It is 1:40 in the morning, the second shift is running a long roll of backlit film, and at roughly twelve meters the operator sees the first faint band. By twenty meters there is a ladder of them. The foreman does what everyone does at that hour: pulls a cleaning, swaps a damper, bumps the heat, then starts pricing a printhead. Four hours later the roll is scrap and the machine is in pieces. The band was never a mystery. It had a spacing, and the spacing was telling him which subsystem was failing.

Before you touch a single part, put a ruler on the print. Banding is one of the few faults that hands you a measurement, and that measurement narrows the suspect list faster than any service manual. The method below works on solvent, eco solvent, UV, latex, and DTF machines, because all of them lay down ink in swaths and all of them move media in steps.

Measure the Gap Before You Touch Anything

You need a steel ruler with millimeter marks, a loupe or a phone camera with a macro setting, a pencil, and the print itself. Do not clean the head first. Cleaning changes the evidence and you will spend the rest of the night chasing a fault that has already moved.

  1. Find a section with at least six consecutive bands and mark them on the back of the print with the pencil.
  2. Measure center to center across ten bands, then divide by ten. One band to the next is too small to read reliably, and ten gives you an average that survives a shaky hand.
  3. Write the number, the pass mode, the print speed, and the media name directly on the print. Without the pass mode the number means nothing.
  4. Note the direction. Bands that run across the media, perpendicular to travel, are feed bands. Bands that run along the media, parallel to travel, are scan bands and point at the carriage or the head instead.
  5. Count how often the pattern repeats across the full width. A repeat that matches the head width is a head or head alignment symptom. A repeat that matches a roller circumference is mechanical.

Photograph the marked bands next to the ruler before you write anything down. A photo carries the pitch, the direction, and the color of the defect in one frame, and it settles the argument with a supplier or a service technician faster than any description you can type into a message.

Two minutes of measurement gives you a number. Everything after this is comparison, and comparison is cheap.

Reading a Pitch That Matches the Feed Step

Every print mode advances the media by a fixed amount after each pass. That amount is the swath height divided by the number of passes used to build it, and your RIP or the machine menu will tell you the pass count. If the measured band pitch lands on that number, or on a simple fraction of it, the ink is being laid down correctly and the media is arriving in the wrong place.

Feed related banding has a signature. It is usually lighter or darker in a regular rhythm, it often starts after a certain length of roll rather than immediately, and it gets worse as the roll gets heavier. That last detail matters: a roll that has been unwinding for twenty meters has a different drag profile than it did at meter one, and the feed motor compensates with a correction table that is only as good as the last calibration.

  1. Mark the feed roller with a strip of tape and watch it turn while the machine prints a test strip. If the band appears once per revolution, you are looking at roller eccentricity or a flat spot, not the head.
  2. Check the encoder strip and the encoder sensor for mist and dust. A dirty strip gives the carriage and the feed a position that is slightly wrong, and the error repeats at the disc or strip resolution, which looks like fine banding at a pitch you cannot match to anything else.
  3. Reduce tension on the feed side and reprint ten meters of the same file. If the pitch changes, the problem is media handling, and no printhead on earth will fix it.
  4. Check the vacuum level and the pinch rollers. A sheet that lifts slightly between passes receives ink at a different distance, and the swath edges show it as a band.

If the pitch tracks the feed step and the roller marks line up, stop looking at the head. You are one calibration, one encoder clean, or one tension adjustment away from a sellable roll.

One caution before you recalibrate. Media feed correction values are stored per media type on most machines, and a value learned on a thin banner will make a heavy canvas drift. If your bands started the day someone loaded a different substrate and reused the old setting, recalibrate for that substrate and print a fresh strip before you look anywhere else. This is also why the pitch tends to move when you switch from film to paper and back: the machine is applying a correction that belongs to a different material.

Reading a Pitch That Matches Nothing You Can Name

Now the spacing is irregular, or it is exactly the width of the head, or it appears only in certain colors. This is the family of faults that lives in the head and the carriage, and the diagnostic tool is the nozzle check, printed at the same pass mode and the same speed as the failing job.

Put the nozzle check next to the banded print and line them up. If the light bands correspond to nozzle columns that are missing or deflected in the check, you have your answer and it is not a feed problem. Deflection is the one people miss: a nozzle that fires at an angle still puts down ink, so the check looks almost complete, but its dot lands beside the line it should be in and leaves a thin pale stripe that runs the full length of the print.

