Enter the code you received by e-mail. It stays on this device, so you won't have to type it again.
Certificate
Check a certificate
No certificate carries this number.
✦Course finished — well done.
Your certificate is waiting.
Share the course with a friend.
With your link, you each get an hour of video lesson with me, free — and you get 10 % off your next purchase (the 2824 course, a watch, a movement).
Your review
A few words about the course help me a lot, and help those still hesitating.
What I do at the bench, on screen: turn the movement, count, regulate, test. Everything can be handled, nothing breaks.
Questions & answers for this chapter
The movement, piece by piece
The same movement as on your bench. Turn it, go down layer by layer, come back to it before each chapter.
Drag to turn · wheel or pinch to come closer · tap a part for its name
Theory · the watchmaker's calculations
How a watch counts time.
A mechanical watch does only one thing: it counts beats. The spring gives the force, the wheels pass it on while
gearing it down, the escapement hands it out drop by drop, and the balance keeps the beat. Everything that follows comes down to a few
divisions — the same ones I do at the bench when I check a wheel train.
Wheel — the teeth, the driving partPinion — the leaves, the driven part (in the going train)
1
The path of the energy
The energy starts in the barrel, where the mainspring is wound. It runs through the wheel train — centre wheel, third wheel,
fourth wheel — to the escape wheel. The pallet fork lets it through one tooth at a time, to the beat of the
balance and hairspring. At each stage a large wheel drives a small pinion: the force goes down, the speed goes up.
On the course's 6497
The course's 6497 is on its way…
0 h 00 min 00 s
The real movement of the course, from its 3D model: drag to turn around it, scroll wheel or two fingers to come closer.
The tooth counts were taken from the model itself.
A watchmaker's note: this model's escape wheel has 18 teeth, hence 80 × 60 × 70 × 18 × 2 ÷ (10 × 8 × 7) = 21,600 vph;
the production 6497 has 15, for 18,000 vph. The same calculation, one wheel that changes.
2
Wheels and pinions: who drives whom
In a watch we don't say "teeth" everywhere. The large wheel has teeth; the small pinion, cut
into the arbor, has leaves. The wheel is the driver (it pushes), the pinion is the driven (it is pushed).
When a wheel of 60 teeth drives a pinion of 10 leaves, each tooth of the wheel moves
the pinion on by one leaf. For one turn of the wheel, 60 leaves have gone by: the pinion has made 60 ÷ 10 = 6 turns.
The ratio of one stage = teeth of the driving wheelleaves of the driven pinion
For several stages, the ratios are multiplied together — the formula you meet everywhere:
product of the driversproduct of the driven
On the 6497: the centre wheel drives the third wheel pinion, 80 ÷ 10 = 8 turns
3
Vibrations: the beat of the watch
The balance swings back and forth. One swing is a vibration (the "tick"), the return another vibration (the "tock"):
one oscillation = two vibrations. At each vibration the pallet fork tips over and lets the escape wheel through by
half a tooth: each tooth gives an impulse on the entry pallet, then one on the exit pallet.
That is why the escape wheel's teeth are multiplied by 2.
On the 6497, in slow motion
The escapement has two jobs. It maintains the balance: at each vibration it gives back, with a small push, the energy
that friction took from it. And it counts: it lets the wheel train move on only one step per vibration.
In the Swiss lever escapement, the balance is free almost all the time. The impulse pin only touches the fork
of the pallet fork over a few degrees, around the dead point. For the rest of the arc the pallet fork doesn't move: it is held against its banking
pin, and the balance goes to the end of its swing without touching anything. That is what makes this escapement accurate.
Unlocking. The impulse pin enters the fork and tips the pallet fork: the pallet slides off the tooth it was holding.
The wheel even recoils by a hair.
Impulse. The tip of the pallet rides up the slope of the tooth, then the heel of the tooth slides along the slope of the pallet:
the pallet fork is pushed, and passes that push to the balance through the impulse pin.
Drop. The wheel, now free, turns by a degree and a half, until a tooth lands on the other pallet. That small knock is what you hear.
Run to the banking. The pressure of the tooth on the pallet, angled on purpose (the draw), pulls the pallet fork the rest of the way against its banking pin.
Lock. The pallet fork stays against the pin, the tooth resting on the pallet. The balance travels freely through its supplementary arc, comes back, and it all starts again on the other side.
A few figures for the production 6497: the balance holds the fork over about 44° (the lift angle,
given by ETA); the pallet fork tips 15° from one banking pin to the other; the wheel moves on 12° per vibration, 1.5° of which is drop.
