Break/Fix: The Timer That Lies

The situation

You printed for four hours on Saturday. One negative, one box of paper, one filter, the aperture untouched all evening, and the timer commanded 11.4 seconds every single time because that is what the test strip said and you never changed it.

On Sunday you lay the prints out in the order you made them, and they are not the same print. The first three match. So do the last three. The first three and the last three do not, and the change between them is smooth enough that no adjacent pair looks wrong — which is why you did not notice on Saturday and would not have noticed on any Saturday.

And at the end of the evening, when you made a fresh five-band third-stop strip to check yourself, the bands were not evenly spaced. The two shortest sat closer together than they should have.

Two faults, or one fault with two faces. The whole of this page is finding out which.

Before you start

Do not change anything yet. The commonest way this diagnosis is lost is that somebody replaces the developer, resets the timer and re-does the strip before collecting the evidence, and then has a working darkroom and no idea what was wrong with it. Everything on the bench stays as it is until the evidence section is complete.

Get the log off the timer. It is the one piece of evidence that expires: it lives in a captured serial session, and if you did not capture it, tonight’s session is the one to capture.

Read the two pages that own most of the mechanisms, because this page will not re-derive them. The calibration page owns the three exposures, the offset and the error budget. The troubleshooting atlas entry for session exposure drift owns the general shape of the fault and the reason a heatsink is a photometric component. What this page adds is the discrimination: nine candidate causes, and a test that tells each one from the others.

Alternative route

Every symptom on this page appears on a print, so the evidence is gathered in a darkroom and there is no route that avoids one. What a reader without a permanent dark room can do is real, though, and it is worth stating: the whole diagnosis fits inside one two-hour session in a temporarily blacked-out room, because the fault is defined by what changes across a session and two hours is a session. Part XVI’s room page lists the blackouts that make one.

Three of the tests need no dark room at all and can be done first, in daylight, which is a good reason to do them first: reading the timer’s log, measuring ten repeat intervals at the easel with the photodiode head, and metering the supply. If any of those three finds the fault, the darkroom session becomes a confirmation rather than an investigation.

The symptoms

Four shapes, and they mean four different things. Before anything is measured, plot what you have: line the prints up in the order you made them, pick one mid-tone patch that appears in all of them, read it on the densitometer, and plot density against print number.

Four session drifts, plotted the same way, that mean four different things

• A source still warming: early excursion, then flat
• Supply variation: smooth, periodic, no trend
• Developer exhausting: flat, then steepening
• Timer jitter: flat mean, no correlation

Four shapes drawn to be told apart, not measured from any session. What is real in the drawing is the four signatures: an excursion that happens early and stops, a smooth oscillation with no trend, a decline that starts late and steepens, and a scatter with no order. Read the shape before you reach for a meter. The curve is drawn to show the shape, not measured from a real material. Your own materials will differ, and measuring them is what the sensitometry part of the course is for.

A monotonic drift through the whole session, smooth, in one direction. Something is changing steadily: a source’s temperature, a developer’s strength, a lamp’s age over a very long session.

An excursion at the start that then stops. That is a warm-up, and it is the cheapest fault on the page to fix because the fix is a rule rather than a part.

A step change after a break. Something switched off and came back different. A head that cooled during a tea break and had to warm up again; a supply that was unplugged; a bath that was topped up.

Random scatter, print to print, with no trend at all. That is jitter or a mechanical fault, and it is the one shape that the developer cannot produce, because a bath cannot get stronger between two prints.

And a fifth, which is not a drift. A systematic error that appears only at short exposures — bands compressing at the bottom of a strip, or a burn of half a second doing nothing visible while a burn of five seconds works as expected. That one has its own section, because it is not a fault in the sense the others are: it is the system’s short-exposure floor, which the calibration page measures.

!The one analysis that costs nothing and separates jitter from everything else

Take your column of densities and compute the difference between each print and the one before it. Then compare the size of those differences with the size of the whole spread.

For any smooth process — a warming source, an exhausting developer, a wandering supply — neighbouring prints are made minutes apart and the process has barely moved, so the consecutive differences are much smaller than the total spread. For jitter, each print is an independent draw, so a consecutive difference is as large as the spread itself.

That is a five-minute arithmetic on data you already have, and it splits the nine causes below into two groups before you touch anything.

