When buyers compare duct machines, they compare the mechanicals: roll tooling, plasma table, flange stations. Fair enough, that is the visible part. But after a decade of commissioning lines in hot, dusty, electrically noisy corners of the world, I have learned that more downtime traces back to the electrical cabinet and its variable frequency drives than to any mechanical component. A machine that cuts beautifully and trips its drives twice a day produces less than a plain machine that runs. So before you sign, here is what to look at and what to ask for.
Every driven section on a line, the decoiler, the leveler feed, the servo feeder, the takeoff conveyor, has a drive sized to its motor. The lazy approach sizes the VFD exactly to motor nameplate current. The correct approach adds headroom, because real loads are not nameplate loads. A decoiler accelerating a full 3-ton coil against inertia can pull 150 percent of rated torque momentarily. Specify drives rated for at least 110 percent overload for 60 seconds and 150 percent for 3 seconds, and insist the supplier shows the calculation, not just the drive model number.
Second question: input protection and harmonics. In plants with weak grids or lots of other VFDs running, drive input distortion and voltage spikes get ugly. Standard practice on a decent line is input chokes or DC chokes on drives above about 5.5 kW, and phase-loss protection that actually trips instead of limping. I have seen a plant in South America run a phase-loss condition for weeks because the drive kept running on two phases with derated output, cooking a motor slowly. The drive knew, nobody had wired the fault relay to anything.
Third: braking. Decoilers under regenerative load, when the strip pulls the coil faster than the drive commands, push energy back into the DC bus. Small regenerative loads ride through on the bus overvoltage limit, but a heavy coil with a long deceleration needs either a braking resistor or a long decel ramp. Ask what decel time the commissioning crew set and whether a braking resistor is fitted. On one line I commissioned, dropping the decel from 5 seconds to 15 seconds and adding a resistor took the overvoltage trip rate from several times a shift to zero.
You do not need to be a drive engineer, but you should know the five parameters that matter and write down their values at handover. Acceleration and deceleration times, usually 10 to 20 seconds on a decoiler with full coil inertia. Minimum and maximum frequency limits, so nobody can accidentally run the line shaft at a speed the mechanicals were never balanced for. Skip frequencies, if a resonance around a specific speed makes the frame sing or the strip flutter, the drive should jump over that band. Motor thermal protection, set to the actual motor service factor. And the fault relay configuration, wired to something visible, a stack light or an HMI banner, not left floating in the cabinet.
Write these into the handover document. When a drive fails two years later and gets replaced by a local distributor, the first thing the new drive needs is this parameter set. Facilities that kept the commissioning parameter sheet recover in an hour. Facilities that did not lose days rediscovering settings by trial and error.
A duct fabrication shop is a hostile electrical environment, and it is worth saying why plainly. The dust is conductive. Steel swarf, graphite from plasma cutting, and zinc particles from galvanized coil all drift through the air, and all of them conduct or corrode. A cabinet that would live happily in a food factory dies in six months next to a plasma table.
The choices that matter:
| Design choice | Fan-filter enclosure | Air-conditioned enclosure | IP65 sealed, passive |
|---|---|---|---|
| Dust ingress | Controlled, filters clog in dusty shops | Positive pressure, dust kept out | Best, no moving air path |
| Heat handling | Limited, derate above 35 C ambient | Full, 40-45 C ambient fine | Poor for high kW drives |
| Maintenance load | Weekly filter check in dusty shops | Low, quarterly service | Lowest |
| Cost | Low | Medium | High, size limits |
| Best fit | Small lines, clean-ish shops | Hot climates, plasma-heavy lines | Small remote stations, sensors boxes |
For most export installations in the Middle East, Africa, or South Asia, I recommend the air-conditioned main cabinet. The compressor adds cost and one more service item, but it also keeps the cabinet pressurized, which keeps conductive dust out, and it lets the drives run at full rating in 45-degree summer heat. A fan-filter cabinet in a Gulf workshop at 48 degrees ambient with a plasma machine downwind is a thermal trip waiting for an afternoon. If your supplier offers it, take it. If not, at minimum position the cabinet away from the plasma table and out of direct sun, ideally in its own shaded corner with clean intake air.
