned, I decided to see if I could make a standard size 1/4 in. hexagon socket cutter in high speed steel (HSS) suitable for a standard hex. drive.  Sourcing 8 mm diameter HSS (the size used by Somma Tools for their rotary broach bit shanks) is not as easy as finding 5/16 in. diameter HSS but can be found.   The tool bit was shaped up using the Quorn (first photograph), and cut to length with a standard 4 inch diameter abrasive cutting disc.  This same disc was also used to provide the flat for the set screw (second photograph).

     On test, the result was a moderate success, but it did show that spiralling was a potential problem.  In the 1/2 in. diameter brass bar shown in the photograph below (pre-drilled to 0.252 in. (6.3 mm)), the hexagon hole is about 7/16 in. deep, and at its cutting tip my HSS cutter measured 0.252 in. across flats.  The Torx bit shown alongside had an across flats measurement of 0.249 in. and started to bind in the hole at a little over half-depth, both my cutting tool hole and the Torx shank having similar tolerances to those used for hex. socket cap screws and their wrenches.  In a second test piece, reversal of the lathe at mid-depth solved the problem, the Torx bit now entering snugly but freely to the full depth.  This suggests two 'rules of thumb':

  • The lathe should be reversed at mid-depth for holes deeper than the across flats dimension, and
  • A rotary broach should not be used for holes deeper than twice the across flats dimension.

Asymmetric crutch

In the last issue I included a download of my thoughts questioning whether the introduction of asymmetry in the action of this crutch has compromised the rate-keeping performance of the Graham deadbeat escapement.  It is clear from correspondence received that the issue of the pendulum being very slightly displaced to the right when at rest is giving some readers concern, arguing that when the crutch starts moving to the left, the pendulum appears to provide a tiny leftward impulse as it returns to its truly vertical position.

     I believe this is incorrect; this is not new, added energy but merely the release of energy stored from its final swing to the right.  In other words, removal of a weight (viz. the horizontal component of the weight of the crutch) will not provide leftward impulse but merely release previously stored energy which must be restored by the rightward movement of the crutch on its next cycle.

     I have added a post script to the article which is included in the download (file name Asymmetric-6) available from the Archive page in this issue of The Micrometer.

Modulus of Elasticity (Young's Modulus)

Many times have I tried to convince unbelievers that, to a first approximation, the modulus of elasticity of steel and its alloys does not change.  Look it up, and you will find that irrespective of the alloy, the modulus of elasticity is around ±5% of 200 GPa (200 x 10⁹ N/sq.m.).  Nor does it change to any extent over the normal temperatures experienced by a clock (see graph).  But the assumption of constant modulus does depend on the application, and for a mainspring such variation in Modulus of elasticity with temperature as there is (defined by the thermoelastic coefficient) is pretty irrelevant considering all the other variables that might affect the torque reaching the balance wheel and altering its amplitude.

     But let us take a step sideways for a moment and consider a balance spring and the error in a standard 18,000 beats per hour lever escapement beating at 17,999 beats per hour.  This 1 beat per hour error represents a percentage error of about 0.005% but will result in the clock losing 5 seconds per day (or over half a minute a week).  Now this is starting to become significant, and the modulus of elasticity may well vary by 0.005% with changes in temperature of even just a degree or two.

     The real point I am trying to make is that the laws that apply to the two spiral springs commonly found in clocks and watches need to be interpreted quite differently in their practical application, and one must consider mainsprings and balance springs separately.


      I suspect much of the confusion stems from the use of the word ‘elasticity’ in a general sense.  A higher yield steel springs back to its original shape after a greater initial deflection than a lower yield steel; it is thus seen as having a greater elasticity.  But this is not due to an increase in the modulus of elasticity; it is simply due to the steel remaining below its higher elastic limit (below its yield strength) to a greater deflection.

