Tensioning a 2.4 m ball-lift belt: 7.5 mm of travel, and a tension you can count by hand

The top pulley tensioner for an open tennis-ball lift: where 7.5 mm of travel comes from, why 10.5 N, how to set it by counting swings, and a flight fix.

The open tennis-ball feeder got a ball lift this morning. A cleated timing belt carries balls from a sump at court level 2.36 m up into the hopper. The CAD for its top pulley said "slot the holes to tension the belt" and nothing else. This afternoon's work turns that into a mechanism, with numbers a builder can check.

How much travel, and where it comes from. The belt is 959 AT5 teeth welded endless: 4,795 mm. Whole teeth don't land exactly on the designed centre distance, so the belt sits 0.65 mm long before anything else. A welder's tolerance of ±2 mm on length (assumed; ask yours) moves the centre by ±1 mm. Tension stretches the steel-cord belt 0.92 mm, which adds 0.46 mm. To put the belt on, the head has to come down past the teeth, so it goes 5 mm below the shortest belt's free length. Added up, the head travels 7.5 mm on the spine: from 5.3 mm below its drawn position to 2.2 mm above.

That travel cost two other changes. The spine is now 5.3 mm shorter, so the lowered pulley's teeth stay 3 mm clear of it. The rod sockets in the head are 6.3 mm deeper, so the channel's rods stay seated.

How tight. The bottom pulley drives, so the down strand is the slack one at the motor. A 16-tooth pulley has only 8 teeth in mesh, and a slack strand there is how teeth jump. With every flight loaded, the lift pulls 10.5 N. Setting the static tension equal to that pull leaves 5.3 N in the slack strand and 15.8 N in the tight one. Each 2.4 m strand also hangs about 2 N of its own weight from the top pulley. The two jack screws holding the head up carry 28 N in all, which is small for a printed M5 nut pocket.

The mechanism is mostly parts already there. The head's four side bolts go into T-nuts in the spine's slots. A loose T-nut slides with the head, so the head needs plain holes, not slots. Two M5 jack screws turn in nuts captive in the head's walls, and their hex heads stand on the top rod clip, which moved down to sit under the head's lowest position. Turn them out to raise the head, tighten the side bolts, then lock each screw with a jam nut.

Setting it without a gauge. A strand of length L, mass m per metre and tension T swings at f = √(T/m) / 2L. For this belt that is about 2.4 swings a second: pluck the down strand at mid-height and count against a phone's stopwatch. Or push it sideways with 1 N (a 100 g weight over a pulley, or a luggage scale) and it should move about 56 mm. Both figures rest on an assumed 0.085 kg per metre, so weigh a metre of your belt and redo the sum.

A fixing that did not fit. The morning's spec fixed each flight with "two M3 through the belt between its cords". Checking it against the belt's profile, it doesn't work. A screw through a timing belt puts its head on the tooth side, which runs over the pulleys' teeth and the spine's sliding strip. An M3 countersunk head is 6 mm across, and an AT5 tooth is about 2.5 mm, so there is nowhere for the head to go. The bill of materials now asks for PU cleats welded to the back by the belt supplier, the usual way cleated belts are made. A printed finger cap that fits over a welded cleat is the next CAD job.

The tests move the head and pulley to both ends of the travel. At each end they check the head clears the spine, the clip and the rods, the rods stay over 10 mm deep in their sockets, the side bolts' T-nuts stay on the spine, and the head never lands on the jack screws. Every figure about the belt is still an assumption until somebody builds one. The full section is "Top pulley tensioner" in ventures/open-tennis-robots/cad/README.md.

All writing · More in Tools and projects · More on Open Tennis Robots · All tools and research