Evidence to collect

Ten items. Collect all ten before forming an opinion, because the expensive mistake on this page is to find the first plausible cause and stop.

Evidence	Where from	What it settles
The timer’s serial log for the session	The captured terminal session	Whether the timer delivered what it was asked for, every time, and whether anything was aborted or fired twice
Ten repeat measurements at the easel	The photodiode head, ten identical commanded exposures now	Whether the light delivered is repeatable today, independently of what happened on Saturday
The supply voltage	A meter, logged over an hour with the head switching as it did	Whether the supply moved, and whether it moved with a period
The head’s temperature, start and end	Infrared thermometer or thermocouple, from the head’s own log	Whether the warm-up rule was actually obeyed, as opposed to written down
The lamp’s age in hours	The head’s record, if you kept one	A slow term that only matters over months, and a reminder to start keeping one if you did not
Developer temperature, and prints through the tray	The thermometer, and the print count	Against the maker’s published capacity for your dilution, not against a feeling
The developer’s mixing date and time	The bottle’s label	An oxidised stock is a different fault from an exhausted tray
The safelight test card and its date	The safelight’s certificate	Whether the room’s fog is where it was when it was last measured
The foot switch, and whether it was used	Your memory, plus the log	Two log lines where you made one press is a bouncing pedal, and it is unmistakable
The prints themselves, in order, with their densities	The densitometer	The plot above, which is the whole of the first diagnosis

📓Why the log is the instrument that makes this tractable

Without a log, “did the prints drift?” is an argument between two people’s memories. With one, it is a column you can plot.

The timer writes one line per exposure with a serial number, a millisecond tick, the mode, the base, the divisions, the band, the stops, the commanded interval, the delivered interval, the offset applied and a flag. Four things fall straight out of it and cost nothing.

Delivered minus commanded, down the whole column. If it is constant, the timer is not your drift, and you have eliminated a third of the candidate list in ten seconds.

The scatter of that difference. That is the jitter figure, measured on the actual session rather than on a soak test.

The ticks. Gaps in the millisecond tick locate your breaks, which is where a step change will sit if you have one.

The flags and the count. An abort where you remember none, or two exposure lines 200 ms apart where you made one press, is a foot switch that bounced — and it names the sheet.

If your log for Saturday does not exist, that is itself a finding, and the fix is to capture the terminal session from now on. A printing session that leaves no record is a session that cannot be argued about.

Hints, in increasing order of disruption

Work down. Stop at the first one that explains everything, then confirm it rather than assume it.

1. Read the log’s delivered-minus-commanded column. Costs nothing, needs no darkroom, and eliminates the whole electrical half of the list if it is flat.
2. Compute the consecutive differences of the print densities. Five minutes. Separates jitter from every smooth process.
3. Count the prints through the developer and look up the published capacity for your dilution. ILFORD tabulate it in 8 × 10 inch prints per litre of working strength for each of their paper developers; find yours, and compare it with what the tray actually did.
4. Check the head’s warm-up record against its certificate’s warm-up rule. Not whether a rule exists — whether it was obeyed on Saturday.
5. Meter the supply for an hour with the head switching as it did, and look for a period.
6. Ten identical exposures at the easel now, with the photodiode head, read as a spread and converted to stops. This is arm 4 of the calibration page run once.
7. Look for the foot switch in the log, then bench-test it: press it fifty times and count the log lines.
8. Run the four-sheet test below, which needs a session and settles exposure against development for good.
9. Re-run the calibration page’s arm 3 on the head you printed with, which is the expensive answer and the one that finds a fault nothing above can see.

The diagnosis

From the log to the cause, with the discriminating question at each branch

1. 1Is delivered minus commanded flat? — the log answers it in ten seconds and eliminates half the tree
2. 2Are consecutive differences as large as the spread? — separates a random process from a smooth one
3. 3Does the excursion happen early and then stop? — that shape is a warm-up and nothing else
4. 4Does the supply oscillate with a matching period? — a shared circuit with something that cycles
5. 5Does it survive fresh developer? — the four-sheet test: the only branch that needs printing

Five questions, in the order of how little they cost. The first four are read off records you already have; only the fifth spends paper.

