Each machine hangs on a few people: HR for special purpose machine builders

Three weeks before the customer's buy-off, the concept designer of a leak-test and marking station hands in his notice. The machine is half wired. Nobody else knows why the clamp cylinder was sized one step up, why the nest has a relief on one side, or which cycle-time trade-off the customer accepted on a phone call. The fitters can build to the drawing. The drawing does not explain itself.

SPM builders feel this harder than any other trade on the industrial machinery hub. No two machines repeat. The page on special purpose machine builders names it as the first pressure point: each machine hangs on a few people.

• The concept designer knows the customer's real problem, the cycle time promised, and why the gripper, cylinder bore and fixture datums were chosen. Very little of that reaches the 3D model.
• The senior fitter knows the fixes made on the floor: the slotted bracket, the re-drilled dowel, the shim pack under the press-fit unit, the guard that had to be cut back for a sensor.
• The controls engineer knows the ladder logic, the interlock sequence, the HMI recipes and the dozen small changes made during trial that were never saved with a note.
• The commissioning engineer knows the customer's floor, the supply voltage quirks, the operator who keeps bypassing the light curtain and the maintenance head who signs the acceptance.

Each of these is genuine expertise. The trouble is that in most SPM shops it sits with one person per machine, and the machines overlap. A small team can carry four live projects, a rivet cell, a screwing station, a bearing press and a gauging fixture, with the same designer, fitter and controls engineer named on all four.

When the designer leaves, change requests slow down because nobody wants to alter a bracket or an actuator stroke whose purpose they cannot see. When the senior fitter leaves, the run-off reveals misfits he used to correct by hand without comment. When the controls engineer leaves, the program in the controller may be newer than any backup in the drawing office. Each case pushes out the customer buy-off. In this trade, buy-off usually triggers a payment milestone, so a resignation becomes a cash problem within weeks.

A quieter loss follows. The two who remain soak up the gap with nights and Sundays, rewiring solenoid valves and re-teaching robot points. The risk of a second resignation rises just when the shop can least afford it.

• Writing things down feels pointless for a rivet cell or leak tester that will never be built again.
• The team is small, and the owner often started as the designer, so knowledge in heads feels normal.
• Every project runs late, so the time for notes is the first thing cut.
• Pairing a junior on a live machine looks like paying two people for one job.
• Controls engineers are scarce, so there is rarely a second one to pair with.

1. 1

   Draw up a dependency sheet for every live machine

   One line per machine: designer, lead fitter, controls engineer, commissioning engineer, and an understudy for each. A blank understudy box is the risk list.

2. 2

   Name an understudy in each role

   A junior designer who reviews the model weekly, an assembler who builds beside the senior fitter, a technician who keeps the PLC backup current. Matching the senior is not the aim. The understudy needs enough to keep the machine moving for a fortnight.

3. 3

   Write a one-page why-sheet at design release

   The customer's real problem, the cycle time promised, the three or four choices that would puzzle someone else, and any trade-off agreed with the customer. Half an hour per machine.

4. 4

   Mark up drawings during assembly

   The fitter marks every hand fix on a print kept at the machine. The designer picks them up weekly, so the next machine with a similar module starts from the corrected version.

5. 5

   Save the program with a note after every trial change

   Version, date, what changed and why. A program on the machine that is newer than the office copy is the most common loss when a controls person leaves.

6. 6

   Hold a lessons review when the machine ships

   Designer, fitter and controls engineer, one hour, a short note filed against the customer's name. The SPM learning and development page covers how this grows into a training plan.

Machine and customer; buy-off date; payment milestone linked to buy-off.
Concept designer; understudy who reviews the model weekly.
Lead fitter; understudy building beside him.
Controls engineer; understudy holding the program backup.
Commissioning engineer; understudy briefed on the site.
Why-sheet filed at design release: yes or no.
Marked-up prints collected this week: yes or no.
Latest program version saved with change note: date.
Lessons review held after dispatch: date and filed location.

Spreading know-how only works if people stay long enough to absorb it. Fitters in SPM shops often learn by years of watching with no grade, so give them one, tied to the modules they can build unaided. Controls engineers field offers from larger automation firms, so show them a ladder and a say in project planning. Because good controls people are rarely free when you need them, start searching before a project calls for one. The recruitment page for SPM builders sets out how. The longer view of who needs cover first is succession planning, and the SPM workforce planning page shows how trial-stage peaks are planned.