<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>The Math on Open Timegrapher</title><link>https://opentimegrapher.org/math/</link><description>Recent content in The Math on Open Timegrapher</description><generator>Hugo</generator><language>en</language><atom:link href="https://opentimegrapher.org/math/index.xml" rel="self" type="application/rss+xml"/><item><title>M1. Beats</title><link>https://opentimegrapher.org/math/m1-beats/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://opentimegrapher.org/math/m1-beats/</guid><description>&lt;p>Every reading starts from a list of beats: a time, a number and a
quality for each tick and tock heard. This page says how the app finds
them in the sound.&lt;/p>
&lt;h2 id="what-it-is">What It Is&lt;/h2>
&lt;p>A &lt;strong>beat&lt;/strong> is one tick or tock found in the sound. Each has:&lt;/p>
&lt;ul>
&lt;li>&lt;strong>a time:&lt;/strong> when its drop was heard, on the sound card&amp;rsquo;s clock;&lt;/li>
&lt;li>&lt;strong>a number:&lt;/strong> counted from the take&amp;rsquo;s start, with any missed beats
counted too, so beat 1,000 is always 1,000 beats after beat 0;&lt;/li>
&lt;li>&lt;strong>a quality:&lt;/strong> how well its sound matches the take&amp;rsquo;s typical beat.
A typical beat scores about 1. Only beats scoring over 0.4, the &lt;strong>good
beats&lt;/strong>, are timed or measured.&lt;/li>
&lt;/ul>
&lt;p>Even-numbered beats are &lt;strong>Tick&lt;/strong>, odd-numbered beats are &lt;strong>Tock&lt;/strong>.&lt;/p></description></item><item><title>M2. Rate</title><link>https://opentimegrapher.org/math/m2-rate/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://opentimegrapher.org/math/m2-rate/</guid><description>&lt;h2 id="what-it-is">What It Is&lt;/h2>
&lt;p>The &lt;strong>rate&lt;/strong> is how many seconds a day the watch would gain (+) or lose
(−) if it kept running exactly as it does over the reading.&lt;/p>
&lt;h2 id="how-it-is-worked-out">How It Is Worked Out&lt;/h2>
&lt;p>The app fits one straight line through the good beats&amp;rsquo; times against
their numbers:&lt;/p>
&lt;div class="highlight">&lt;pre tabindex="0" class="chroma">&lt;code class="language-text" data-lang="text">&lt;span class="line">&lt;span class="cl">time of beat k = t0 + P × k ± e/2
&lt;/span>&lt;/span>&lt;/code>&lt;/pre>&lt;/div>&lt;ul>
&lt;li>&lt;code>P&lt;/code> is the watch&amp;rsquo;s beat period, the time from one beat to the next.&lt;/li>
&lt;li>&lt;code>e/2&lt;/code> is added to Ticks and taken from Tocks: half the beat error, so
the two sides&amp;rsquo; uneven timing doesn&amp;rsquo;t tilt the line (M5).&lt;/li>
&lt;li>The line is fitted by least squares, then fitted twice more without the
beats over 5 robust standard deviations from it.&lt;/li>
&lt;/ul>
&lt;p>The rate compares &lt;code>P&lt;/code> with the beat period the movement is made for,
3,600 s over the beat rate:&lt;/p></description></item><item><title>M3. The Sound Card's Clock</title><link>https://opentimegrapher.org/math/m3-the-sound-cards-clock/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://opentimegrapher.org/math/m3-the-sound-cards-clock/</guid><description>&lt;h2 id="what-it-is">What It Is&lt;/h2>
&lt;p>The sound card counts samples by its own quartz crystal. If the crystal
runs fast, the card&amp;rsquo;s second is short, and every beat seems to come a
little late: the watch reads slow. The error is given in &lt;strong>parts per
million (ppm)&lt;/strong>, + when the card runs fast.&lt;/p>
&lt;table>
 &lt;thead>
 &lt;tr>
 &lt;th>Card error&lt;/th>
 &lt;th>Rate error&lt;/th>
 &lt;/tr>
 &lt;/thead>
 &lt;tbody>
 &lt;tr>
 &lt;td>1 ppm&lt;/td>
 &lt;td>0.09 s/d&lt;/td>
 &lt;/tr>
 &lt;tr>
 &lt;td>10 ppm&lt;/td>
 &lt;td>0.86 s/d&lt;/td>
 &lt;/tr>
 &lt;tr>
 &lt;td>20 ppm&lt;/td>
 &lt;td>1.73 s/d&lt;/td>
 &lt;/tr>
 &lt;tr>
 &lt;td>50 ppm&lt;/td>
 &lt;td>4.32 s/d&lt;/td>
 &lt;/tr>
 &lt;/tbody>
&lt;/table>
&lt;p>A sound card&amp;rsquo;s crystal is commonly 10 to 50 ppm off, and holds that error
steadily. Two sound cards can read the same watch a few s/d apart.&lt;/p></description></item><item><title>M4. Amplitude and Lift Angle</title><link>https://opentimegrapher.org/math/m4-amplitude-and-lift-angle/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://opentimegrapher.org/math/m4-amplitude-and-lift-angle/</guid><description>&lt;h2 id="what-it-is">What It Is&lt;/h2>
&lt;p>The &lt;strong>amplitude&lt;/strong> is how far the balance turns from rest to the end of its
swing, in degrees.&lt;/p>
&lt;p>The &lt;strong>lift angle&lt;/strong> is the arc of the balance&amp;rsquo;s swing during which the