Inside the cabinet, look for the quiet markers of quality. Power and control wiring separated, control bundles twisted and routed away from drive output cables. A real grounding bus, generously sized, with every ground landed on it rather than daisy-chained to a door hinge. EMC cable glands on shielded cables, actually grounded at the gland, not just shielded cable stuffed through a plain hole. And spare space, at least 20 percent, because you will add sensors, lights, and a remote monitoring box someday, and a cramped cabinet means drilling into a live panel.
Beyond the drives, a modern line cabinet should carry a proper safety architecture. Emergency stops wired as a safety circuit, not just into the PLC, with a safety relay that drops the drive enable outputs. Door interlocks on the cabinet and on any guarded section. Light curtains or safety mats around the plasma cutting area, because that is where hands wander. And 24-volt control power fed from a protected supply with fusing per branch, so one shorted proximity sensor does not black out the whole control system.
Grounding deserves its own paragraph, because it is the number one cause of "random" faults in overseas installations. VFD output is high-frequency switching, and if the motor cables and machine frame are not well grounded, that switching noise radiates into every encoder and sensor line on the machine. Symptoms: position errors on the servo feeder that appear only when the plasma arc starts, spindle speed wobble, false sensor counts. The fix is boring and cheap: a proper earth electrode system under 4 ohms, bonding of every frame section, shielded motor cables grounded at both ends. A plant in Nigeria chased a phantom servo fault for three months. It was a paint-covered grounding strap at one frame joint. Four ohms of paint.
For the price of a 4G router and an afternoon of configuration, your line cabinet can talk to the factory. Most modern PLCs support Modbus TCP or similar, and any decent supplier can expose a read-only view of drive status, fault history, and cycle counts. When something trips, the supplier's engineer can see the fault log before anyone uploads a photo of the HMI screen. On the auto duct production line 5 configuration, this matters most around the plasma and servo subsystems, where fault codes carry real diagnostic content. Remote access turns a week of email ping-pong into a twenty-minute session.
Security note from experience: put the remote access behind a VPN or at least change every default password, and tell the supplier you want read-only access. I have seen factory networks where the machine HMI shared a password with the office printer.
Turn this article into a short annex to your next machine contract. Drives sized with stated overload ratings, calculated per motor, not catalog-guessed. Input chokes on drives above 5.5 kW. Air-conditioned main cabinet for ambient up to 45 degrees, IP54 minimum for remote terminal boxes. Safety relay based e-stop circuit and light curtain at the plasma station. English electrical drawings as-built, drive parameter sheets recorded at handover, and PLC program backup handed over on media you keep. Remote monitoring port configured. Terminal re-torque and thermographic check included in commissioning scope.
A supplier who accepts that list without arguing is a supplier whose cabinets you will not have to open on Sundays. A supplier who pushes back on each line item is telling you where their cost savings live. Either way you learn something useful before the money moves. And when the line is five years old and a drive module finally fails on a Friday afternoon, you will care about one thing: whether the spare is in your store, the parameter sheet is in your drawer, and the drawings actually match the cabinet. The power slitting machine and every other single-machine buyer in your group should demand the same annex, by the way. It works for the whole machine family, from a standalone spiral tubeformer to a full line.
You can, and in some markets it shortens the spare chain. But mixing brands means the supplier cannot support parameter sets, and warranty claims get murky. A workable middle path: take the supplier's brand as fitted, then buy one spare drive of the same model from a local distributor at commissioning time. You get warranty support from the factory and same-day hardware availability from town.
Weekly checks during the first month, then set the interval by how dirty they come out. In a heavy plasma-grinding environment, plan on monthly changes in dry season and fortnightly when the workshop does more sanding. A clogged filter raises internal temperature slowly, and the first symptom is usually a drive thermal trip on the hottest afternoon, the worst possible time to be troubleshooting.
Yes. Plasma cutting on an open table with a moving gantry is the one area of a duct line where an operator can lose fingers. Insurance requirements aside, every serious HSE audit flags it. The curtain costs a fraction of one accident and does not slow production if it is positioned properly.
As a rule, drives tolerate minus 10 to plus 10 percent continuously. If your plant sees dips beyond that at peak hours, fit a stabilizer or servo voltage corrector sized at 120 percent of line load. Measure before you buy: log the voltage at the machine feeder for a full week including weekends, because peak-hour sag is invisible on a morning visit.