1. Start with your own process window, not the machine datasheet
Before you compare suppliers, write down your own limits. A coating run is defined by a small number of interacting variables: target material and power type (DC, mid-frequency AC, or RF), working gas and pressure, substrate web material and thickness, line speed, web tension, and the temperature budget of the substrate. Write down the film you need, the sheet resistance or optical spec it must reach, the width of the web, and the metres per minute you need to stay profitable.
Two numbers matter most and should be agreed in writing early. The first is the uniformity budget: how much the film thickness, and therefore the electrical or optical property, may vary across the coated width and along the length of the roll. The second is the steady-state line speed at your target thickness and sputter yield. Any evaluation that skips these numbers will end up comparing machine features rather than production capability.
2. Vacuum, pumping, and the coating chamber
Ask how the chamber reaches and holds its operating pressure, and what the base pressure is after a normal load-lock cycle. Pump type, pumping speed, and the gas load the chamber can tolerate decide how stable the process pressure stays as the web runs. A machine that drifts in pressure as web speed rises will drift in film properties, and no amount of operator tuning will fix that at production scale.
Then ask about the chamber interior. Rotatable versus fixed targets change the usable target life and the depth of the racetrack erosion, which feeds straight into your consumable cost per square metre. A closed-field magnet design and magnetic containment close to the substrate can improve ionisation at lower pressures. What counts is not the design name on the brochure but whether the vendor can show you coating results on a film close to yours.
3. Web handling, tension control, and through-put loss
In a coil coating tool, most unplanned downtime is mechanical. Check the unwind and rewind design: maximum roll diameter, core size, maximum tension, and whether tension is actively closed-loop controlled or simply set by a brake. Ask how splicing is done and how long a splice takes the line down. Roll-to-roll lines lose most of their availability to threading, splice failures, and web breaks; a supplier who cannot describe their threading procedure and break protection is hiding something.
Ask about the web path: guide roller count, dancer roll or tension transducer placement, edge-position control, and whether the web runs over coated or uncoated surface on its return pass. A mis-set edge-position controller will cost you width, and a contaminated return path will cost you adhesion and defect rate.
4. Power supplies, process control, and repeatability
Request details of the power generation: type of supply, control mode (constant power, constant voltage, constant current, or closed-loop on a process sensor), arc detection and suppression, and how fast the supply recovers after an arc. Arcs are part of sputtering; the question is whether the tool protects the film and the target or just stops the run. Confirm that recipes can be stored, recalled, and locked so that the same job runs the same way across shifts.
Also ask what is actually logged. Pressure, gas flow, power per station, web speed, tension, temperature, and run duration should be recorded per roll or per length, exportable in a plain format, and timestamped. Process logging is what lets you trace a customer complaint back to a single machine event instead of scrapping a batch.
5. Uniformity, thickness monitoring, and metrology
Ask how the vendor intends to prove uniformity, not how they claim it. Uniformity is usually shaped by target-to-substrate distance, gas distribution, and shutter or screen geometry, and it is tuned mechanically before it is tuned electronically. Request a witness-sample protocol: cross-direction and machine-direction profiles measured on a representative run, using a method you can also use in your own lab.
If the line carries in-situ monitoring, check what is actually closed-loop. Thickness or resistance sensors that only trend without correction move the tool from monitoring to control, and they decide how much scrap you accept when a target erodes or gas purity changes.
6. Utilities, footprint, and the operating envelope
Get the utility list in writing: installed electrical load, chilled-water flow and temperature, compressed-air quality and volume, exhaust and venting, and any gas handling or safety interlocks. Confirm the footprint, service access, and ceiling height, including door and airlock dimensions. A tool that cannot be moved into your building or reached with a maintenance cart becomes a project before it becomes a production line.
7. What the vendor must demonstrate, not describe
Ask for a coated trial on your own substrate, or on the closest substrate you can supply. Evaluate the sample for sheet resistance or optical performance, cross-direction uniformity, adhesion, and defect count. A serious supplier will run this trial and share the run data, because it is the fastest way for them to win the order.
Put the acceptance test into the contract before the order is placed: base pressure, uniformity across the coated width, steady-state line speed at the agreed thickness, defect limits on a defined inspection length, and the duration of the endurance run. A tool that meets those numbers on your floor, on your web, is a machine you can schedule against.
8. Cost per square metre beats purchase price
Model the running cost before the capital cost. Target utilisation, sputter yield on your material, power draw per coated square metre, consumable life for targets and shields, downtime for maintenance, and the number of operators decide what a coated square metre costs. Two tools with different purchase price can swap order in a three-year cost model. A vendor confident in their process will walk through this arithmetic with you; a vendor who only discusses specifications will not.
9. Integration, service, and long-term support
A roll-to-roll magnetron sputtering coating machine rarely runs alone. It sits with unwinding, pretreatment or plasma sources, in-line inspection, laminating or winding, and material handling. Ask which of these the vendor integrates themselves and which they leave to you, and how the machine talks to your line control. The automated production line solutions category on our site shows how single process tools become a scheduled line rather than an isolated machine.
Then ask what happens after commissioning: which spare parts are stocked, the response time for a service call, whether remote diagnostics are available, and whether operator training is included. Shenzhen PowerDe Automation Precision Equipment provides the engineering services that sit around such lines, from equipment customisation to manufacturing automation support.
10. A short evaluation timeline that works
Run the evaluation in four passes rather than one. First, a specification pass against your own process window. Second, a reference and trial pass on real samples with data you can reproduce. Third, a cost pass modelling the coated square metre over the tool's life. Fourth, a contract pass that converts every accepted number into a measurable acceptance test. Buyers who compress these into a single vendor visit usually end up negotiating on price, which is the wrong variable.
For a production decision, ask us for the trial protocol and the acceptance checklist we use on coating lines, and send your substrate and film target. We will start from the numbers your process needs, then show you the machine specification that reaches them. If your requirement extends beyond coating, review the thin-film solar equipment range and contact our engineering team directly.
