Densitometer daily five-minute check

Purpose

To establish, in about five minutes, that the instrument is still the instrument its certificate describes — and to know before the first real reading rather than after the last one.

When to use it

At the start of every session, before any reading anybody will act on. Again whenever anything at all has been touched: the head refitted, the aperture plate moved, the drive current changed, the firmware reflashed, or the instrument carried between rooms.

It is not a calibration. The scale, the linearity, the repeatability figures and the working range are established by the calibration experiment, at the interval the certificate names — X-Rite recommend once a week for their own instrument, and a home instrument in a warmer, dustier, more mobile place should not stretch that.

Before you start

You should already have the instrument certificate: the warm-up rule in a sentence from your own drift curve, the step you calibrate on and its certified value, the working range, the stray-light ceiling, the placement tolerance, the repeatability figures, and the uncertainty band table. Without it there is nothing for this check to compare against.

On the bench: the reference wedge, handled at the edges only; the white tile and the black trap if you will read prints today; a soft camel-hair brush; the terminal, logging to a file before the first reading rather than after the third; the certificate; the notebook.

Bring the wedge into the room an hour before if it has come from a cold cupboard. A cold reference puts a transient into the zero that you will spend the session chasing.

Procedure

1. Read the instrument’s own label against the build record — aperture, gap, drive current, firmware version — and stop if any of them has changed without your knowing.
2. Switch on and start the warm-up your certificate specifies. Open the log at the same moment.
3. Take the dark reading, lamp off, head closed. It is a property of the electronics and it moves with temperature.
4. Take the zero, lamp on and the stage empty. It is a property of the lamp and the optics, and it is the reading that drifts most: a commercial instrument has a whole procedure for re-taking the zero alone, for exactly that reason.
5. Put the calibration step on the stage, against both fences, and read it.
6. Compare it with the certificate’s value for that step. Within 0.02 D, the instrument is calibrated and you may work. Outside it, recalibrate before anything else.
7. If you will read prints, fit the reflection head and re-anchor in this order: the white tile first, which sets the gain, then the black trap, which is measured on that gain. Reversed, the ratio means nothing.
8. Re-anchor again every time the head comes off and goes back on, without exception. A refitted head is a different geometry until proved otherwise.
9. Place every sample within the placement tolerance on the certificate, against the fences for film and flat against the port for a print, with the same backing behind every print and the same settling time.
10. Take readings. Re-take the zero whenever you have been reading for a while, and always before a reading you care about: the drift is multiplicative and lives in the zero, so a re-zero removes it.
11. Quote every density with the uncertainty for its band, from the certificate’s table, and not with a single figure for the whole range.
12. Report nothing above the density at which your residual left the band. Above the ceiling, the instrument is reporting its own stray light rather than the sample.

How you know it worked

• The dark and the zero are two readings in the log, taken at two moments, not one reading used twice.
• The calibration step read within 0.02 D of its certified value, and the log says by how much rather than merely that it passed.
• The warm-up that actually ran is the one on the certificate.
• Every reading in the session has a zero behind it that is not stale.
• Every density leaving the session carries a band and an uncertainty, and any reflection density carries the tile’s serial with it.

If a step fails

Step 2, you have no certificate and therefore no warm-up rule, no working range and no bands. Then this check has nothing to check against. Run the calibration experiment first. An instrument that repeats itself is not an instrument that is right.

Step 3, the dark reading has moved since the last session. That is the electronics and it usually means temperature. Let the head reach room temperature and take it again. A dark that keeps moving while the room does not is a fault rather than a warm-up.

Step 5, the reference has a fingerprint or a mark on the measurement area. Do not read through it. Fingerprints and foreign substances on the measurement area cause errors, and nothing but a soft camel-hair brush goes near it — a cloth, a solvent or a tissue may change the densities themselves and then the certificate describes a wedge you no longer own.

Step 6, the step reads outside 0.02 D. Recalibrate, and look at the zero first: it is the term that drifts. If a fresh zero brings it back inside, the instrument was fine and the zero was stale. If it does not, work through the head page’s commissioning tests in their order — dark, stray light, ambient leakage, mechanical repeatability — because each one holds the others fixed and tells you which stage to go back to.

Step 6, the computed slope is far from unity. More than about ten per cent from 1.000 is a fault rather than a calibration. The usual causes, in order: a stale zero, a certificate value entered wrongly, or the step read at the wrong place on the wedge.

Step 7, the tile is not to hand. Then there is no reflection scale today. Read transmission only. A reflection density without its tile is not a number anybody else can use, and writing one down without the serial is the omission that costs you the day you compare it with somebody else’s.

Step 10, a series drifts steadily through the session. Finish it, then re-read the first sample. If the two disagree the zero went stale mid-series: re-zero and repeat the series rather than correcting it arithmetically.

Step 11, you are about to write a single accuracy figure. Do not. “Accurate to 0.01 D” confuses accuracy with repeatability, and “traceable” claims an unbroken chain of comparisons that a calibrated wedge gives you exactly one link of. Write what the certificate is entitled to say and no more.

What to record

Three lines in the log before the session’s first real reading: the dark, with its spread; the zero, with its spread; and the calibration step, with the certificate’s value beside it and the difference between them. Then, for a reflection session, the tile reading and the current it was set at. Ten of those daily lines on one page make a drifting instrument visible, which is something no single check can do — and the day the calibration step drifts out, the page tells you whether it went suddenly or had been going for a fortnight.