impulse pin is in the fork: from the unlock to the drop. It is set by the
movement&amp;rsquo;s design. The calibre table gives it where it is known.&lt;/p>
&lt;h2 id="how-it-is-worked-out">How It Is Worked Out&lt;/h2>
&lt;p>The balance swings like a pendulum: its angle follows a sine wave of
amplitude &lt;code>A&lt;/code> and period &lt;code>T&lt;/code>, one full oscillation, two beats. Near the
middle of its swing it passes through the lift angle &lt;code>L&lt;/code>, centred on the
rest point. The time &lt;code>t&lt;/code> it takes is short when the swing is wide, since
the balance is moving fastest there.&lt;/p></description></item><item><title>M5. Beat Error</title><link>https://opentimegrapher.org/math/m5-beat-error/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://opentimegrapher.org/math/m5-beat-error/</guid><description>&lt;h2 id="what-it-is">What It Is&lt;/h2>
&lt;p>The &lt;strong>beat error&lt;/strong> is how unevenly the two halves of the balance&amp;rsquo;s swing
are timed: half the difference between the Tick-to-Tock and Tock-to-Tick
intervals, in milliseconds.&lt;/p>
&lt;p>If one interval is 124.7 ms and the next 125.3 ms, the beat error is
half of 0.6 ms: &lt;strong>0.30 ms&lt;/strong>.&lt;/p>
&lt;p>It is signed: &lt;strong>+ when the Tick comes late&lt;/strong>. The cards and charts show
its size; the sign is in the command line&amp;rsquo;s JSON.&lt;/p></description></item><item><title>M6. Averaging, Windows and Cuts</title><link>https://opentimegrapher.org/math/m6-averaging-windows-and-cuts/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://opentimegrapher.org/math/m6-averaging-windows-and-cuts/</guid><description>&lt;h2 id="what-it-is">What It Is&lt;/h2>
&lt;p>A reading is never one beat. It is built in layers, each made only from
the one above it:&lt;/p>
&lt;div class="highlight">&lt;pre tabindex="0" class="chroma">&lt;code class="language-text" data-lang="text">&lt;span class="line">&lt;span class="cl">Sound
&lt;/span>&lt;/span>&lt;span class="line">&lt;span class="cl"> ↓ beats found (M1)
&lt;/span>&lt;/span>&lt;span class="line">&lt;span class="cl">Beats: time, number, quality
&lt;/span>&lt;/span>&lt;span class="line">&lt;span class="cl"> ↓ placed on true time (M3); cuts and masks applied
&lt;/span>&lt;/span>&lt;span class="line">&lt;span class="cl">Rate fit: one line through every good beat in the span (M2)
&lt;/span>&lt;/span>&lt;span class="line">&lt;span class="cl">10 s rate windows: a rate fit every 5 s, overlapping by half
&lt;/span>&lt;/span>&lt;span class="line">&lt;span class="cl">2 s windows: each side&amp;#39;s beats averaged into one shape (M4, M5)
&lt;/span>&lt;/span>&lt;span class="line">&lt;span class="cl"> ↓
&lt;/span>&lt;/span>&lt;span class="line">&lt;span class="cl">Reading: rate as fitted; amplitude and beat error as the median of
&lt;/span>&lt;/span>&lt;span class="line">&lt;span class="cl">the 2 s windows; each with its ± (M7) and chips
&lt;/span>&lt;/span>&lt;/code>&lt;/pre>&lt;/div>&lt;h2 id="how-it-is-worked-out">How It Is Worked Out&lt;/h2>
&lt;p>&lt;strong>The span.&lt;/strong> While listening, a reading covers the last &lt;strong>Average Over&lt;/strong>
seconds (10 s unless changed). For a stopped or opened take, the whole
take&amp;rsquo;s reading covers each position&amp;rsquo;s measured stretch.&lt;/p></description></item><item><title>M7. The ± and Correlated Readings</title><link>https://opentimegrapher.org/math/m7-the-plus-minus/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://opentimegrapher.org/math/m7-the-plus-minus/</guid><description>&lt;h2 id="what-it-is">What It Is&lt;/h2>
&lt;p>The &lt;strong>±&lt;/strong> beside a figure is its &lt;strong>standard error&lt;/strong>: an estimate, from the
take itself, of how far the figure would likely move if the same stretch
were heard again under the same conditions.&lt;/p>
&lt;p>It is &lt;strong>precision, not accuracy&lt;/strong>. A wrong lift angle or an uncorrected
sound card clock moves the figure without widening the ±.&lt;/p>
&lt;h2 id="how-it-is-worked-out">How It Is Worked Out&lt;/h2>
&lt;p>The ± comes from the short readings inside the span, in time order: the
2 s windows for amplitude and beat error, and every other 10 s rate
window for the rate.&lt;/p></description></item><item><title>M8. Positions Test Figures</title><link>https://opentimegrapher.org/math/m8-positions-test-figures/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://opentimegrapher.org/math/m8-positions-test-figures/</guid><description>&lt;h2 id="what-it-is">What It Is&lt;/h2>