The nine causes, and what tells each one apart

Cause	The mechanism	The discriminating test
LED warm-up and thermal droop	An emitter’s flux falls as its junction warms. Cree plot relative flux against junction temperature and the atlas entry reads the direction off that chart and deliberately not the ordinates. A junction that is still climbing is a brightness still falling	Repeat the first print after a full warm-up from cold and again after an hour. If the difference disappears, the source was warming. The shape is the giveaway: early excursion, then flat
A tungsten lamp on a moving supply	A filament’s temperature follows its power, and its output follows its temperature faster than linearly. A supply that wanders makes an output that wanders	Meter the supply for an hour and look for a period. The calibration page has you fit your own exponent, because no lighting reference giving one is in this course’s corpus
A tungsten lamp that has aged	A filament evaporates through its life and its output falls permanently. No lamp manufacturer’s figure for the rate is in this course’s corpus, so no number is given here	It cannot drift within one evening: the timescale is hundreds of hours. If your session drift is real and the lamp is old, these are two separate findings, and the second is fixed by writing the lamp’s hours on the head
Relay contact degradation or chatter	An arcing contact pits and its resistance rises; a marginal coil chatters. Neither exists in this instrument, which switches a transistor	If you are on the bought-timer route for a mains enlarger, this belongs to that appliance and the answer is the maker’s, not this course’s. Nothing in the Pure Silver timer has a moving contact in the lamp path
Leakage lighting the lamp between exposures	A switch that is off is not disconnected. The transistor’s own leakage is bounded at 25 µA by its datasheet, twelve parts per million of a 2 A array; a solid-state relay is quoted at 1.50 mA, and Omron note that it flows through the snubber even when there is no power at the input. If your mains enlarger is on the bought-timer route, that timer switches mains through a semiconductor and a semiconductor leaks	The switched output page’s T3: paper under the dark lamp for a session, half covered, against a control. A meter bounds the current and only the paper answers the question
Firmware timing drift or a scheduling pause	A garbage collection inside an exposure, or a loop that has grown a blocking call	Delivered minus commanded, in the log. A pause shows as one line tens of milliseconds long in a column that is otherwise tight
A foot switch that bounces and fires twice	Contacts bounce on closing; if the firmware’s debounce interval is shorter than the bounce, one press becomes two events	The log names the sheet: two exposure lines a fraction of a second apart where you made one press. Then bench-test it — fifty presses, fifty lines
The negative popping in a warm head	The film absorbs heat, expands against the carrier and shifts part way through the exposure, so the print is sharp for part of the time and displaced for the rest. No popping temperature or timescale for the films this course uses is sourced anywhere in it	It is a sharpness fault, not a density fault, and that is what tells it apart. A loupe on the print finds a displaced second image; a densitometer finds nothing
Developer exhaustion and cooling	Each print consumes developing agent from a bath of a few hundred millilitres. ILFORD publish the capacity in 8 × 10 prints per litre for each dilution and note that high temperatures reduce solution life considerably	The four-sheet test below. Nothing else settles it, because an exhausting developer and a dimming lamp produce the same monotonic drift on a print
Safelight fog accumulating	Kodak state that virtually all exposures are cumulative, so a sheet that spends longer out fogs more; and fog degrades highlights and lowers contrast before it becomes visible in the borders	It attacks the highlights and the contrast, not the mid-tone. If your drift is a mid-tone that moved with clean whites, it is not fog. Then re-run the fog test anyway, because filters fade

✕The two rows on that table that are not this page's

Relay contacts and lamp ageing are on the list because they are the two causes a printer is likeliest to be told about by somebody else, and both are dead ends here. This instrument has no contact in the lamp path at all — it switches a transistor, and the isolation lesson explains why. And a lamp’s ageing is measured in hundreds of hours, which cannot produce a change inside one evening whatever anybody says.

Recognising a cause that cannot produce the symptom you have is as much of a diagnosis as finding one that can, and it is faster.

The one test that separates exposure from development, for good

The two commonest causes — a source that is dimming and a developer that is exhausting — produce the same monotonic drift on a print, in the same direction, over the same timescale. No amount of looking at prints will separate them. This will, and it costs four sheets and 300 mL of extra developer.

Four sheets, two trays, two moments.

