Work log — Pallet stacker project, cylinder cushioning week
Engineer: A. Benetti (pneumatics commissioning)
Project: PSR-900 palletising suction stacker, 900 mm lift arm
Log period: Day 01 to Day 06
Document ref: WL-2026-0911-CS03
Day 01 — morning stand-up, the arm that slams
Stand-up 08:20. The call that brought me in was short and it was a noise complaint with numbers behind it: the lift cylinder on the PSR-900 hits its extend end with an audible thump at 18 pallets a minute, and the last maintenance report logged an accelerometer peak of 14 g at the arm tip during the stop. The product damage claim came in the same week, dented carton edges on the two top layers, dead on the line between 90 and 110 per shift. The cylinder is a 63 mm bore, 500 mm stroke, with adjustable cushioning, and my first job is to find out whether the thump is the cushion not tuned or the cushion not able.
Morning on the machine with the acoustic pick-up: the thump broadband peak sits right at end of stroke, and the cylinder speed into the cushion is the first suspect. I videotaped the extend motion frame-stepped, the free stroke runs at 260 mm/s and the cushioning engages 32 mm from the end, which is on the knife edge of the datasheet’s 25 to 35 mm cushion entry. The 14 g figure, though, is not a normal cushion arrest, so my suspicion is the cushion needle is open too far or the cushioning seal is bypassing, which can happen on a cylinder that has been slammed for months.
Evening note: I will start tomorrow by measuring the flow needle setting and then closing it down in three steps instead of flipping it to full cushion and being sorry. There is almost nothing worse on a pneumatic line than a shock that has become normal.
Day 02 — the needle, the seal, and the discovery under the cap
Morning, cushion needle inspection first. The datasheet box position for this cylinder is 2.5 turns open from the closed seat, and the found setting was 4 turns open, which explains a lot. The needle screw on the extend cushion was nearly backed out, someone had opened it fully at some earlier session and the mark on the cover said nobody closed it after. I set it to 2.5 turns as a starting point and ran one low-speed trial, 120 mm/s, and the thump dropped from the loud knock to a soft kiss, but the stop time rose and the arm overshoot increased 6 mm. So the cushion absorbs, but the needle alone is not the whole story.
Then the seal check, because a bypassing cushion seal would eat all the needle’s work. I pulled the front end cap of the 63 mm cylinder, cushion spear and seal out, and found the anti-extrusion ring chewed on one side and the cushion seal lip folded, classic damage from the months of hammering at 4 turns open. The parts kit was in stock, and the swap took forty minutes with the arm blocked. New seal, new anti-extrusion ring, thin film of grease on the spear, reassembled, and the cushion hold test came back clean, no air hiss past the seal at the crush point.
Afternoon with needle at 2.5 turns, cushion seal new: the 18 pallets per minute run gave 6.2 g at the arm tip, down from 14 g, and the dent count in a two-hour run fell from 90 to 12 per shift. The line is already better, but the product engineer wants under 4 g and I want the velocity curve flat through the cushion, so Day 03 is about the approach speed.
Evening thought: the 14 g was doing genuine damage and it was hiding in plain sight because the thump was the line’s normal rhythm. The needle had been opened, the seal had been damaged by the needle being open, and nobody re-closed either. When in doubt, the first thing to do on a cushion complaint is read the needle setting and check the seal, before touching the flow control valve on the supply.
Day 03 — the approach speed budget and the two-valve fix
Stand-up: I bagged the problem to the fluid team, the approach speed was the last fat in the cushion. The cushion can only do its work if the piston arrives at the cushion entry slowly enough to let the trapped air build the arrest, but the line used one supply flow control for both the fast approach and the cushion phase, which is the classic design mistake on a fast pick-and-place cylinder. I laid out the two-valve scheme on the pneumatic sketch: a main flow control sized for the 300 mm/s approach, plus a separate small needle that throttles only the final 32 mm of cushioning travel through the cylinder’s integral cushion valve. The arm reaches the end slowly because the cushion flow is limited, regardless of the approach speed.
The numbers argue for it. At 260 mm/s free stroke, the piston carries the 40 kg arm and 18 kg of suction head, kinetic energy at cushion entry of about 3.1 J, and the cushion has to eat that within 32 mm, which at the old 4-turn setting it simply did not, the needle dumped the air straight out and the piston hit metal. With the needle at 2.5 turns the cushion traps air, the arrest energy is absorbed over the cushion length, and the stop becomes controlled. I fit the remote needle into the cushion exhaust line and gave the whole circuit a detent so the two valves stay independent.
Afternoon trial confirmed the mechanism: arm tip acceleration dropped to 3.4 g at 18 pallets per minute, under the product engineer’s 4 g target, and the stop time went to 0.18 seconds with the overshoot back to the 3 mm the mechanism was drawn for. Two-hour run, dent count dropped to 3 per shift, all from cushioning tuning, no mechanical change.