  1. Print the check twice. A defect that stays identical is hardware. A defect that moves or clears is ink, air, or temperature.
  2. Look at the edges of the band. A sharp edge with a clean white line means a missing nozzle. A soft, gradual transition means overlap or a height problem.
  3. Check head height and head slant. A head that is slightly too high throws dots off center; a head that is slanted produces a band whose pitch equals the head width and whose position drifts across the roll.
  4. Check for static on film and synthetic media. Static pulls the mist sideways and produces bands that appear and disappear with the weather and the room humidity.

One useful habit: keep a nozzle check from the day the machine was installed. Comparing today against a known good pattern takes seconds and settles arguments about whether a nozzle was always like that.

Four Mistakes That Send You Shopping Too Early

Most unnecessary printhead purchases in this trade start with one of these four moves. Each one destroys information you cannot recover.

  1. Cleaning before measuring. A powerful cleaning can hide a supply fault for twenty minutes and make a head fault look better or worse. Measure, photograph, then clean.
  2. Changing the profile or dropping the pass count. Running more passes makes finer bands invisible without fixing anything. The fault is still there, now buried in a slower job that costs more ink per square meter.
  3. Blaming the head when the encoder strip is dirty. Mist contamination on the strip and the sensor is common on solvent machines and produces banding that looks exactly like a failing head. It wipes clean with the right fluid and a lint free swab.
  4. Ignoring heat and vacuum. A platen that is a few degrees off, or a vacuum fan that has lost suction because the filter is loaded, changes how the ink dries and how the media sits. Both produce bands that move when you change media width, which is the tell.

The fourth mistake deserves one more sentence because it is the most expensive. Replacing a head removes the part that was showing you the fault, and the new head often prints clean for a week. Then the same band returns, because the encoder, the damper, or the vacuum was the cause all along. You have now paid twice and learned nothing.

The pattern in all four is the same: an action taken before a measurement. The measurement is free.

Decision Tree: Match the Spacing to the Subsystem

Before you start, write the measured pitch at the top of a sheet of paper and leave space under each branch for a result. It sounds like paperwork, but the value is in the record: the next time this happens, the sheet tells you what the pitch was and which branch you already ruled out. Without it, every event starts from zero and the same wrong part gets ordered twice.

Work down this list and stop at the first match. Do not run two branches at once, because you will not know which change fixed it, and the fault will come back on the next long roll with no explanation.

If the pitch equals the feed step or a fraction of it: go to media feed. Encoder strip and sensor first, then feed roller condition, then tension, then vacuum. This branch costs cleaning fluid and time, not parts.

If the pitch equals the roller circumference: go to the feed roller itself. Look for a flat spot, dried ink build up, or a pinch roller that is not gripping. Replace or resurface, then recalibrate feed.

If the pitch equals the head width: go to head alignment and height. Check slant, check the head gap, check that the carriage is level. A band at head width is a stitching problem, not a clog.

If the bands line up with specific nozzle columns: go to the head and its supply. Nozzle check, damper, cap, then head. This is the only branch where a printhead purchase is on the table.

If the bands appear only on one edge of the media: go to media handling and heat. Edge curl, platen gaps, and uneven heater zones all behave this way, and all of them get worse as the roll diameter shrinks.

If the bands appear and vanish with humidity or the time of day: go to static and ink viscosity. Ground the machine, control the room, and check that the ink is within its working temperature range.

When the Band Travels With the Image, Not the Media

There is one more case worth separating, because it fools experienced operators. Print the same file twice, once at the start of a roll and once at the end, and mark the band positions relative to a fixed graphic element. If the band stays in the same place on the image while the media moves underneath it, the fault is in the data path or the carriage, not in the feed.

Typical causes are a failing encoder sensor that drops counts at a specific carriage speed, a ribbon cable that makes intermittent contact as the carriage flexes, or a RIP that is dithering a gradient into a visible step. The last one is more common than people admit. A smooth sky printed with a coarse dither can produce a band that is perfectly regular and has nothing to do with the machine.

Test it by rotating the image ninety degrees. If the bands rotate with the image, the file or the RIP is producing them. If the bands stay perpendicular to media travel, the machine is producing them. That single test has saved more heads than any service call.

One last detail worth knowing: the pitch can be correct and the band can still be caused by a profile that is clipping a channel. If the band is a step in density rather than a line, look at the ink limit and the linearization before you look at hardware, because a machine cannot fix a curve that was built for another media.

Write the measurement, the pass mode, the branch you took, and the result into a notebook that stays with the machine. After three or four events you will recognize the pitch before you reach for the ruler, and the 1:40 a.m. roll will stop becoming scrap.