Work log — Lift table project, worm gear drive diary
Engineer: E. Kowalski (mechanical designer)
Project: LFT-1400 service lift, 1,400 kg capacity, 1.1 m travel
Log period: Day 01 to Day 08
Document ref: WL-2026-0911-WG02
Day 01 — morning stand-up, the elevator that must never fall
Stand-up 08:15. The customer’s brief is short and nervous: a maintenance lift that carries a service technician and a pallet of spares up 1.1 metres, holds position for an hour while the man works underneath, and drops to zero if the power dies mid-raise. The key phrase in the spec sheet is self-locking drive, no brake add-on accepted as the primary safety. That sends me straight to the worm gear, because a properly calculated worm and wheel locks by geometry alone when the input stops, and that is the load-holding behaviour the job needs without a brake solenoid in the picture.
Morning I started the ratio and helix arithmetic. The drive pair in mind is a single-start worm, 4 modules, 2-start considered and rejected, because only single-start worms give the self-locking margin the customer demands. I ran the numbers first with the datasheet’s worst friction: for a steel worm on phosphor-bronze wheel, lubricated, the friction angle runs 4 to 6 degrees at 10 to 20 m/min sliding speed, and self-locking holds when the lead angle stays below the friction angle plus a margin. The draft design carries a lead angle of 4.2 degrees against a 5.5 degree friction angle, which braces the lock, and the margin is 1.3 degrees, thin enough that I flagged it as the day’s open point.
Evening note: the margin is the whole game. A worm that self-locks in the catalogue test on a bench may creep under vibration on a live platform when the oil warms and the friction drops, so I will spend Day 02 turning the 4.2 degree lead angle decision into a number I can defend to the safety reviewer.
Day 02 — the efficiency trap and the holding torque budget
Stand-up: none, I was in the drawing office the whole morning, which is a stand-up of a different kind. The open point from Day 01, the 1.3 degree friction margin, did not close itself overnight, so I pulled the drive efficiency and torque budget together to see the whole machine.
The lift moves 1,400 kg plus a 120 kg table and the column, so call it 1,550 kg acting through a 5 mm lead screw pitch… no, through the worm pair and a separate screw this lift uses. The design is a worm gearbox driving a trapezoidal screw, 40 mm diameter, 8 mm pitch, so the hold is a double stage: the screw holds its own load by friction and the worm holds its own end, and the two mechanisms share the responsibility in a way that makes the reviewer happy and my load case clear.
I ran the worm efficiency first, because self-locking and efficiency live on the same line. Single-start worm at 4.2 degree lead angle and 5.5 degree friction angle gives an efficiency of roughly 42 percent for the raising direction, and the reverse efficiency is negative, which is the mathematical way of saying it cannot backdrive, the load cannot push the worm around. The raising efficiency of 42 percent is the price of the lock, a two-start worm would lift 72 percent but would backdrive on the bench, and on this machine that trade buys the customer’s hour-long hold.
Afternoon I converted the hold to numbers on the check sheet. Screw stage: 8 mm trapezoidal pitch at 1,550 kg gives about 62 N-m of input torque at the screw nut in the raising sense, and the screw’s own lead angle of 3.6 degrees against its friction angle of 9 degrees braces the screw against backdrive on its own. Worm stage: the 62 N-m at the screw nut reflects back to the worm shaft divided by the 38:1 worm ratio and the 42 percent efficiency, which puts about 3.9 N-m on the motor shaft at rated load. The hold case is the figure that matters: with power off, the total system must hold 62 N-m at the nut with no input, and both stages resisting backdrive gives it a recorded margin of 1.9 over the no-backdrive threshold at cold start, dropping to 1.3 at warmed oil, the number from Day 01.
Evening thought: what nags me is the drop from 1.9 to 1.3 when the lubricant warms. A warm box is the honest operating state of this lift after forty minutes of duty, so the 1.3 margin is the real design margin, and 1.3 is respectable but I have seen a greasy, warm worm walk. Tomorrow I get the supplier’s actual friction-vs-temperature curve instead of the catalogue band.
Day 03 — the friction curve arrived, and it was worse than the catalogue
Supplier reply came at 09:40 with the friction-vs-temperature curve for the phosphor-bronze wheel on hardened 42CrMo worm at the sliding speeds this box sees, 12 to 18 m/min. The catalogue told me 5.5 degrees. The curve told me: at 30 C oil, 4.8 degrees; at 60 C, 4.3; and the no-load idle detector on the rig showed 4.1 degrees in steady warmed running. That drops my real margin from the Day 01 figure of 1.3 down to 1.15 at operating temperature, and 1.15 is a number I am not happy to sign.
The options, written on the whiteboard: (a) accept 1.15 with a brake as belt-and-braces, but the customer banned a brake as primary; (b) reduce the lead angle by changing the module or the pitch, which cleans the margin but forces a 46:1 ratio and a slower lift; (c) keep the geometry and fit a specially honed worm thread, which the supplier says raises friction by about 0.7 degree at the cost of a four-week lead time. I argued (b) into the afternoon with the safety reviewer on the phone, and the decision recorded at 16:30 was (b) with a compromise: drop to a 5 mm module, single-start, lead angle 3.4 degrees, giving a warmed margin of 1.65 against the no-backdrive line, at the price of raising time from 38 seconds to 46 seconds over the 1.1 m travel. The customer’s spec wanted 45 seconds max, so we gave away the last second.
