RTU Belt and Sheave Alignment

Purpose

A belt that broke six weeks after this shop replaced it did not fail early. It was destroyed by a drive geometry nobody measured, and the second belt will die the same way. Sheaves wear into a dish, motor bases shift on their slides, and a replacement motor almost never lands with its sheave in the plane of the one that came off. This procedure guarantees a belt leaves the roof running in the right plane at a measured tension. The quieter failure it prevents: an over-tensioned drive does not squeal, so it reads as a good repair, then takes out the motor and blower bearings over the season, and nobody connects that to the belt job because nobody wrote the tension down.

Scope

Covers V-belt drives on packaged rooftop units and light commercial air handlers a two-tech shop services with a ladder and a service truck: belt replacement, sheave inspection and condemnation, alignment, tensioning, and the run-in re-tension.

Does not cover changing the drive ratio to re-balance a system, which is an airflow decision needing a static pressure survey. Does not cover the annual service pass, owned by the light commercial RTU service SOP, or edge protection and anchorage, owned by the rooftop fall protection SOP.

Roles and responsibilities

Procedure

1. Open the disconnect, lock and tag it, and prove dead before the blower door comes off. Electrical utilization work is controlled by 29 CFR 1910.333(b)(2), not 1910.147, which excludes it at its own (a)(1)(ii)(C). Prove the meter on a known live source, test all conductors at the load side, prove the meter again, per NFPA 70E-2021, 120.5. Acceptance: zero volts line to line and line to ground on a meter that read live before and after. What wrong looks like: a clean, believable zero from a meter with a blown fuse. Stop rule: no live reading afterward voids everything since the first check and you start over. Hazard in this step: many RTUs carry a second supply for an electric heat section or a convenience receptacle, so an open unit disconnect does not mean a dead cabinet.

2. Record the drive exactly as it sits, before you disturb anything. Belt cross-section and length off the belt, sheave and bushing numbers off the faces, motor horsepower, speed, full load amps and frame off the nameplate, both pitch diameters, center distance, and turns open on a variable-pitch flange. Acceptance: every field filled or marked unreadable. What wrong looks like: ordering from the counter's guess after the old belt is in the trash. Stop rule: worn-off markings means taping the outside circumference around both sheaves and recording it as measured, not as a part number. Hazard in this step: brushing out belt dust and degraded duct liner puts rubber dust and respirable fibers into the compartment you are leaning into, so vacuum with a HEPA-filtered unit rather than blowing it out, and where liner sheds on contact the work needs respiratory protection under a written program per 29 CFR 1910.134.

3. Gauge the sheave grooves and condemn a worn sheave before fitting a belt to it. Set a groove gauge for the cross-section into each groove and read the light behind it. Acceptance: the belt's top surface sits flush with the groove top or up to about 1/16 in proud, sidewalls straight, and the belt clear of the groove bottom. What wrong looks like: dished, polished sidewalls with the belt riding low, a sheave that has spent a year machining itself on the old belt. Stop rule: a belt that bottoms condemns the sheave, so phone the service manager for sheave and bushing rather than fitting a new belt into a groove that will dish it to match within weeks. Hazard in this step: a taper-lock bushing drops the sheave the moment the last bolt clears, so support it from underneath and keep fingers off the groove edges.

4. Slack the drive with the motor base, never over the sheave rim. Back off the adjusting screw until the belt lifts clear of both grooves. Acceptance: the belt comes off with one hand and no tool. What wrong looks like: prying a belt over the sheave edge, which breaks tensile cords where you cannot see them, so the belt you just fought on is the next callback. Stop rule: a base that will not give enough travel is seized or was installed at the end of its adjustment, and both are findings to write up. Hazard in this step: a blower motor released on its rails drops under its own weight, so keep the hold-down bolts snug enough to control it and fingers out of the base slots.

5. Align the sheaves and measure the error rather than eyeing it. Lay a straightedge or taut string across the machined outer faces of both sheaves and look for contact at all four points. Acceptance: four-point contact, with residual offset inside the drive manufacturer's published limit, commonly about 1/16 in of offset per foot of centre distance for a standard V-belt drive, which is the form this procedure measures in; where the drive sheet gives an angular figure instead, convert it before comparing, because half a degree is nearer 1/10 in per foot and the two are not interchangeable. Use the published figure, since a narrow-section drive runs tighter than a classical one. What wrong looks like: one belt sidewall shiny and the other dull, or a belt walking up a groove flange under load. Stop rule: correct at the motor sheave on its shaft first and only then by shimming the motor, and do not tension a drive still outside the limit. Hazard in this step: hands go between sheave and belt here, so confirm the lock is still on the disconnect and never turn the drive by pulling the belt span.

6. Fit and tension the belt to a measured deflection. Measure the span between the points where the belt leaves the sheaves, then deflect the middle of that span 1/64 in per inch of span at the deflection force the belt manufacturer publishes for that cross-section. Acceptance: the reading matches the computed target within the tension checker's resolution, both written down. What wrong looks like: tension set by thumb and ear. Under-tension slips, glazes the sidewalls and throws belt dust; over-tension is silent and loads motor and blower bearings until one runs hot. Stop rule: a drive that will not hold the computed deflection anywhere in its range has the wrong belt length in it. Hazard in this step: fingers clear of the groove throat as you roll the base back, and torque the hold-down bolts before checking deflection, since a base that shifts under the checker reads nothing.