Skeleton clocks

     Not my favourite, primarily because of the acres of lacquered brass that the owner frequently wants re-polishing.  But I must admit they make attractive mantel clocks in the correct room setting, and one such clock that I did take on is illustrated (it was included briefly in the May-June 2017 issue of The Micrometer to illustrate the manufacture of a new repeat cord knob).  As is my custom, I leave the glass dome at the owner's home as to replace an accidentally broken dome is both difficult to source and very expensive.

     This customer did want a full re-polish which added hugely to the work involved in the service.  Rather than re-lacquer the brass, we agreed that I would treat all surfaces and edges (4.5 mm thick) with micro-crystalline wax ('Renaissance Wax') after removal of the old lacquer and repolishing.  This might not give such a long life as lacquer, but it would be run-free (which I find is not easy to achieve on multi-faceted plates), repairable if the brass were rubbed during winding, and make future refinishing far easier than having to remove ageing lacquer and the cost of huge quantities of grit paper (ranging between 600 and 2000 grit) clogged with lacquer.

     The movement was in good condition but all oils were heavily congealed.  The shouldered plate pillars had not been correctly seated in their plate holes by a previous repairer and, once correctly seated, the negative end-shake of two previously re-bushed arbors needed to be adjusted.  This adjustment was done by simply pressing the bushes into their correct axial position so the inner face of the bush aligned with the face of the plate.  The centre arbor back pivot needed replacement, but otherwise the remaining work needed was largely cosmetic.

     All screw heads were polished and re-blued (no point in having shiny plates with rust-spotted and bruised screw heads showing), and the black paint trowelled onto the hands was removed and the hands correctly polished and heat-blued before lacquering.  It looked good, so to complete the job I persuaded the owner to buy a good-looking key of the correct size, my view being that the Meccano-type clockwork motor key was wholly unfit for purpose.




Tubal Cain

The pen-name adopted by the late TD (Tom) Walshaw, Tubal Cain was an eminent writer in Model Engineer magazine between 1971 until his death in 1998 and from whose writings I learned much of my engineering craft. He was also educated at the former incarnation of Loughborough University, which is the university at which I gained my degree in Mechanical Engineering.

     Tubal Cain is, perhaps, the only engineer (or, more strictly, metalworker) named in the Bible.  To paraphrase Genesis 4.22 "Tubal-Cain was the son of Zillah, who forged all kinds of tools out of bronze and iron".  Some versions call him an artificer, and another variation suggests he instructed all who work in bronze and iron, though the formella* shows Tubal-Cain making rather than instructing.

     Of course, this perhaps ignores the obvious in that both Jesus and Joseph, husband of Mary were both carpenters (Mark 6:3 and Matthew 13:55).  Carpenter translated from the ancient Greek, I understand that τέκτων could also refer to craftsmen working metals and stone.  Anyway, with numerous examples of prodigious engineering structures in wood, why should a worker in wood not be considered an engineer?  Those familiar with the Greek alphabet should have no difficulty in enunciating τέκτων and hence deducing the everyday English word derived from it (answer at the bottom of this page).


*  The formella (or tile) was made around 1335 and is attributed to the sculptor Andrea Pisano of Orvieto, which depicts him forging tools on his anvil with the furnace in the background and the bellows to his rear.

Photo: Sailko, Wikimedia Commons

The clock auction...

The first time I ever bid at an auction room was about 15 years ago.  I was in need of clocks in order to meet the 'Record of Repairs' criteria for my successful achievement of full Membership of the BHI.  The Record of Repairs not only stipulated that a number of clocks of different types had to be serviced and/or repaired, but that a range of specific operations must also be carried out.  This meant that, like collecting cards from a cereal packet as a child, the first 50% were easy to achieve, but as one neared the end of the course of study, finding a clock needing the last one or two specific operations became increasingly difficult.

     Buying at auction can be fun (especially if caterers are in attendance) and I still attend clock auctions being, on average, successful in offering the winning bid about one time in three.  For a non-dealer who is always after a specific lot (i.e. not any lot providing it is at a price low enough to provide a margin for resale) the key to success seems to be being absolutely clear as to which lots one wants and what one is prepared to pay for the enjoyment (or, in the case of undertaking one's Record of Repairs, what educational value) it will bring.