With an amplitude of 270°, the balance therefore spends most of its travel — the supplementary arc — touching nothing.
To find the number of vibrations per hour, start from the centre wheel, which makes exactly one turn per hour
(it carries the minute hand), and work down the train to the escape wheel:
vph = centre × third × fourth × escape × 2third pinion × fourth pinion × escape pinion
The course example: centre wheel 54, third 40 / 6 leaves,
fourth 40 / 6 leaves, escape 15 / 6 leaves:
Without the × 2, you get the number of oscillations per hour: 6,000.
The watchmaker's check. The fourth wheel carries the small seconds: it must make one turn a minute, that is 60 turns an hour.
Centre wheel → fourth wheel: 54 × 406 × 6 = 60 turns an hour. ✓
4
Frequency and period
The frequency is the number of oscillations in one second, in hertz (Hz). The period is the length of one oscillation, in seconds.
One is the inverse of the other.
f (Hz) = vph2 × 3,600 = vph7,200
· T (s) = 1f
For our example: 12,000 ÷ 7,200 = 1.67 Hz, and a period of 1 ÷ 1.67 = 0.6 s.
The higher the frequency, the more finely the watch "cuts up" time: it is less sensitive to knocks and positions,
so more regular. But it uses more energy and wears the escapement more — that is the whole balancing act of a calibre.
vph
Oscillations/h
Frequency
Period
You find it in…
12,000
6,000
1.67 Hz
0.6 s
the course example
18,000
9,000
2.5 Hz
0.4 s
the 6497 / 6498, pocket watches
21,600
10,800
3 Hz
0.33 s
many classic calibres
28,800
14,400
4 Hz
0.25 s
most of today's automatics (2824…)
36,000
18,000
5 Hz
0.2 s
"high-frequency" calibres
On the 6497, in real time: the balance beats three times a second
5
The balance and hairspring: it sets the beat
The wheel train counts, but it doesn't decide the pace. The balance and hairspring does. The balance is a small flywheel;
the hairspring, fixed at one end to its staff (by the collet) and at the other to the balance cock (by the stud), always brings it back towards its rest position,
the dead point. Once set going, the balance overshoots the dead point on its momentum, the hairspring slows it, stops it and sends it back: it oscillates.
A few words to know:
Displacement — the angle of the balance from the dead point, at a given moment.
Amplitude — the largest displacement. On a wound watch, dial up, I look for between 270° and 310°. Below about 200°,
something is braking: a tired mainspring, dried-up oil, a damaged pivot.
Vibration — from one end of the arc to the other. Oscillation — there and back. Period — the length of one oscillation.
T = 2π × √IC
I — the balance's moment of inertia: its mass, and above all how far that mass is from the axis. C — the hairspring's torque: its "stiffness".
What that means at the bench:
The heavier or larger the balance, the slower it is: the watch loses.
The stiffer the hairspring, the faster the balance: the watch gains.
The regulator shortens the part of the hairspring that works: shorter, it is stiffer. Regulator towards "A", the watch gains.
Ideally, the period doesn't depend on the amplitude: large or small oscillation, same length. That is isochronism, and it is what lets
a watch keep time whether it is fully wound or nearly stopped. In practice, the escapement, the positions and the hairspring itself
create small errors. Regulating is the work of making them cancel out.
On the 6497
The regulator — try it
One per cent less inertia — a speck of metal taken from under the rim — and the watch gains more than seven minutes a day.
That is why regulating is done in tiny touches.
6
The power reserve
The barrel itself turns very slowly. Its teeth drive the centre pinion, which makes one turn an hour.
Dividing the barrel's teeth by the centre pinion's leaves tells you how many
hours one turn of the barrel lasts:
60 teeth10 leaves = 6 hours per turn of the barrel
The mainspring, for its part, unwinds over a certain number of turns (its development). If it makes 6:
6 h × 6 turns = 36 hours of power reserve
For comparison, the 6497 you take apart in the course is given at about 46 hours.
The mainspring is never completely "empty": when the watch stops, it still has some tension, but no longer enough to keep the balance going.
Its development is the number of turns between the fully wound spring and the spring that no longer runs the watch.
A mainspring develops over 8 to 10 turns at most.
And its force is not the same from start to finish: strong when it is wound, it drops as it unwinds, and the amplitude drops with it.
That is why I advise winding a hand-wound watch every day, at the same time: it always works in the most
regular part of the spring.
On the 6497: the spring in its drum, winding and running
A word about words: strictly speaking, what we calculate here is the running time — from the fully wound spring until the watch stops.
The power reserve is what is left at a given moment. Running time = time since winding + power reserve.
