Pull the carriage cover off a machine that is printing crooked and most shops start pricing a CR motor before they have touched anything. That is the expensive guess. On a repair bench the motor is usually fine. The belt is stretched, a pulley is glazed, the rail is dry, or a bundle of ink tube is dragging on the frame. The motor gets blamed because it is the part that runs hot and makes noise. It is also the part that costs the most and takes the longest to land on your floor. Before you order one, give twenty minutes to the parts around it. This covers the two ends of the carriage drive: the CR motor and the belt and pulley set it turns. What each one does, how each one fails, and the order to check them so you do not buy a motor to fix a belt.
Why the Belt Wears Out Long Before the Motor Does
The CR belt is a toothed timing belt running from the motor pulley across the frame to an idler or tensioner at the far end, with the carriage clamped to one span of it. Every pass, every head clean, every homing cycle drags those teeth through the same two pulleys. Teeth do not snap. They round off. The belt also creeps a few thousandths longer over thousands of cycles. The rubber hardens with age and with solvent mist in the air, and a hard belt loses its grip on the teeth long before it looks bad. Neither change shows up on a nozzle check.
Two things hide that wear. Firmware compensates. Modern machines adjust carriage timing inside a tolerance window, so a slightly stretched belt still puts down acceptable output. The correction masks the fault right up to the point where it cannot. And the symptoms look like something else. Banding reads as an alignment issue. Dot placement error reads as ink starvation. Shops burn an afternoon on cleaning and alignment, chasing a mechanical fault with software tools.
The eye test on a belt is weak. It sits behind the carriage, half hidden by the head assembly, black on black. From a foot away a worn belt and a healthy one look the same. You need a light, a cloth, and your fingers.
Power the machine down and unplug it. Open the carriage cover and wipe the back face of the belt once, in one direction, with a clean lint free cloth. Then look at the cloth. Light gray dust is normal. A dark smear of black rubber powder means the belt is shedding material and the teeth are going. Next, press the middle of the long span with one finger. A healthy belt feels like a guitar string, tight with a short spring back. If it sinks and stays soft, tension is low or the belt has crept long.
The pulley tells on the belt too. Run a finger around the motor pulley groove. It should feel like clean machined teeth. If the groove is polished bright, or you can catch a fingernail on a step where the belt has worn a shoulder into it, the pulley is finished. A worn pulley will shred a new belt in weeks, and then everyone blames the belt.
The motor is a different animal. Nothing inside it wears against a toothed surface all day. Most of these are brushless, with a rotor on bearings and no consumable contact face. When a CR motor dies it is usually because something outside it made it work too hard, or because its driver on the main board failed and took the winding with it. That is the whole reason to check the belt, the rail, and the tube bundle first. You are not being cheap. You are refusing to replace the part that reports the fault instead of the part that caused it.
Reading the Belt and Pulley Symptoms
Five signatures point at the belt and pulley set rather than the head. None of them is proof on its own. Two together, on a machine with a clean nozzle check, is enough reason to pull the cover.
Banding That Ignores Cleaning
Print a solid patch in one color at production speed, then look at the lines. If they run the full width of the media and stay evenly spaced, the head is no longer the suspect. Run a nozzle check. A perfect check next to a banded patch moves the fault to media feed or carriage drive. Belt micro slip lands here because the carriage is a fraction late on some passes and on time on others. Cleaning cycles never touch it, which is the tell.
Bi-Directional Misalignment That Comes Back
Print the bi-directional alignment pattern, run the adjustment, print it again. If the left to right and the right to left passes still sit slightly off after the machine accepted the correction, the carriage is not returning to the same position between passes. That is a tension or tooth engagement problem. A belt that has crept long cannot hold registration across the width, and the firmware keeps applying a correction that is already wrong by the time it lands.