1. At the start of the session, with the working tray fresh, expose two identical sheets from your reference negative at your reference exposure. Process one in the working tray and one in a small tray of freshly mixed developer, immediately, at the same temperature and for the same time. Call them S_work and S_fresh.
2. Print your evening.
3. At the end, expose two more identical sheets. Process one in the working tray, which is now whatever it has become, and one in a tray of developer mixed fresh a minute ago. Call them E_work and E_fresh.
4. Read all four on the densitometer, at the same patch, above the same base plus fog.

Now read the four numbers, which decompose the drift completely.

Comparison	What it isolates	Because
Efresh − Sfresh	The exposure side, alone	Both were developed in identical fresh chemistry, immediately after exposure. Nothing about the development differs, so any difference came from the lamp, the supply or the timer
Ework − Efresh	The developer’s decline, alone	Both were exposed within a minute of each other. Nothing about the exposure differs, so any difference is what the working tray has become
Ework − Swork	The total drift you noticed on Saturday	This is the fault as it presented
The identity	A check on all three	The first two should add up to the third. If they do not, something is varying that you have not controlled — most often the temperature of one tray

Note what the design avoids. Every sheet is developed immediately after it is exposed, so no sheet sits as a latent image for hours while another does not, and latent-image keeping never enters the comparison. That is the confound that ruins the obvious version of this test, in which four sheets are exposed at the start and developed at intervals.

The uneven strip, which is a different question

The bands of a third-stop strip are supposed to be evenly spaced in density wherever the paper’s curve is straight, because equal steps in stops are equal steps in log exposure — a third of a stop is 0.301 ÷ 3 = 0.100 log H, and against MULTIGRADE RC DELUXE’s published range of 0.90 at filter 2 that is eleven per cent of the paper’s whole usable scale per band.

Three things make them uneven, and they sit at different ends of the ladder.

Three ways a third-stop ladder goes uneven, and where each one shows

1. 1Compression at the bottom — the lamp's rise, or the paper's reciprocity. Both act only at short exposures
2. 2Compression at the top — the paper's shoulder, working normally. It moves with density, not with band number
3. 3Rounding, everywhere and tiny — eight millistops at worst on the cumulative rule: a fortieth of a band

The diagnosis is where on the ladder the unevenness sits. Nothing about a strip needs an instrument to read in this much detail; a ruler on the plotted densities is enough.

Compression at the short end is the lamp’s rise and fall, or the paper’s reciprocity, and those two are separated by the calibration page’s second strip: repeat every band at four times the time with the aperture closed two stops, so the exposure is unchanged and the route to it is not. If the compression follows the times, it is the lamp. If it follows the illuminance, it is the paper.

Compression at the long end is the paper’s shoulder and is the paper working correctly. Test it by making the whole strip a stop lighter: a shoulder moves with density, so the compression follows it up the ladder.

Rounding is never the visible cause. The f-stop lesson establishes that rounding the cumulative rather than the increment leaves the worst band eight millistops out, against a third-stop band of 333 millistops. That is a fortieth of a band. If somebody tells you a strip is uneven because of rounding, the arithmetic says otherwise — unless they rounded the increments, in which case the firmware has been modified and the fault is that modification.

=And the honest limit on the short end

Below about a second the paper’s own response is related to the exposure it received by a factor this course cannot state. ILFORD publish a reciprocity correction for their films and say that exposures of one second or less need none of it; no equivalent figure for a developing-out paper was found in this course’s corpus at all.

So a strip whose bottom band is at 0.4 s and looks wrong may be showing you a timer fault, a lamp fault or a property of the paper, and the only one of the three this course can quantify for you is the first two. The calibration page states the floor as a floor for exactly this reason, and the practical instruction that follows is short: print in the range where the strip is readable, and reach for the aperture rather than the timer when the exposure falls below it.

The fix

The fix depends on the finding, and each one has a retest that must be passed before the instrument goes back into use.