In practice, people often say "reserve" for both.
The reverse calculation is useful too: for a running time you want, how many turns of the spring do you need?
turns = running time wantedhours per turn of the barrel
· for 48 h with our barrel: 48 ÷ 6 = 8 turns
It all adds up. Over the 36 hours, the barrel makes 6 turns and the centre wheel 36. The escape wheel turns
54 × 40 × 406 × 6 × 6 = 400 times faster than the centre wheel:
36 × 400 = 14,400 turns. Each turn is 15 teeth × 2 = 30 vibrations: 14,400 × 30 = 432,000 vibrations.
And 36 h × 12,000 vph = 432,000. ✓
Winding
When you turn the crown, the stem drives the sliding pinion, which sits on its square. Through its Breguet teeth it drives the winding pinion,
which meshes with the crown wheel on the barrel bridge. The crown wheel drives the ratchet wheel, squared on the barrel arbor:
the arbor turns, the spring winds itself around it. The click jumps from tooth to tooth and stops the ratchet wheel going back — that is the little
clicking you hear when you wind. Turned the other way, the crown winds nothing: the Breguet teeth of the sliding pinion slip.
turns of the crown = turns of the barrel arbor × ratchet wheel teethwinding pinion teeth
On the model's 6497: winding pinion 13, crown wheel 35, ratchet wheel 56. The crown wheel only changes the direction and the plane — the ratio
is 56 ÷ 13 ≈ 4.3 turns of the crown per turn of the arbor. To wind the spring by about 6 turns: 6 × 4.3 ≈ 26 turns of the crown.
And each turn of the arbor is 88 ÷ 12 ≈ 7.3 hours of running.
The direction is no accident: the crown winds when you turn it clockwise, seen from its end.
While the watch runs, it's the other way round: the arbor and the ratchet wheel stay still, held by the click, and the drum turns — in the same direction as the ratchet wheel when winding.
In the 3D view, touch a part for its name; the buttons pull out the crown and turn it one way or the other.
In the barrel view above, the drum is see-through: when winding, the spring coils around the arbor; when running, it unwinds
against the wall. The arrows show the direction of rotation — of the ratchet wheel when winding (orange), of the drum when running (blue): it is the same.
What turns the opposite way is the crown wheel, which drives the ratchet wheel. And when the spring is let down, the watch stops.
On the 6497: winding
7
The motion works: from minutes to hours
Under the dial, the cannon pinion turns with the centre wheel: one turn an hour, it carries the minute hand.
It drives the minute wheel, whose pinion drives the hour wheel, which carries the hour hand.
That one must turn 12 times more slowly:
For example: cannon pinion of 10, minute wheel of 30, minute wheel pinion of 8,
hour wheel of 32: 30 × 3210 × 8 = 3 × 4 = 12 ✓
Here the calculation reads "backwards": we want a reduction, so the large wheels are the driven ones.
On the watch, the hour wheel sits on the cannon pinion, around the same axis; the minute wheel is beside it, under its pinion — the 3D view below shows it from the dial side.
Notice: 10 + 30 = 8 + 32 = 40. That is no accident. The cannon pinion and the hour wheel turn on the same centre, and the minute
wheel and its pinion on a single arbor: the two gearings therefore have the same centre distance. With teeth of the same size
(the same module), the sum of the teeth must be the same on both sides.
For a 24-hour dial, same calculation, with 24 instead of 12.
And when setting the time? The crown turns the sliding pinion, which drives the minute wheel through the setting wheel.
The cannon pinion then turns by friction on the centre arbor: the hands move without dragging the wheel train.
On the 6497, dial side
Setting the time
You pull the crown: the stem moves out one notch. The setting lever, whose pin sits in the groove of the stem, pivots on its screw;
it tips the yoke, which pushes the sliding pinion inwards. It leaves the winding pinion and meshes with the setting wheel.
The setting lever spring holds the stem in that position.
You turn: the sliding pinion drives the setting wheels, then the minute wheel, which turns the cannon pinion and the hour wheel.
The cannon pinion slips by friction on the centre wheel's arbor: the hands turn, the wheel train doesn't move, the watch keeps running.
You push the crown back in: the yoke spring brings the sliding pinion back onto the winding pinion.
turns of the minute hand per turn of the crown = 10 × 16 × 3616 × 36 × 12
= 1012: one turn of the crown moves the hands on by 50 minutes
Sliding pinion 10, setting wheels 16, minute wheel 36, cannon pinion 12: the setting wheels and the minute wheel cancel out in the calculation, they only pass the motion on.