7. Reinstall the guard, close the cabinet, then re-energize and prove the drive under load. Belt guard back with every fastener, blower door latched, all hands out of the compartment before the disconnect closes. Acceptance: correct rotation, motor current at or below nameplate full load amps, no visible whip in the span, and the blower door interlock proven to stop the blower when the door is opened. What wrong looks like: a guard left off "until we see it run", which is how a hand ends up in a running drive. Stop rule: an interlock that does not stop the blower keeps the unit out of service, because that switch protects whoever opens this cabinet next. Hazard in this step: stand out of the drive plane and to the hinge side of the disconnect, call it before closing, and re-prove dead before a hand goes back in.

8. Re-tension after run-in before you leave the roof. New belts seat into the grooves and lose tension in the first minutes. Run the blower at least 30 minutes, open and prove dead again, re-measure and reset to the same computed target. Acceptance: a second deflection value logged against the run time it followed. What wrong looks like: leaving on the first tension and returning in three weeks to a glazed belt. Stop rule: a drive needing a third correction in one visit has a defect step 5 or step 3 missed, so go back to those steps rather than chase tension. Hazard in this step: sheaves and motor are hot and the drive coasts after shutdown, so the disconnect is open and proven dead again before the guard comes off, never reached through a guard opening.

The record this produces

One drive block on the ticket, filled on the roof: belt cross-section and length as read off the belt, sheave and bushing part numbers, motor horsepower, speed and nameplate full load amps, center distance and span, computed deflection target, first deflection reading, alignment offset in inches per foot with the tool used, motor current under load, interlock result, run time before the re-check, and the second deflection reading. A condemned sheave gets its own line with the part number and who authorized it.

That block turns a belt into an asset history. Next visit compares this offset and these amps against the last set, and a drive whose current has climbed at unchanged tension is telling you about bearings before the bearing does. Three belts in eighteen months at one unit is a geometry investigation, and these figures are the evidence that decides it.

Worked pass: 5-ton packaged unit on a strip retail bay, squeal on start

Ticket read "loud on startup, mornings only." Blower motor 1 hp, 208/230 V three-phase, nameplate full load amps 4.0.

Step 1 passed: disconnect open, locked and tagged, meter proved live on the curb receptacle circuit, zero across all three phases at the load lugs, meter proved live again. Step 2 recorded an A-section belt, motor sheave 3.4 in pitch diameter, blower sheave 8.0 in, center distance 24.0 in, variable-pitch flange two turns open. Step 3 gauged both sheaves: sidewalls straight, belt flush, both accepted.

Step 4 slacked the drive four turns on the base screw, belt off by hand. Step 5 FAILED. The straightedge touched at three points with a gap at the far face of the blower sheave, and a feeler read 3/16 in of offset. Center distance 24.0 in is 2.0 ft, so 0.1875 in over 2.0 ft is 0.094 in per foot against this drive's 0.0625 in per foot limit. Under the stop rule no belt goes on. The motor sheave bushing was loosened, the sheave moved in on the shaft, and the offset re-read at 1/32 in over the same 2.0 ft, which is 0.016 in per foot and inside the limit. Bolts torqued, straightedge re-laid, four-point contact confirmed.

Step 6: span between tangent points measured 23.5 in, so the target is 23.5 divided by 64, or 0.37 in. The checker reads in 1/32 in marks, so 12/32 (0.375 in) is the usable mark, reached at the published deflection force for an A-section belt on a 3.4 in sheave. Logged target 0.37 in, set 0.375 in.

Step 7: guard on with all four fasteners, door latched, disconnect closed from the hinge side after calling it. Rotation correct, current 3.6 A against nameplate 4.0 A, no whip. Door interlock opened and the blower stopped in under two seconds. Step 8: after 45 minutes the disconnect was opened and proved dead again and the deflection re-read at 14/32 in (0.44 in) against the 0.37 in target, the belt seating into freshly gauged grooves. Reset to 12/32 in and logged with the run time.

What the failure teaches: steps 1 through 4 passed and step 3 called both sheaves serviceable, so a tech running the fast version fits a new belt to a drive 50 percent past its alignment limit, tensions it perfectly, hears nothing because the belt is new, and books the callback for six weeks out. The morning squeal was a cold, stiff belt slipping on a misaligned drive, and every hour it ran was cutting the sheave the shop would otherwise not have bought.

References

• 29 CFR 1910.333(b)(2), work practices for electrical utilization equipment, which 29 CFR 1910.147 excludes at its own (a)(1)(ii)(C); 29 CFR 1926.417 is the construction counterpart for electrical work
• NFPA 70E-2021, 120.5, for the before-and-after instrument check that makes "proved dead" mean something
• 29 CFR 1910.134, the written respiratory program any respirator use in a lined blower compartment depends on
• The belt manufacturer's span-deflection tension table for the cross section, and the drive manufacturer's alignment tolerance
• See related: Light Commercial Rooftop Unit Service; Rooftop Work Fall Protection and Roof Access