Finding	The fix	The retest that proves it
A source still warming	Write the warm-up rule on the head, from its own drift curve, and obey it	Repeat the four-sheet test’s first pair after the full warm-up. Efresh − Sfresh should collapse
Developer exhausting	Mix enough for the session, or replace part way through and record the change on the print record	Print through a full session again with a control strip in every batch
Supply variation	Move the head to a circuit with nothing cyclic on it, or convert the head to a constant-current LED, which removes the term entirely	Meter for an hour again. A flat supply and a flat density column together
Timer jitter	Strip the exposure loop back — the firmware page’s further experiment measures the cost of each feature — or accept it and raise the shortest honest exposure on the certificate	The calibration page’s arm 4, at the exposures you actually print at
A garbage collection inside an exposure	gc.collect() immediately before the assertion is already in the published firmware; if it is happening anyway, re-run the soak after a fresh reset	Sorted extremes of a hundred exposures, not the mean
A bouncing foot switch	Raise the debounce interval to cover the measured bounce, having measured it rather than guessed	Fifty presses, fifty log lines. Not forty-nine and not fifty-two
Leakage in something you built	Find the leaking element — driver standby, an indicator, the transistor — and remove or replace it	The switched-output page’s T3 again, on paper, for a full session length
Leakage in a bought mains timer	Not yours to fix. The unit is a sealed certified appliance and it is not opened, drilled or rewired. Recognise it, and take the enlarger’s plug out of the timer between sessions so the lamp is disconnected rather than merely switched off	T3 on paper with the enlarger plugged in and the timer idle, then again with the enlarger unplugged. Two sheets settle it
Safelight fog	Replace the filter, move the lamp, or shorten the time paper is out	The fog test, with the timer’s display and the safelight together
A popping negative	Give the head a glass carrier, or pre-warm the negative in the beam and focus after it has settled	The loupe, on a print of a subject with a hard edge in it

Then re-print the negative you started with, at the exposure the log says you gave it on Saturday, and put the new print beside the first one from that session. Two prints, one variable — everything you changed. If they match, you are done. If they do not, you have fixed a real fault and there is a second one, which is commoner than anybody expects and is why the evidence section asks for all ten items rather than the first plausible three.

What to record

On the timer’s certificate, which the calibration page created and which this page amends rather than replaces:

• the fault, in one sentence, with the date it was found and the date it was fixed;
• the evidence that identified it, named — “the log’s delivered column, flat; the four-sheet test, all of the drift on the developer side”;
• any figure on the certificate that this changes: the repeatability, the shortest honest exposure, the warm-up rule, the offset;
• what was changed on the instrument, so the next person knows this is not the build the pages describe.

In the session record, which for this instrument is the captured serial log:

• the log file itself, dated and kept, because it is the raw data behind everything above;
• the print densities you plotted, with the patch you read and the densitometer’s certificate date;
• the four-sheet test’s four numbers and the identity check.

And one line on the instrument itself, in permanent marker, because that is where somebody looks in the dark: the calibration date, and the shortest exposure it is trusted for.

📓Where the course keeps these, and one gap it has not filled

The calibration record set carries sheet CA-7, which is the timer against a second clock: the trials, the deviation in seconds and as a percentage, whether the offset is constant or proportional, and the action. Use it for the annual check and for the check after any repair; this page does not duplicate the form.

The gap is the printing session record. The course’s exposure log is a camera log, not a printing one, and no printed sheet in this course collects a session of print exposures. For this instrument the serial log fills the hole and fills it better than paper would. For a reader on the bought-timer route it does not, and until the printing part lands, the honest instruction is a ruled page in the notebook with the negative, the filter, the aperture, the commanded time and the tray count on it.

Where this page stops

Three neighbouring faults belong elsewhere and are named here so that a search finds them.

Prints that are wrong from the first sheet rather than drifting are not this page. A print that was never right is a base-exposure or a grade problem, and it belongs to the printing parts.

Fog, flare and uneven illumination belong to Part XVI’s break/fix page, which separates five fog sources and eight sources of unevenness. If your whites are grey or one corner is lighter, start there instead.

Faults in the negative — popping aside, which is on the list above because it is diagnosed at the enlarger — belong to the failure-analysis part. A thin or contrasty negative produces prints that are hard to make and not prints that drift.

Two entries in the troubleshooting atlas carry the compressed versions of what is on this page: session exposure drift for the drift itself and timer display fog for the instrument as a fog source. Both were written before this part and both name it; where an atlas cell and this page disagree, the page is right and the cell is a bug.