The same direction of rotation — clockwise, seen from the end of the crown — winds when it is pushed in and moves the hands forward when it is pulled out.
On the 6497, dial side: pull, turn, push back
8
The calculator
Count the teeth and leaves of a wheel train — your own, on the bench — and let the calculation do the rest. The list of speeds in part 1 uses these figures.
9
Exercises
Do them on paper first. The answer opens under each question.
1. A barrel of 72 teeth drives a centre pinion of 12 leaves. The mainspring develops over 7 turns. What is the power reserve?
See the answer
72 ÷ 12 = 6 hours per turn of the barrel; 6 × 7 = 42 hours.
2. Centre wheel 80; third wheel 75 teeth, pinion 10 leaves; fourth wheel 80 teeth,
pinion 10 leaves; escape wheel 15 teeth, pinion 8 leaves. How many vibrations per hour? What frequency, what period?
See the answer
(80 × 75 × 80 × 15 × 2) ÷ (10 × 10 × 8) = 14,400,000 ÷ 800 = 18,000 vph;
18,000 ÷ 7,200 = 2.5 Hz; period 1 ÷ 2.5 = 0.4 s. And the fourth wheel does make (80 × 75) ÷ (10 × 10) = 60 turns an hour.
3. A watch beats at 28,800 vph. How many oscillations per hour? What frequency? How many "ticks" do you hear in one second?
See the answer
28,800 ÷ 2 = 14,400 oscillations per hour; 28,800 ÷ 7,200 = 4 Hz;
and 28,800 ÷ 3,600 = 8 vibrations per second — you hear 8 beats.
4. Motion works: cannon pinion 15, minute wheel 45, minute wheel pinion 12, hour wheel 48.
Will the hour hand be right? And is the centre distance the same for both gearings?
See the answer
(45 × 48) ÷ (15 × 12) = 2,160 ÷ 180 = 12: yes, the hour wheel makes one turn while the cannon pinion makes twelve.
And 15 + 45 = 12 + 48 = 60: with the same module, both gearings do have the same centre distance.
5. With the course example (12,000 vph), by how many degrees does the escape wheel move on at each vibration?
See the answer
15 teeth, two vibrations per tooth: one turn = 30 vibrations, so 360 ÷ 30 = 12° per vibration.
6. I want a watch that runs for 40 hours. The barrel has 80 teeth, the centre pinion 10 leaves.
Over how many turns must the mainspring develop?
See the answer
80 ÷ 10 = 8 hours per turn of the barrel; 40 ÷ 8 = 5 turns.
7. The fourth wheel is missing. The watch beats at 21,600 vph (3 Hz) and has a small seconds. Centre wheel 80;
third 75 / 10 leaves; fourth pinion 10 leaves;
escape 15 / 7 leaves. How many teeth on the fourth wheel?
See the answer
First the check: (80 × 75) ÷ (10 × 10) = 60 turns an hour, the small seconds is right.
Then, 21,600 = 60 × S × 15 × 2 ÷ 7, so S = 21,600 × 7 ÷ (60 × 30) = 84 teeth.
8. A watch loses 2 minutes a day. By how much must the balance's moment of inertia change to correct it?
See the answer
2 minutes = 120 s out of 86,400 s: it must run 0.14 % faster. The period varies as the square root of I,
so a small change in I counts for half: the inertia must come down by about 2 × 0.14 = 0.28 %.
In practice you don't touch the balance for so little: you turn the regulator, or the balance screws if it has them.
The words in this chapter — barrel, cannon pinion, pallet fork, pallets, leaves… — are all in the Watchmaking glossary, under Documents.
Practice · regulating
Regulating a watch on the timegrapher.
When a watch comes out of a service, I put it on the timegrapher before closing it. In twenty seconds it tells me whether it gains
or loses, whether the balance has strength, and whether the tick and the tock are equal. It is the tool that tells me whether the work is good. Here you
have one on screen: it reacts like mine, and you can put it out of adjustment as much as you like.
1
What it measures
The machine has a microphone in its clamp. It hears every beat — the noise of the lever and the escape wheel — and compares it with its
quartz, which doesn't make mistakes. Out of it come three figures:
The rate, in seconds per day (s/d). What the watch gains (+) or loses (−) in 24 hours.
The amplitude, in degrees. How far the balance turns either side of its rest point. It is the health of the movement: a
balance with good amplitude is getting its energy properly, from the barrel to the escapement.
The beat error, in milliseconds. The difference between the tick and the tock. If the balance at rest is not exactly
in the middle of the lever's fork, one beat is longer than the other.