A Carriage That Sounds Different
A healthy carriage has a flat, even cadence. Stand next to the machine for thirty seconds with the cover open and listen. A worn belt gives a slightly irregular rhythm, and the pass sounds softer at one end of travel than at the other. Experienced operators catch this weeks before the prints go bad. If you own two of the same model, run them side by side and compare. Your ears are a free diagnostic and almost nobody uses them.
Quality That Improves When You Slow Down
Reprint the failing file at half speed. If the banding and the misregistration mostly clear up, tension or tooth engagement is marginal. At low speed the carriage has time to settle and slack does not matter. At production speed it does. A machine that only fails fast is telling you the drive is at the edge, not that the head is tired. This test costs one piece of scrap media. Do it before you touch the head, because a slow print that recovers proves the ink path was never the problem.
What You See Under a Flashlight
Look along the belt, not at it. Edge fraying means the belt is running off center. A shiny, glazed tooth face means friction heat. Teeth that should have crisp square edges but look rounded are past saving. Check the bottom of the belt channel for fine black powder. Then confirm the belt sits in the middle of each pulley groove instead of riding a flange. A belt climbing a flange wears its edge, drifts the registration, and eventually walks off the pulley.
What a Real CR Motor Failure Looks Like
The motor does fail. It is simply not the first thing to test. Here is how to prove it instead of guessing.
Kill the power at the wall, not just at the panel. Then separate the two suspects with your hands. First, unhook the belt tension or release the carriage from the span and push the carriage from one end to the other. It should glide with even resistance and no tight spots. A gritty patch, a scrape, or a hard stop points at the rail, a foreign object, or a tube bundle. That is drag, and no motor will fix it. Then spin the motor shaft by hand with the belt off. It should turn with light, even drag and a faint cogging feel. A shaft that feels rough, catches, or rocks side to side has a bearing or debris problem.
Heat tells you about load, not about the motor. Warm after a long run is normal. Too hot to hold your fingers on after a single print is not, and it almost always means the motor is fighting something. Walk the carriage path and look for the usual causes: a dry or contaminated rail, old grease mixed with ink mist turned into paste, a belt tensioned too high, an ink tube bundle pulling or rubbing the frame, a scrap of cured ink or a displaced plastic part in the travel path. On converted machines with external ink supply, check tube routing first. A tube that snags once per pass adds drag on every stroke.
Winding failure is measurable. Get the coil resistance for your model from your own service manual and measure across the motor connector with a meter. Do not compare against a number from a forum post about a different machine. If the winding reads shorted or open, stop and think before you fit a new motor. A shorted winding usually means the driver on the main board has failed, and it can take the replacement with it. Some manuals tell you to replace the main board at the same time. Believe the manual.
The manufacturer's own paperwork puts the motor near the end of the list. Service literature for wide format Epson models ties the CR overload condition, reported as code 113A on the models that use that code, to carriage load, overcurrent, jams, belt tension, insufficient lubrication, and frame distortion. The same documentation, for a CR servo parameter error, lists the checks in this order: clean and inspect the CR scale, adjust belt tension, replace the encoder sensor, replace the CR motor, replace the main board. That is the vendor telling you the motor is fourth, not first. Copy that order into your own routine.
One more trap. A dirty or scratched encoder strip makes the carriage hunt. It overshoots, corrects, and slams back, and the motor heats up because the machine keeps asking it to fix a position error that is not real. If you already replaced a motor and the fault returned in a few weeks, look at the strip and the sensor before you buy the second motor. Clean the strip with the correct solvent, one direction, on a lint free cloth, then hold it against a light and look for a bright scratch line. If you can see a scratch, cleaning will not save it.
Too Loose Is Not the Only Failure
Shops worry about a loose belt because a loose belt is obvious. A tight belt is the quieter killer, and the manufacturer documents it.
Service manuals for the wide format Epson line describe both ends of the tolerance band. Over tensioned, belt life shortens, the carriage runs under a higher load, the machine spends more time in wait control to keep from overheating, and the motor shaft is pulled off axis until the brush inside the motor wears. The documented end of that chain is a CR overload error. Under tensioned, the belt teeth slip, the carriage swings, the active damper correction stops working, and vertical bands show up near the media edges. Both ends land you at the same fault from opposite directions.