Evening note: I should have asked for the curve on Day 01. The catalogue friction angle is a marketing number, the fitted curve is an engineering number, and I designed to the wrong one for a whole day.
Day 04 — the test rig and the no-backdrive fixture
The gearbox builder loaned a test box on the 5 mm module geometry, and I spent the day on the bench rig. The fixture holds the worm input locked with a torque arm and applies load to the wheel shaft through a hanging weight frame, opposite to the working sense, to prove the box will not drive backwards. Cold start: applied 70 N-m to the wheel, equivalent to 1.1 times the rated hold, and the wheel held, no rotation, torque arm resting on the stop, 28 degrees oil. Then I warmed the box to 60 C by running it unloaded for forty minutes and applied the same 70 N-m, held again. The rig won’t walk, and the reading confirmed the geometric margin rather than the oil-film luck.
Afternoon I added the dynamic case nobody had asked for: a reversal at half load with power on. Applied 35 N-m steady, induced a 6 Hz vibration with an eccentric mass to mimic the technician stepping on the platform, and the wheel held at its position within 0.02 mm of dial travel over the three-minute run. The self-locking holds through vibration, which was my Day 01 worry.
Day 05 to Day 07 — soft laps on the lift and the design record
Days 05 to 07 folded into one job because the bench and the machine ran in parallel. The lift ran hundreds of cycles with the 3.4 degree geometry at 1,400 kg load and the 46 second raising time came out at 45.2 seconds average, inside the customer’s 45 second line by a margin I called a gift from the supplier’s motor oversize. The hold measurement every morning: the platform, loaded, sat at its height for an hour with a dial gauge on the corner, maximum drift recorded 0.03 mm over the hour, and no measurable backdrive on any of the five recorded holds.
Evening of Day 07 I closed the design record with the decision chain so a future engineer reads the why, not just the what: single-start, 5 mm module, lead angle 3.4 degrees, hardening 42CrMo worm on phosphor-bronze wheel, 46:1 ratio, warmed friction 4.1 degrees from the fitted curve, no-backdrive margin 1.65, screw stage adding its own 9 degree friction hold on the 8 mm trapezoidal screw. The two-stage hold is the reason the customer’s one-hour requirement is met by geometry and not by electrics.
Day 08 — handover and the record table
Handover to production with the test rig results and the release drawing. The safety reviewer signed the no-backdrive test certificate after watching the 70 N-m cold and warm holds, and the maintenance manual now carries my bench procedure as the standard load-hold check. The log closes with the one sentence I most want future me to read: a worm gear that self-locks on the supplier’s test bench still has to prove it at operating temperature with the oil warm, and the margin belongs to the curve, not to the catalogue.
Attachment A — drive parameter record
| Item | Value |
|---|---|
| Worm starts / module | Single-start, 5 mm module |
| Lead angle | 3.4 degrees |
| Ratio | 46:1 |
| Materials | 42CrMo worm, phosphor-bronze wheel |
| Warmed friction angle | 4.1 degrees (fitted curve) |
| No-backdrive margin | 1.65 at 60 C oil |
| Hold test | 70 N-m cold and warm, no rotation |
| Raising time | 45.2 s average vs 45 s spec |
Attachment B — no-backdrive test check list
1. Lock the worm input with a torque arm against a stop. 2. Apply the rated hold torque to the wheel shaft. 3. Confirm zero rotation on the wheel dial. 4. Warm the box to 60 C by a forty minute idle run. 5. Repeat the hold and confirm zero rotation warm. 6. Record the oil temperature with every reading. 7. Repeat with a vibration mass if the duty includes stepping loads.
Glossary
Self-locking: a worm gear that cannot be driven backwards by the load when the input stops. Lead angle: the thread helix angle of the worm. Friction angle: the arctangent of the coefficient of friction between worm and wheel. Backdrive: reverse rotation of the input caused by the load. No-backdrive margin: the ratio of the friction angle to the lead angle at the self-locking boundary.
Week-end reflection — what the diary taught me about self-locking drives
Writing this in the machine office on Friday evening with the box sitting on the pallet behind me. Three lessons I will carry: first, the friction angle changes with the oil temperature more than any other single input, and the design must be frozen against the warm curve, not the cold one. Second, a single mechanical lock on a person-carrying machine feels thin no matter what the margin says, which is why this lift carries both a self-locking worm and a self-locking screw, and the two together turn a catalogue specification into a hold that survived a dial gauge and a 70 N-m hammering without a micron of drift. Third, the customer’s ban on a brake was not a gift to me, it forced the honest geometric answer, and the honest answer cost one second of raising time to buy an hour of holding. For a machine that spends most of its working day parked at height with a man underneath, that trade is the only one that makes sense.