Questions

1. Your log shows delivered minus commanded flat to within 2 ms down the whole session, and your print densities fall by 0.14 from the first print to the sixteenth. Name every cause on the table you have just eliminated, and say in one sentence why the log eliminated them.
2. A foot switch bounces and occasionally fires twice. Say exactly what the print would show and exactly what the log would show, and say why the log is the faster evidence even though the print is the thing you care about.
3. Design the shortest sequence of tests that separates a lamp fault from a developer fault without re-printing the whole session, and state how many sheets it costs.
4. The four-sheet test gives E_fresh − S_fresh = −0.02, E_work − E_fresh = −0.10 and E_work − S_work = −0.06. Something is wrong with this set. Say what, and give two things that could have caused it.
5. Your consecutive print-to-print differences average 0.04 and the whole session’s spread is 0.05. What process is that, and what is it not? Name one further check.
6. Somebody tells you the uneven bands at the bottom of your strip are a rounding error in the timer. Refute it with the arithmetic, then name the one circumstance in which they would be right.
7. You find and fix a warming head, retest, and the drift is halved rather than removed. Say what you do next, and why “halved” is a more useful result than “removed” would have been.

Further work

Instrument your next three sessions before you have a fault. Capture the serial log every time, read one patch on the first and last print of each session, and keep the numbers. Three clean sessions give you a baseline, and a baseline is what turns “is this drifting?” into “this is drifting by more than it did in March”.

Measure your foot switch properly. Put a second Pico on the pedal’s line, timestamp every edge with ticks_us, and press it fifty times. The distribution of bounce durations is a number the hardware page could only ask you to estimate, and it tells you whether the firmware’s debounce interval is generous or marginal for your pedal.

Make the developer’s decline a curve rather than a suspicion. Process one control strip per ten prints through a whole session and plot them. Against the maker’s published capacity for your dilution, you will find out whether the tabulated figure describes your practice — and if it does not, yours is the number that governs your darkroom.

Try the four-sheet test on somebody else’s darkroom. It needs no knowledge of their equipment, takes twenty minutes of their session and settles an argument they have probably been having for years.

Recap

• Four drift shapes mean four different things: an early excursion that stops is a warm-up; a smooth oscillation with no trend is the supply; a decline that starts late and steepens is the developer; and a scatter with no order is jitter.
• Consecutive differences separate the fourth from the other three, in five minutes, on data you already have.
• The log is the instrument. Delivered minus commanded, its scatter, the ticks and the flags eliminate half the candidate list before anything is printed.
• Two causes produce the same drift and cannot be told apart by looking. The four-sheet test — two trays, two moments, everything developed immediately — decomposes it completely and avoids the latent-image confound.
• Where the unevenness sits on a strip is the diagnosis: bottom is the lamp or reciprocity, top is the paper’s shoulder, and rounding is a fortieth of a band and never the answer.
• Three things this course could not source and does not state: a tungsten lamp’s decline with hours, a negative’s popping temperature, and any paper reciprocity figure at short exposures. Each is named as a gap on the row it belongs to rather than filled with a plausible number.

Check your understanding

Question 1. Your print densities fall smoothly from 0.96 to 0.85 over the first five prints and then sit at 0.85 for the remaining eleven. Which cause does that shape point at, and which does it rule out?

• Developer exhaustion, and it rules out the lamp
• A source still warming, and it rules out developer exhaustion — because a bath gets weaker as work goes through it, so its effect starts small and grows, which is the opposite shape
• Timer jitter, and it rules out the supply
• Safelight fog, and it rules out everything electrical

Show the answer and why

Answer: A source still warming, and it rules out developer exhaustion — because a bath gets weaker as work goes through it, so its effect starts small and grows, which is the opposite shape

The two shapes are mirror images and that is the whole diagnostic value of plotting the column. A warming source does most of its changing early, while the junction or the filament is climbing towards a temperature it then holds; an exhausting developer does almost nothing for the first few prints and then accelerates, because the agent is consumed in proportion to the work done. Jitter has no trend at all, and fog attacks the highlights and the contrast before it moves a mid-tone. Note that the shape is a hypothesis rather than a proof: the confirming test is to repeat the first print after a full warm-up.

Question 2. Why does the four-sheet test develop every sheet immediately after exposing it, rather than exposing four at the start and developing them at intervals?