This is why a belt change done with a wrench and a feeling is a bad repair. Tension is a specification. On machines that call for it, the manual specifies a sonic tensimeter and a target value for your model. If you do not own the gauge, follow the manual adjustment procedure and the reference position it gives instead of tightening until it feels right. The instinct to add a quarter turn is how a machine ends up with a stretched belt and a worn motor brush in the same quarter.
Pulley alignment is part of the same job. After the belt goes on, turn the pulley by hand through a full revolution and watch the tracking. The belt should sit centered in each groove, not climbing a flange and not hanging off the outer edge. A belt that runs against a flange frays on that edge, sheds rubber, and slowly walks the registration out. If it will not center, the pulleys are not parallel or the idler bearing is going, and the belt will not last no matter how carefully you tensioned it.
Rail lubrication belongs here because it is the most common cause of a motor that runs hot. Use the lubricant the manual specifies. Do not use a general purpose penetrating oil, and do not use light machine oil. Both attract dust and ink mist and turn into an abrasive paste that grinds the carriage shoes. If the rail already carries old grease, wipe it off completely before you add anything. Fresh grease over contaminated old grease is worse than a clean dry rail. Wipe the rail with a dry cloth first and look at what comes off the cloth.
Then write it down. Date, machine hours, condition of the belt, tension reading if you took one. After three changes you will know your own interval from your own machines, which beats any number from a supplier page. A shop running one shift and a shop running two shifts do not wear a belt on the same schedule, and no chart knows your duty cycle.
The Check Order and What to Keep on the Shelf
Here is the sequence, cheapest first. Stop at the first check that fails and fix that one thing before you move down the list.
- Hand travel test. Power off at the wall, push the carriage the full length. Tight spot, scrape, or hard stop means rail, object, or tube. No cost.
- Tube bundle. Trace every line from the carriage to the bottle. Look for a tie wrap pulling the bundle tight, a tube rubbing the frame, a line sitting where the carriage will catch it. No cost.
- Encoder strip and sensor. Wipe the strip one direction with the correct solvent and check the sensor window for ink mist. Low cost.
- Belt surface, pulley grooves, rubber powder. A light and a cloth. No cost.
- Belt tension against the manual. Gauge if you have one, procedure and reference position if you do not. Low cost.
- Belt and pulley replacement. Half a day of downtime on most frames. Moderate cost.
- Coil resistance on the motor. A meter against your manual value. Low cost and decisive.
- Motor replacement. Expensive, and only after the driver question is settled.
- Main board. Last, and only with evidence of a failed driver.
Notice what is not near the top of that list: the printhead. If the nozzle check is clean, the head is out of the conversation. Chasing alignment and cleaning on a mechanical fault burns a day and produces nothing you can bill for.
For the shelf, keep belts, an idler or tensioner, the specified rail grease, lint free cloths, and the correct cleaning solvent for the strip. That covers most carriage faults for a small spend. A CR motor is a different decision. It fails rarely, but when it does the machine is down until the part lands. If you run several machines of one family, one motor on the shelf pays for itself the first time you skip a three day wait. If you run a single machine, put that money into a belt and a tension gauge instead.
Price the downtime, not the part. Take your hourly output on that machine, multiply by the hours actually lost including the days spent waiting for delivery, then add the media you scrapped, the express freight, and the technician visit. On a machine that earns money every hour, a three day wait costs more than the gap between a belt and a motor, and far more than the gap between a good belt and a cheap one. That arithmetic is the reason the cheap check order is also the profitable one.
Run the hand test first. It takes five minutes, it costs nothing, and it splits the whole problem in two: drag somewhere in the mechanism, or a drive that cannot hold position. Everything after that is a smaller question.