Day 04 to Day 05 — full shift runs and the hot-cold consistency
Days 04 and 05 ran the machine at the production rhythm with the two-valve circuit. First full shift, 10,000 pallets, arm tip peak held between 3.2 and 3.6 g all day, dents closed to 2 and 3 per shift, and the cushion needle did not drift, the set screw I applied after the trial held. Second shift at the warmer afternoon temperature, supply at 6.2 bar, the cushion phase was 4 percent slower but still inside the cycle time, and the 3.4 g figure reproduced. The temperature effect on a tuned cushion is small when the needle is closed enough to make the air do the work, which is the argument for not running it wide open.
One thing I logged with a wry note: the 63 mm cylinder is a perfectly ordinary catalogue cylinder, nothing special, the whole fix was reading the needle, replacing a 30-euro seal kit, and adding a 70-euro needle valve. The catastrophic noise was 100 euros from sorted, and the months of damage were the price of nobody turning a screwdriver.
Day 06 — handover and the procedure that prevents the repeat
Handover day. I wrote the cushioning set-up procedure into the maintenance file, with the needle baseline (2.5 turns from seat), the seal inspection on the PM schedule (every 40,000 cycles or when the thump returns), and the two-valve sketch so nobody is ever tempted to use the supply flow control for the cushion. The maintenance plan now checks cushion needle turns as a monthly item, on a row right next to the filter drain. Last run of the day with the customer’s quality rep watching: three pallets, no dent, arm tip at 3.4 g, and the rep signed the release.
Closing the log with the numbers the job owed: 63 mm bore, 500 mm stroke cylinder, needle from 4 turns down to 2.5, new cushion seal and anti-extrusion ring, two-valve cushion circuit with a 3 mm overshoot, arm tip from 14 g to 3.4 g, and dents per shift from 90 plus down to 2 to 3. The product damage claim is dead, and the thump is gone from the plant floor.
Attachment A — cushioning parameter record
| Item | Before | After |
|---|---|---|
| Cushion needle | 4 turns open | 2.5 turns from seat |
| Cushion seal | folded, bypassing | new, clean hold |
| Approach speed | 260 mm/s | 260 mm/s (cushion independent) |
| Arm tip peak | 14 g | 3.4 g |
| Overshoot | +6 mm | 3 mm |
| Dents per shift | 90 to 110 | 2 to 3 |
Attachment B — cushioning set-up check list
1. Read and record the needle setting before any adjustment. 2. Check the cushion seal and anti-extrusion ring for bypass. 3. Set the needle to the datasheet baseline, 2.5 turns for this cylinder. 4. Run one low-speed trial and measure the arm tip shock. 5. Fit a separate needle for the cushion phase if the approach valve is shared. 6. Measure the stop time and overshoot, set overshoot against the drawn value. 7. Lock the needle screw and log the setting.
Glossary
Cushioning: the internal arrest of a cylinder piston near end of stroke by trapping exhaust air. Cushion needle: the adjustable restrictor that controls the trapped-air bleed rate. Float: the uncontrolled contact of piston and end cap when the cushion is bypassed. Approach speed: the piston velocity before the cushion engages. Overshoot: the distance the load travels past the cushion entry while stopping.
Week-end reflection — the log as a warning note
Friday evening, and this diary has turned into the warning I wish had been written before the first dent. Three things worth keeping. First, a cushion complaint is rarely a cushion problem in one part; this one was a needle setting, a damaged seal, and a shared flow valve, three separate faults stacked in one thump, and fixing any one alone would have left the noise partly alive. Second, the numbers matter more than the noise, the product engineer’s 4 g target gave me a stopping point, without the accelerometer the thump would have been called cosmetic and the dents explained away as packaging. Third, the cost of the fix, a seal kit and a needle valve, is small next to the damage and the claim paperwork, which is a sentence I want every colleague to read before they call the thump normal.
The last entry for this project goes to the PM sheet, not to the log: cushion needle turns, once a month, next to the filter drain, because the needle that drifted to 4 turns took a year to drift, and a monthly glance would have caught it in week one. The machine ends the week at 3.4 g, the line ends the week with dents in the single digits, and my screwdriver finally found the job it was supposed to do months ago.
A final arithmetic note for the file: at 18 pallets per minute the cushion cycle is about 0.55 seconds per extension including the hold, and the cushion phase occupies roughly 32 mm of the 500 mm stroke, so the arrest work happens in about 6 percent of the stroke length. That is the whole reason a needle setting of half a turn makes the difference between a controlled 3.4 g stop and a violent 14 g metal contact: with 94 percent of the stroke left to approach at speed, the cushion has a short runway to do all the slowing, and every turn of the needle either gives it that runway or takes it away.
