Servo and Drive Retrofit

Motionwell Automation does servo retrofit and industrial drive replacement in Singapore inside our largest line of work this year, control system modernisation of existing production machines: replacing ageing PLCs, servo drives and VFDs across Allen-Bradley, Siemens, Omron, Mitsubishi, Beckhoff and Inovance platforms. On the Allen-Bradley side that means CompactLogix and ControlLogix controllers, Kinetix servo drives and PowerFlex 755 VFDs, and our Allen-Bradley spend for this work grew roughly fourfold in a single year. This page covers the motion half of it, meaning industrial drive replacement on an axis whose frame, guides, screw and gearbox are staying exactly where they are. What a servo axis is worth once it is in is visible on the machines we build new: the rotary capping heads on our GMP filling and sealing platform run Mitsubishi HG-KR servo motors with torque feedback, torque programmable from 0.5 to 5.0 Nm with the curve logged for every container. Machines are designed, assembled and tested at our Woodlands Link facility, and the company has delivered more than 150 special purpose machines since 2014 under ISO 9001:2015 and bizSAFE Level 3.

The short answer, before the detail. The mechanics of a well-built machine usually outlast its electronics, which is the whole reason this work exists: you are buying a second service life for a frame, a screw and a set of guides that are already paid for. A drive retrofit buys two things reliably, adjustability and data, and one thing conditionally, which is rate. It will not repair worn mechanics, and a stiffer position loop closed around a worn mechanism tends to expose the wear rather than hide it.

Where we stand, said plainly. We do not manufacture motors, drives, encoders, gearboxes or cable, and we are not a distributor for any of them. We buy them and re-engineer the axis around them on whichever platform your plant already runs, which is a question we ask before selecting drives rather than after. We are not a notified body and we do not issue CE certificates. Two boundaries decide which page you should be reading. If the controller rather than the motion hardware has run out of support, that is a different project, on our PLC migration and upgrade page. If the axis does not exist yet and has to be designed from a mass, a stroke and a move time, that is new-build work on our servo lift and precision positioning page.

What follows takes the decisions in the order they arrive: why the electronics run out first, how a new motor gets matched to an old mechanism, what absolute feedback changes, whether the axis needs a servo at all, why tuning is the hard part, what the drive gives you beyond motion, what a retrofit exposes, when not to do it, and what belongs in the survey and the acceptance plan. The motor-selection argument in general form is in our note on servo and stepper motor drives. If you have a motor list and a downtime window, skip ahead and talk to an engineer.

Why Do a Machine’s Mechanics Usually Outlive Its Electronics?

Because the two age against different clocks, and only one of them is under your control.

A frame, a set of linear guides, a ballscrew and a gearbox wear in proportion to use, in ways a maintenance schedule can measure and stay ahead of, so the mechanism degrades slowly and visibly. Electronics do not fail that way. A drive can be in perfect working order and still be unsupportable, because the part number is out of production, the programming software will not install on a current operating system, or the last engineer who knew the machine has retired. None of that shows up in a vibration reading.

That gap is what we see across the retrofit work: much of it has a compliance or supportability driver rather than a mechanical one, and the machine still runs fine while its control system cannot do something the plant now needs. The second driver is adjustability. A cam-driven axis is not adjustable, so a new product variant becomes a machining job, while a servo with an electronic cam profile turns it into a recipe change. That is usually the whole business case, ahead of any speed gain.

Read the last row as a cost item rather than a lament. On legacy machines the control system is often undocumented, and working out what the existing circuits do before replacing them is frequently the largest single item in a retrofit safety scope.

How Do You Match a New Motor and Drive to a Mechanism You Are Not Changing?

This is a part quotations get wrong, and the trap is that torque is the number that gets checked first. A modern motor rated for the same continuous torque as the one it replaces is often not the same physical object, because a frame sized for a given continuous torque today is commonly shorter and lighter than the one it replaces, which changes the shaft diameter, the keyway, the flange pilot, the bolt circle, the length behind the mount and the position of the connectors.