For the amplitude, the machine needs to know the movement's lift angle, the angle the balance turns through while the lever
gives it its impulse. For most calibres it is 52°; some are 50° or 53°. With the wrong angle the amplitude shown is
wrong, not the rate.
The values I consider good, watch fully wound
Rate
Amplitude flat
Amplitude vertical
Beat error
Men's watch
−5 to +15 s/d
250 to 330°
220 to 270°
0 to 0.5 ms
Ladies' watch
−5 to +25 s/d
250 to 330°
220 to 270°
0 to 0.5 ms
Chronometer
−2 to +6 s/d
250 to 330°
220 to 270°
0 to 0.5 ms
The next day, let down, the amplitude drops 10 to 15 %: that's normal. If it collapses, the barrel or the oils have a problem.
2
Before measuring
I wind the watch: ten to fifteen turns of the crown. Then I let it run five to ten minutes. A watch that has just been wound is not
stable yet, and the trace shows it.
I look at the time: never just before midnight. The date change takes force from the train and spoils the measurement.
The crown is pushed in, or screwed down.
I demagnetise. A magnetised watch gives a very messy trace, and you end up looking for a fault that doesn't exist. A phone, a
speaker, a bag with a magnetic clasp are enough to magnetise a hairspring.
3
The positions
A watch doesn't run the same lying flat and standing up: flat, the balance rests on the tip of its pivot; vertical, it rubs on its side. So we measure
in several positions. I start vertical — crown down, crown left, crown up — then flat, dial up and dial
down. Between two positions I wait ten to fifteen seconds for the trace to settle, then I measure for fifteen to twenty seconds.
What interests me, more than the rate in one position, is the difference between the positions. A watch doing +3 flat and +40 vertical
is not a watch to regulate: it has a balance to poise, or a damaged pivot.
4
The timegrapher
Your turn. The watch is in the clamp, running. The machine is made like mine: at the top the rate, the beat error and the amplitude, in the middle
the trace, at the bottom its settings. With cursor you choose a setting, with ▲ ▼ you change it. speaker lets you hear the watch,
print notes the measurement for the position, freeze stops the trace.
RATE s/dayBEAT ERROR (count down s)AMPLITUDE deg
----------
AUT2880020522GRAPH
BEAT MODEBEATS PER HOURMEAS. TIMELIFT ANGLEGRAPH RESOL.TEST MODE
timegrapher · l'atelier TL
siglevel
freezeprintcursorspeaker
The watch in the clamp
Turn the stud carrier or the regulator right on the balance, with a finger or the mouse. The stud carrier takes the regulator with it: the pins stay where the hairspring passes.
The readings
Position
Rate
Amplitude
Beat error
5
Regulating
Always in this order: the beat error first, the rate second. The stud shifts the rate a little when you turn it; the regulator doesn't
touch the beat error. If I set the rate first, I lose it while setting the beat error.
The beat error. Under 0.5 ms, I leave it alone. Above, I turn the stud carrier a hair (or the stud itself on calibres without
a stud carrier), wait for the trace to settle, and look. The two lines should come together.
The rate. With the regulator, dial up, I aim for +2 to +5 s/d: a watch that gains a little on the wrist is nicer than a
watch that loses. Then I check it vertical.
The difference between positions. If it is small and steady, it is isochronism: a few seconds' difference between flat and
vertical. On a hairspring with regulator pins, you correct it by closing the pins if it loses vertical, opening them if it
gains. If it is big, it is no longer regulating: the balance needs poising, or the regulating organ needs work.
6
Reading the trace
The trace says more than the figures. Choose each fault in the timegrapher above and compare it with what follows.
Two straight lines, close together
A movement in good order. The slope gives the rate.
Two straight lines, but apart
The beat error. Nothing to do but set it, then redo the rate.
Lines that wave regularly
A bent escape wheel, or a wheel of the train running out of true: one wave per turn. An
escape wheel with 15 teeth gives a wave every 15 teeth. The wheel is replaced.
Scattered dots, a low amplitude
The oils have dried, the movement is dirty. Service. If the amplitude takes a long time to
come back when you change position, the pivots of the balance or the train are short of oil.
A cloud of dots, a wild rate
A magnetised watch. Demagnetise and measure again before taking anything apart.
A line that splits in two
Knocking: the amplitude is so high (over 330°) that the impulse pin comes back and hits the lever's fork
from the outside. You can hear it too, a double tick-tock. Mainspring too strong, or the lever and escape wheel need work.
Jumping dots, a scraping
The hairspring touches: the regulator pins, the stud, or the coils against each other. Centre it and
put the hairspring back flat.