Photograph courtesy of Gardiner Houlgate




... and the BHI Distance Learning Course

With the help of the clocks bought at auction I gained my qualification over two years of study between 2005 and 2006.  The clocks were largely bought from Gardiner Houlgate at their auction rooms near Bath, and a view of just a fraction of the total number of clocks on offer at one of their their four-monthly sales can be seen in the photograph above.

     I studied for my BHI examinations at home, supported by Trevor Waddington and a couple of week-end platform escapement courses run by Paul Shrouder.  After spending the first 24 years of my life in infancy and formal education, I vowed never to sit an examination ever again, but I succumbed, primarily because I thought it was not fair on the public at large to entrust their clocks for professional repair by one purely self-taught.  My view on this developed as the course went on, and the main benefit I gained from the course was that it 'forced' me to undertake repairs that I never thought I would be able to do, viz. platform escapements with their less than 0.5 mm diameter screws and 0.15 mm diameter hardened pivots.

     That I achieved success was in no small part due to Mr Shrouder's easy but mischievous style of teaching: "What do you think of my straightened-out hairspring*, Paul?" as I showed him my neatly spiralled hairspring from a box of close-on ten thousand from street lamp timers (photograph below).

     "Hmmm, let me see" as he took his loupe to his eye, and with a deft flick of the wrist said "See it's still bent - do it again...!"  I did and, while far from perfect, I was so proud of my first 'repair' (below).  It is a great loss to the current generation of horology students that, shortly after my course, the BHI let Paul go.


*  No starchy 'It's not a hairspring; it's a balance spring' from Paul.  Like me, Paul considered unambiguous understanding to be the basis of communication, not uniformity of terminology.

So old it’s growing whiskers?

One of the early clocks I repaired was a small striking mantel clock which involved a technique by a previous repairer that I had not been taught as part of the Distance Leaning Course.  From a study of the end-shake* in the escape wheel arbor, it was apparent that there was none.   Rather than separate the plates to release the escape wheel arbor (and the consequential inconvenience of having to reassemble the trains after attending to the end-shake), the two upper plate/pillar taper pins had previously simply been eased so that a few strands of garden twine could be wound around the pillar shoulders before re-pinning.  Job done – or it would have been if the scissor trimmings after re-pinning had not chosen to lodge on the lubricated pallets and pivots.

     Lest there be any doubt, neither the DLC nor I recommend the use of garden twine for the improvement of end-shake...  Had I felt it necessary to undertake such a 'bodge' repair, I would have cut a slot in a brass shim washer or two and slotted it around the turned down end of the post.  As for a proper repair, where the plates had previously been re-bushed (as in this case**), the simplest way is to lightly tap the offending bush in the plate so as to increase the end shake (clearance).


*   End-shake: the axial clearance or 'rattle' of the arbor between the plates.  Side-shake is the radial clearance of the pivot in the pivot hole.

** In practise, and on the assumption the manufacturer did not use garden twine, such a problem will invariably only occur when a clock plate has been re-bushed.

Soft soldering

As with all soldering operations, success will only result if there is sufficient heat so that the solder melts on the components to be joined and not just in the flame.  An example of a bad repair to the hammer arm of a Japy Frères French clock and my 'repair of the repair' is shown below.  Starting at the left you can see the appalling attempt at securing the loose 1.2 mm diameter arm into the boss, where it is obvious that the solder has just dripped onto cool components.  There is also evidence of the remains of corrosive flux.  After cleaning up (third photograph) and fluxing, both components are heated so that the solder melts (flashes) onto both components to form a good joint.  The finished joint is then thoroughly scrubbed with a nylon toothbrush in hot detergent and rinsed in water before thorough drying.