• To save time in the session
• To keep latent-image keeping out of the comparison: a sheet that waits hours before development is not comparable with one that does not, and the difference would be attributed to the fault under investigation
• Because paper cannot be stored between exposure and development
• Because the developer would exhaust while the sheets waited

Show the answer and why

Answer: To keep latent-image keeping out of the comparison: a sheet that waits hours before development is not comparable with one that does not, and the difference would be attributed to the fault under investigation

The design exists to isolate two variables — the exposure side and the developer state — and a third variable that moves with the same clock would ruin it. Latent-image keeping is exactly such a variable: whatever it does over four hours, it does to the sheet exposed first and not to the sheet exposed last, so it would appear in the same column as the exposure drift and be indistinguishable from it. Immediate development removes it from the comparison entirely. Paper stores perfectly well as a latent image, which is why the naive version of the test is tempting.

Question 3. The log shows two exposure lines 180 ms apart on one sheet where you remember making a single press of the foot switch. What is the fault and what would the print show?

• A garbage collection; the print would be slightly dark
• A bouncing pedal whose bounce outlasts the firmware's debounce interval; the print would carry two exposures and be about a stop dark, and the log is the faster evidence because it names the sheet without your having to compare prints
• A firmware scheduling fault; the print would be unaffected
• The interlock refusing a second start; the print would be light

Show the answer and why

Answer: A bouncing pedal whose bounce outlasts the firmware's debounce interval; the print would carry two exposures and be about a stop dark, and the log is the faster evidence because it names the sheet without your having to compare prints

Two exposure lines for one press is a debounce failure and nothing else: contacts bounce on closing, and if the bounce lasts longer than the interval the firmware ignores edges for, the second closure is read as a new press. A sheet given its exposure twice is a stop dark. The log wins on speed because the print only tells you that one sheet is wrong, while the log tells you which sheet, when, and that the cause was two commands rather than one long one. Note the interlock is not implicated: it refuses a start while the output is already asserted, and these two exposures did not overlap.

Question 4. A five-band third-stop strip has its two lowest bands closer together in density than the rest. Which single test separates the two candidate causes?

• Repeat the strip at a harder grade
• Repeat every band at four times the time with the aperture closed two stops, so each band receives the same exposure by a different route: if the compression follows the times it is the lamp's rise, and if it follows the illuminance it is the paper's reciprocity
• Repeat the strip with the increments rounded rather than the cumulative
• Read the strip on a different densitometer

Show the answer and why

Answer: Repeat every band at four times the time with the aperture closed two stops, so each band receives the same exposure by a different route: if the compression follows the times it is the lamp's rise, and if it follows the illuminance it is the paper's reciprocity

Both candidates act only at short exposures, so neither can be excluded by where on the ladder the compression sits — which is why a further test is needed rather than a further look. Holding the product of illuminance and time constant while changing both by four is the classic way to split a time-dependent effect from an intensity-dependent one, and it is exactly what the calibration page's second strip does. A harder grade changes the paper's exposure range and would confuse the picture; changing the rounding rule would introduce a fault rather than diagnose one; and a second densitometer tests the densitometer.

Question 5. Which of these does this course decline to give you a number for, and why does that matter to a diagnosis?

• The developer's capacity, because ILFORD do not publish it
• A tungsten lamp's output decline with operating hours, a negative's popping temperature, and any paper reciprocity correction — none of the three is sourced in this course's corpus, so each is named as a gap on its own row rather than filled with a plausible figure, and a diagnosis that turned on one of them would be resting on an invention
• The transistor's off-state leakage, because datasheets do not specify it
• The safelight's safe time, because it depends on the room

Show the answer and why

Answer: A tungsten lamp's output decline with operating hours, a negative's popping temperature, and any paper reciprocity correction — none of the three is sourced in this course's corpus, so each is named as a gap on its own row rather than filled with a plausible figure, and a diagnosis that turned on one of them would be resting on an invention

Two of the distractors name figures the course does have. ILFORD publish paper developer capacity in prints per litre at each dilution, and the IRLZ44N datasheet bounds leakage at 25 microamps. The safelight's safe time is genuinely room-dependent, which is why the course gives a test rather than a number — a different situation from having no source at all. The three in the correct answer are true gaps, and the reason it matters is procedural: a row that says "unsourced" sends you to a measurement, while a row with an invented number sends you to a conclusion.

Check my answersClear