A big difference between the vertical positions
Out of poise: the balance is heavier on one side. Poise it or replace
the regulating organ.
7
Afterwards
The timegrapher tells how the watch runs for twenty seconds. Before handing it back, I check the date change, the power
reserve, the automatic winding if there is one, and I let it run for twenty-four hours. The next day I put it back on the machine,
let down, to see the amplitude drop normally.
Practice · water resistance
A water-resistant watch, and how to check it.
A watch that lets water in is a movement that rusts within days. After every service I change the seals and test
the case before handing the watch back. Here is what "water-resistant" means, how it is measured, and why a watch stops being so.
1
What "water-resistant" means
The word is defined by a standard, ISO 2281. A watch marked "water-resistant" must withstand sweat, rain, splashes,
an hour under ten centimetres of water, and a minute under a pressure of 2 bar — the equivalent of twenty metres of water.
The "30 m", "50 m" or "100 m" engraved on the caseback is therefore not a diving depth: it is a test pressure in a laboratory, with the watch
still. On the wrist, an arm moving through water or a shower jet creates much stronger pressure peaks. In practice: 30 m, you can
wash your hands; 100 m, you can swim; to dive you need a diver's watch, which follows its own standard.
The limit set by the standard is very small: under 2 bar, in one minute, no more than 50 micrograms of air may get into the case.
2
How it is measured
In the workshop we don't put the watch in water: if it isn't water-resistant, the water is already in the movement. We test it with air.
By deformation. The watch sits in a closed chamber, under a sensor. The chamber is put under pressure (or vacuum): the
case deforms very slightly, the crystal and the caseback sink in (or bulge out) by a few microns. If the case is water-resistant, it stays
deformed. If it leaks, air gets in (or out), the pressures even out, and the deformation goes back by itself. The sensor measures that return:
under about 1 % for a man's watch, it is water-resistant.
Under pressure and under vacuum. Under pressure (2 bar and more), the chamber presses the parts against each other: you mostly test
the case itself, its seals, its crystal. Under vacuum (−0.4 to −0.8 bar), the parts hold only by their own fit: you find
the small leaks and the fittings that are too weak, a crystal pressed in too loose, a caseback not tight enough. I do both.
By condensation, when I have no other way to settle a doubt: I warm the watch to about 40-45 °C for half an hour, then
put a drop of lukewarm water on the crystal. If mist appears inside, moisture has got in: the case is not water-resistant.
3
The test bench
Choose the test and the case, then start. On the left, the watch in the chamber, cut in half: watch the crystal and the caseback sink in
(or bulge out), and the air get through where the case leaks. The deformation is magnified thousands of times: in reality, it is a few microns.
On the right, the curve I look at on my machine: the pressure, and the deformation of the case.
Put the watch in, choose the test, and start.
Negative return deformation: the case comes back, air gets through. Beyond −1 % (−2.5 % for a lady's watch), it is not water-resistant.
Zero: a very rigid case, it hardly moves. Water-resistant.
Positive: the material keeps giving a little during the measurement, a thin caseback for example. Water-resistant.
4
Why it isn't any more
There are three main reasons, and the test that fails often tells you which:
Worn parts
Old, dry or dirty seals — crown, caseback, crystal —, a cracked acrylic crystal, a damaged crown.
It shows mostly under pressure. The seals are changed at every service: a ten-year-old seal has no elasticity left.
Parts that deform under pressure
The case holds in the light test and leaks under high pressure: a caseback too thin, a crystal
that sinks in. It shows only under pressure.
A fit that is too weak
A crystal pressed in too loose, a seal too thin, an oval crown tube, a caseback not screwed down properly. The parts hold
only by their fit, and it shows under vacuum.
5
Before and during the test
I look at the case and the crystal. A cracked crystal I don't test: I replace it.
The watch must be clean, the crown pushed in, or screwed down all the way — it's the most common thing forgotten.
I remove the protective films stuck on the caseback or the crystal: the air trapped under them spoils the measurement.
I don't hold the watch in my hand for long: it must be at the same temperature as the machine, otherwise the air inside it expands
during the test.
I test at the pressure given by the brand, and with the right sensor: very domed crystals go dial down.
If the result is bad, I don't restart straight away. I pull the crown out, wait fifteen seconds, push it back in, then do
the test again. And I keep the measurement sheet with the watch's record.
Practice · quartz
A quartz watch, and why it stops.
A quartz watch that has stopped is not always the battery. Before changing anything, I measure: the battery, the coil, the circuit,
the motor, and only then the mechanics. Here is how it works, what I measure, in what order, and a tester on screen to find
the fault in a watch you don't know.