&lt;p>A watch runs at a slightly different rate in each position, because
gravity acts on the balance differently in each. A &lt;strong>positions test&lt;/strong>
measures it in several positions and sums the differences in a few
figures. The Positions Test pane (chapter 20) gives them, and so does
&lt;code>timegrapher session&lt;/code>. They follow the figures Witschi&amp;rsquo;s timegraphers
use and the Swiss standard NIHS 93-10 (NIHS: Normes de l&amp;rsquo;Industrie Horlogère
Suisse, the Swiss watch industry&amp;rsquo;s standards; the project&amp;rsquo;s
&lt;a href="https://github.com/peterhudson/timegrapher/blob/v1.0.1/docs/developer/math-standards.md">math standards&lt;/a> list them).&lt;/p></description></item><item><title>M9. Steadiness Statistics</title><link>https://opentimegrapher.org/math/m9-steadiness-statistics/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://opentimegrapher.org/math/m9-steadiness-statistics/</guid><description>&lt;h2 id="what-it-is">What It Is&lt;/h2>
&lt;p>Steadiness (chapters 9 to 18) runs a family of tests on the readings of
each series in time order, and names what passes. This note gives the
statistics behind the Info panel&amp;rsquo;s figures and the views: the scatter,
the outlier rule, the Allan deviation, the trend, the CUSUM (cumulative sum), the change
finder and Holm&amp;rsquo;s correction. The period search has its own note (M10),
and the autocorrelation is in chapter 13.&lt;/p></description></item><item><title>M10. Finding Cycles</title><link>https://opentimegrapher.org/math/m10-finding-cycles/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://opentimegrapher.org/math/m10-finding-cycles/</guid><description>&lt;h2 id="what-it-is">What It Is&lt;/h2>
&lt;p>The &lt;strong>period search&lt;/strong> looks for periods the readings repeat at, scores how
sure it is of each, and names the wheel each one matches. Its scores are
what Periods draws (chapter 11), its lines are what the headline names,
and its cycle is what Cycle folds (chapter 12).&lt;/p>
&lt;h2 id="how-it-is-worked-out">How It Is Worked Out&lt;/h2>
&lt;ol>
&lt;li>
&lt;p>&lt;strong>The watch&amp;rsquo;s own clock.&lt;/strong> Each beat is placed at its number times the
nominal beat period, not at the time the sound card heard it. A wheel
turns once in a fixed number of beats: the fourth wheel once a minute
is 480 beats at 28,800 bph, and the escape wheel&amp;rsquo;s turn is twice its
teeth in beats:&lt;/p></description></item><item><title>M11. Time and Frequency Maps</title><link>https://opentimegrapher.org/math/m11-time-and-frequency-maps/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://opentimegrapher.org/math/m11-time-and-frequency-maps/</guid><description>&lt;h2 id="what-it-is">What It Is&lt;/h2>
&lt;p>Three views break a series into periods: the &lt;strong>FFT&lt;/strong> (Fast Fourier Transform, chapter 16) over the
whole take, and the &lt;strong>STFT Map&lt;/strong> (Short-Time Fourier Transform) and &lt;strong>Wavelet Map&lt;/strong> (chapters 17 and 18)
at each time of it. This note gives how each size is worked out, and
what each method trades.&lt;/p>
&lt;p>All three give a period&amp;rsquo;s size as a &lt;strong>sine&amp;rsquo;s size&lt;/strong>: half its peak to
peak, in the series&amp;rsquo; unit. For the rate that is the beat timing, in ms,
except the FFT&amp;rsquo;s default, which converts it to the swing in rate readings
(below).&lt;/p></description></item><item><title>M12. Isochronism and Rate against Amplitude</title><link>https://opentimegrapher.org/math/m12-isochronism-and-rate-against-amplitude/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://opentimegrapher.org/math/m12-isochronism-and-rate-against-amplitude/</guid><description>&lt;h2 id="what-it-is">What It Is&lt;/h2>
&lt;p>A balance is &lt;strong>isochronous&lt;/strong> when its period doesn&amp;rsquo;t depend on how far it
swings. A real one isn&amp;rsquo;t quite: the hairspring, the escapement&amp;rsquo;s impulse
and the balance&amp;rsquo;s poise all make the period move a little with the
amplitude. As the mainspring lets down, the amplitude falls, so a watch&amp;rsquo;s
rate moves through the day with it.&lt;/p>
&lt;p>The Isochronism pane (chapter 19) measures that: the rate at each
amplitude, the slope of rate against amplitude, and where the rate
changes least. The Positions Test&amp;rsquo;s full wind against 24 hours (chapter
20, M8) measures it another way: the change in each position between two
states of wind.&lt;/p></description></item></channel></rss>