     Incidentally there is a design flaw in these clocks, which always makes this joint prone to loosening.  Fit the arm with a conventional right-hand thread as the manufacturer did and the bend in the arm as shown top left, and the action of striking the bell (or, in this case, the gong) will tend to unscrew the arm from the boss.  This is why you will often see them repaired with soft solder.

Calendar work

With a standard French movement fitted to a cast dial plate, this Ca. 1870 clock by Delettez* illustrated has some interesting calendar work.  Triggered by the nib (just below the bell in the photograph) once every 24 hours, the linkage and calendar wheel diameters are arranged so that the 7 tooth day wheel moves one seventh of a revolution and the 31 tooth date wheel one thirty-first of a revolution.

    Simple mathematics suggest this needs a ratio between the rotation of the periphery of the two wheels of 31/7 = 4.43, and by simple inspection I think it can be seen that half of this comes from the separation of the pivot points on the centrally-pivoted vertical link, and the rest from the difference in diameter of the two ratchet wheels.  Ingenious, and far more robust and reliable than the sort of calendar work one sees in English clocks of the same period.


*  Several clocks by this maker appear to have calendar work with a mechanism of the type illustrated.  Sometimes having a large second dial also incorporating months of the year, one such clock was recently sold for a hammer price of £800 by Gardiner Houlgate at their February 2018 auction.

Quad 306...

I bought my Quad 306 power amplifier and ESL63 electrostatic loudspeakers back in 1990, and they have been in pretty much constant use for close-on 30 years.  But recently I thought the sound quality was not what it once was, and something needed to be done.  In my set-up, the 306 is connected to the line output of a Bang & Olufsen CD player and FM tuner, which are intended for use with the stylish B&O active loudspeakers (the power amplifier is built into the speaker).  Nowadays, I believe B&O use a highly efficient Class D amplifier in their speakers but, requiring semiconductors  capable of high speed switching at reasonable power levels, the Class D amplifier could not be engineered for power amplification back in the 1990.

     The Quad 306 is a reasonably efficient (up to 70% or thereabouts) Class B power amplifier with very low distortion*, but each channel does incorporate a huge electrolytic DC-blocking output capacitor in each output to the loudspeaker.  And like all electrolytics, their performance degrades with time.  Perhaps the obvious thing to do was chuck the system away and invest in new equipment, but I do have a soft spot for the Quad ESL63s, which I know can be driven by the 306.  So I bought the Quad 306 upgrade kit available from DaDa Electronics in Antwerp and, after an hour or two's work with the soldering iron, it really did improve the sound, no doubt helped by allowing the original two output and two power supply capacitors to be increased to 6800µF within the same envelope as the original 4700µF cap's (i.e. the more compact, modern construction of the electrolytics still allowed the chassis to fit in the case).


* Strictly all high quality Class B amplifiers are Class AB as some slight bias is needed to eliminate crossover distortion caused by the necessary 0.7 volt forward bias in order for a silicon transistor to start conducting. There is also something called 'current dumping' in the Quad 306 (which I do not understand), so at this point I will refrain from further comment.

     The photographs show the internal chassis after installing the upgrade components which, in addition to the 6800µF electrolytics (the four black 'dustbins' in a row - two power supply ripple filters plus two DC output blockers), also include a number of additional 0.1µF decoupling capacitors soldered on the underside of the board (inset, second photograph).  The two pairs of Class B push-pull output transistors and their drivers can be seen bolted to the heat sink at the front (towards the bottom in the first photograph).

     For those interested in reading a little more about the different classes of power amplifier, the following link may be useful:



... neat cabling...

Neat cabling is always something I strive for, be it associated with computer, television or hi-fi equipment.  For example, wherever possible I fit separate wall-plates for each UK 13 amp plug so as to avoid the use of two- and three-way adapters.  In the set-up shown in the photograph, the Quad 306 is concealed behind the right hand loudspeaker on a specially-constructed plinth to give adequate ventilation (cooling).  To get the speaker cable from the Quad 306 across the fireplace and hearth to the left hand loudspeaker, I used a pair of blank wall-plates (one each side of the fireplace) to which I fitted 4 mm terminals so the cable disappears below the sitting room floor.  The right-hand wall plate (which is behind the curtain just to the right of the CD player) can be seen in the second photograph (the wall-plate with the red and black terminals).