1
What's inside
The battery. It gives the energy, 1.55 V for a silver-oxide battery, 3 V for a lithium battery.
The quartz. A small crystal tuning fork that vibrates 32,768 times a second. It gives the precision, like the balance in
a mechanical watch.
The circuit (IC). It divides those 32,768 vibrations by two, by two again, fifteen times, down to one a second. Every second
it sends a pulse to the motor.
The stepping motor. A coil and a small magnet, the rotor. Each pulse in the coil makes the rotor turn half a turn.
The train and the motion works. As in a mechanical watch: they carry the rotor's movement to the hands.
Everything is electrical up to the rotor, everything is mechanical after it. So a fault is on one side or the other, and the tester is there to find out
which.
2
Measuring the rate
The tester can listen to two things:
The quartz, with the acoustic microphone (or capacitive on an open watch). It measures the 32,768 Hz directly. It is the quick test:
if the "quartz" signal lights up and the rate is between 0 and about +6 s/d, the quartz and the circuit are working.
The motor, with the magnetic sensor. It feels the pulse going through the coil, one a second (or every 5, 10, 20 seconds
depending on the calibre). For a true rate measurement, I measure over 60 seconds.
Why both? Because on many watches the quartz is no longer adjusted. There used to be a trimmer, a small capacitor you
turned. Today the quartz is left a little fast, +3 to +5 s/d, and the circuit is programmed to "forget" a few vibrations once
a minute (or every 20 s, 30 s, 2, 4 or 8 minutes depending on the circuit). That is inhibition. On those watches, the quartz alone
says it gains 4 seconds a day, when it is spot on. Only the measurement at the motor, over 60 seconds, gives the true rate.
3
The values
What I consider good
Measurement
Good
What it means otherwise
Rate at the quartz
0 to +6 s/d
No signal: the quartz or the circuit
Rate at the motor (60 s)
−0.05 to +0.35 s/d (−2 to +15 s a month)
The watch's true rate
1.55 V battery, under load
1.45 V minimum
Battery to replace
3 V lithium battery, under load
2.85 V minimum
Battery to replace
Coil
1 to 3 kΩ
Infinite: broken. Too low: short-circuited
Insulation (coil, or battery −, to the mainplate)
700 kΩ and more
Current leaking: corrosion, moisture
Starting voltage
1.3 V max (3 hands) 1.15 V max (2 hands)
The train is forcing
Minimum pulse with the generator
3.9 to 4.8 ms very good 4.8 to 5.8 ms good 5.8 to 6.8 ms just enough
Beyond, the train is forcing
The battery test is done under load: unloaded, an almost flat battery still shows a good voltage. The tester draws current from it
like the motor (about 750 µA), or like a light (15 mA) for "high drain" batteries.
4
The tester
A stopped watch on the bench. Choose a fault, or take a mystery watch, and make the measurements. Each measurement is done in the
conditions it needs: with its battery, without a battery, or powered by the tester.
RATE
Signal
quartz
LCD
motor
----
sec/day
sec/month
MODULE
Continuity
open
short
----
µA
mA
V
kOhm
PARAMETER
1.55
V
sec
msec
The watch open, on the tester's table.
Battery test
Testmode rate
Testmode module
Parameter
Choose a measurement.
5
A stopped watch: in order
First of all, I open it and look: corrosion around the battery, a badly insulated battery housing, that is already an answer.
The battery. I take it out and test it under load. Bad: I replace it, and I don't stop there. A new battery that runs flat in
a few months means something else is drawing current.
The coil and the insulation, with no battery and no power. The coil between 1 and 3 kΩ, the insulation of the coil and of the battery − to
the mainplate above 700 kΩ. A broken coil, a low insulation: I replace the coil, or I look for moisture.
The quartz and the circuit. The watch on the table, powered by the tester instead of the battery, crown pulled out (time-setting
position: the motor stops). I look at the consumption. No signal, or a wild consumption: I replace the electronic module.
The motor. Crown pushed in, rate measurement over 60 seconds, and the consumption at the same time. The pulses arrive but the
hands don't turn: the motor or the train.
The starting voltage. I lower the voltage little by little until the train stops. Three hands that stop
above 1.3 V: the train is forcing somewhere.
The pulse generator. The tester takes the circuit's place and sends pulses into the coil itself. I start from a
short pulse, 2.9 ms, and lengthen it until the watch runs. If it needs more than 6.8 ms, it is forcing. If it runs
well with the generator and not with its own circuit, it is the circuit.
When everything is good: a new battery, I close it, and I test the water resistance.
6
The mechanics
When the electronics are good and the watch is forcing, it is almost always one of these:
A steel particle
The rotor is a magnet: it attracts the smallest filing. A particle on the rotor or in the train, and it
jams. Take it apart, clean it, and never put a rotor down on the bench without looking at what it has picked up.
The crown
Dirt between the crown and the case holds the stem in the time-setting position: the circuit thinks the time is being set
and stops the motor.
The hands
A hand touching the crystal, or the seconds hand touching the minute hand. Or hands pressed on so
low that they have no play left.
The calendar
The date change asks an effort of the train. A watch that always stops around midnight: that's where.
Pratique · votre montrePractice · your own watchOefenen · uw eigen horlogeÜben · Ihre eigene Uhr
Le chronocomparateur, sur votre montre.The timegrapher, on your own watch.De uurwerktester, op uw eigen horloge.Die Zeitwaage, an Ihrer eigenen Uhr.
Posez la montre sur le micro, branché à l'ordinateur ou au téléphone : la page mesure la marche, le repère et l'amplitude, comme l'appareil de mon atelier. Choisissez le calibre, l'angle de levée vient de ma bibliothèque de mouvements. Le son ne quitte pas votre appareil.Put the watch on the microphone, plugged into your computer or phone: the page measures the rate, the beat error and the amplitude, like the instrument in my workshop. Choose the calibre, and the lift angle comes from my movement library. The sound never leaves your device.Leg het horloge op de microfoon, aangesloten op de computer of de telefoon: de pagina meet de gang, de slagfout en de amplitude, zoals het apparaat in mijn werkplaats. Kies het kaliber; de heffingshoek komt uit mijn uurwerkbibliotheek. Het geluid verlaat uw apparaat niet.Legen Sie die Uhr auf das Mikrofon, angeschlossen an Computer oder Telefon: Die Seite misst Gang, Abfallfehler und Amplitude, wie das Gerät in meiner Werkstatt. Wählen Sie das Kaliber, der Hebungswinkel kommt aus meiner Werkbibliothek. Der Ton verlässt Ihr Gerät nicht.
MODEBEAT MODEMODUSMODUSALTERNANCES/HBEATS PER HOURSLAGEN/UURA/HMESUREMEAS. TIMEMETINGMESSUNGANGLE DE LEVÉELIFT ANGLEHEFFINGSHOEKHEBUNGSWINKELRÉSOLUTIONGRAPH RESOL.RESOLUTIEAUFLÖSUNGPOSITIONPOSITIONPOSITIELAGE
Les six positionsThe six positionsDe zes positiesDie sechs Lagen
Choisissez la position où la montre est posée, attendez que les chiffres se calment (une trentaine de secondes), puis noter pour l'inscrire. Faites-en au moins deux : cadran en haut et couronne en bas.Pick the position the watch is held in, wait until the numbers settle (about thirty seconds), then print to note it. Do at least two: dial up and crown down.Kies de positie waarin het horloge ligt, wacht tot de cijfers rustig worden (een halve minuut), en druk op noteren. Doe er minstens twee: wijzerplaat boven en kroon onder.Wählen Sie die Lage, in der die Uhr liegt, warten Sie, bis sich die Zahlen beruhigen (etwa dreißig Sekunden), dann notieren. Mindestens zwei: Zifferblatt oben und Krone unten.
Avec un bon micro, la marche et le repère sont justes ; l'amplitude reste une estimation. Une carte son bon marché peut décaler la marche de quelques secondes par jour : le décalage est le même dans toutes les positions, l'écart entre elles reste juste.With a good microphone the rate and the beat error are right; the amplitude stays an estimate. A cheap sound card can shift the rate by a few seconds a day: the shift is the same in every position, so the spread between them stays right.Met een goede microfoon kloppen de gang en de slagfout; de amplitude blijft een schatting. Een goedkope geluidskaart kan de gang een paar seconden per dag verschuiven: die verschuiving is in elke positie gelijk, dus het verschil tussen de posities klopt.Mit einem guten Mikrofon stimmen Gang und Abfallfehler; die Amplitude bleibt eine Schätzung. Eine billige Soundkarte kann den Gang um einige Sekunden pro Tag verschieben: Die Verschiebung ist in allen Lagen gleich, die Differenz zwischen ihnen stimmt also.
Send me a photo and your question, even when it is not about the course: a movement of your own, a part you cannot place, an adjustment that will not come right. You can circle what you mean.
Sent — the answer will appear here.
Sending failed, try again in a moment.
A video lesson, with me
Request sent — the confirmation will appear here.
That slot has just been taken. Please pick another.