     Another technique I adopt is to sleeve together cables running in the same direction.  In the inset the left hand speaker cable, the power cable to the 306 amplifier, and the screened signal cables between the B&O CD player and 306 are all sleeved together using a large diameter length of pvc sleeving available in a range of sizes from electronic component suppliers.  Of course the cables need to be cut to length and sleeved before the second end connector is fitted to each cable, but I think you can see from the main image that the cabling is barely visible along the edge of the floor*.   Some might say running signal cables next to a power cable inadvisable, but I have not found any problems discernible to my ears.


* In my experience, spiral wound sleeving, which would avoid the need for fitting the second connector before sleeving, is neither successful nor neat.


     And for readers who are interested in interior design, the curtain fabric is Christian Lacroix/Designers' Guild in the Manaos print.  The 22 mm thick strip hardwood floor (solid tongue and groove strips directly onto the joists below and fitted when the house was built some seventy years ago) is merely wax polished, and we find any damage is much easier to repair when waxed than when varnished or lacquered.



... and HDMI connectors

Seemingly ubiquitous for connecting peripherals to digital televisions, I find standard HDMI (High-Definition Multimedia Interface) connectors offer poor security of connection when the cable is subject to any movement, and much of this I ascribe to the weight and stiffness of the standard 19 core cable (4 shielded twisted pairs plus 7 additional cores) compared to the weight of the connector.  On initial set-up, I had all sorts of intermittent (and irritating) connection problems with the standard cables supplied, and this needed to be overcome so that the television might be turned to give the best viewing angle for the seated viewer(s).

     The way in which I solved this problem was firstly to throw away the stiff, as-supplied cables and purchase cables with as small an outside diameter cable as I could get, and secondly to make full use of nylon cable ties to relieve the connectors of as much stress as possible when the television is moved.  The photographs below show our successful arrangement, the second photograph showing the two HDMI connectors with their thin cables secured by cable ties at close intervals to one another plus an additional two points securing the cables to the back of the television.

     As I am not up to date with the various versions of HDMI, I can offer no view on the relative performance of cables of different size; all I can say is that on short runs of less than 1 metre as in this set-up, the smaller diameter HDMI cables appear to work very reliably.

     Returning to the previous two photographs, again I hope you will agree that the sleeved cabling running along the base of the skirting board to the left hand Quad electrostatic loudspeaker (just visible on the extreme left) is also discreet, while I like to think that even the cabling to the back of the television (above) has a certain functional neatness.  The television stand is a B&O Attyca from the 1980s, while just to the right of the HDD recorder and DVD player, those who remember the 1960s will have no trouble whatsoever in recognising the Mathmos lava lamp...

West Country Clock and Watch Fair

The next West Country Clock and Watch Fair will be held on Sunday 22nd April 2018 at the Holiday Inn, Taunton, TA1 2UA, which is close to Junction 25 on the M5 motorway.  Doors open at 9.00am and last entrance is at 2.15pm  For further details visit


     Said Fairs Co-ordinator Donald Hamilton: "With nearly forty stallholders at the November 2017 Fair, as well as providing an opportunity for numerous horological discussions, I am sure visitors to the Fair will find much of interest to make their visit rewarding and worthwhile".

Next time

A Router Collet Chuck for holding woodcutting router cutters in a No. 2 Morse taper spindle.



Guy Gibbons




τέκτων = tektōn, from which the English words technology/technical are derived.

Copyright (c) 2018  G E Gibbons

The Micrometer

A two-monthly review of the engineering of time

Issue 10   March - April 2018

The Micrometer

A two-monthly review of the engineering of time

Issue 10   March - April 2018

Copyright (c) 2018  G E Gibbons