The mechanical fit is therefore a drawing exercise before it is a catalogue one, and it usually resolves into an adaptor plate, a coupling bored to the new shaft, and a check that the motor still fits the space the guard leaves. None of it is difficult, and all of it is expensive to discover during the shutdown.

Rotor inertia is worth pulling out of the list, because it is invisible on a nameplate comparison and it decides how the axis behaves afterwards. The ratio between the load inertia reflected to the motor and the rotor’s own inertia is what makes an axis tuneable, and a smaller, lighter rotor on an unchanged load raises that ratio. A gearbox is the lever that fixes it, since a ratio reduces reflected load inertia by the square of the ratio, which is what makes a large mass controllable by a small rotor. Changing that ratio is a decision about top speed at the same time, and the resolution is usually the motion profile rather than the ratio.

Two habits save most of this. Record the nameplate of every motor and drive before quoting anything, including the ones nobody plans to touch, because a machine rarely carries one generation of hardware. And treat existing cable as suspect rather than as an asset, since a cable that has flexed in a carrier for years is cheap to renew and expensive to diagnose. Where the panel is rebuilt around the new drives, name EN 60204-1 in the scope so the rebuild is not quoted as like-for-like rewiring.

What Actually Changes When You Gain Absolute Position Feedback?

More than the homing routine, which is the part everyone expects.

An incremental device counts from wherever it was switched on, so the machine must find a datum before it can be trusted, every time; an absolute device knows its position at power-up. On long gantry and track axes we specify servos with absolute encoders for exactly that: it removes the homing routine, holds position knowledge after a power loss, and supports dual-motor gantry synchronisation.

The operational change is in recovery rather than in start-up. A machine that has to home cannot resume after a stop mid-cycle until it has been cleared, which on a legacy machine means a person opening a guard and taking a part out of a fixture by hand. On a vertical axis it matters more again, because homing a suspended load means moving it while nobody yet knows where it is.

One consequence follows from the third and fourth rows. The reference that used to live in a sensor bracket now lives in a parameter file, so the parameter backup becomes the thing that lets the machine be repaired. Somebody has to own it, and it has to be re-made after every commissioning change rather than once at the end.

Is a VFD Enough, or Does the Axis Genuinely Need a Servo?

Worth asking per axis, because converting everything that spins into a servo is how a retrofit scope doubles without buying anything. Where an axis has no positioning requirement, an induction motor and a VFD is the cheaper right answer, and a PowerFlex-class drive covers a fixed-speed conveyor or pump.

The line is drawn by what the axis has to prove rather than by what it has to do. A VFD controls speed and, with a vector mode and a feedback device, torque. What it does not give you is a repeatable stop position, a profile that changes by recipe, coordination with another axis, or a position the controller can verify. When any of those four is in the requirement, the axis has become a servo axis whatever it used to be.

The row that saves money without buying anything is the last-but-one: an electric axis bought to replace a working cylinder buys tuning, cabling and a parameter backup for a move with two ends. The same argument on a filling machine, where a controls retrofit is often the cheapest accuracy improvement available on mechanically sound equipment, is on our drum and pail filling page.

Why Is Tuning the Hard Part on an Old Mechanism?

Because a servo axis cannot be tuned stiffer than the structure and transmission it is bolted to, and an old mechanism has more compliance and lost motion in it than the drive’s autotune expects to find.

The reason this bites on retrofits specifically is what the old control did. A cam defines motion geometrically, and a DC drive behind it only had to keep the speed roughly right, so backlash in a worn gearbox, a stretched belt, lead error in a screw and wind-up in a coupling were all tolerated because nothing closed a tight position loop around them. A modern drive does close that loop, at a bandwidth the mechanism has never been asked to accept, and the mechanism answers by ringing.

Four responses, in the order they are worth trying:

Lower the gains and accept the settling time. Free, and it tells you how much of the machine’s performance was hiding in its mechanics. Autotune gets most of the way; the last part is manual.

Approach every station from the same direction. A single-direction approach takes most of the backlash error out for the price of a little extra travel, and where an axis indexes between fixed stations it often costs nothing.

Filter the resonance in the drive. Notching out a narrow structural resonance raises the useful gain. It does not touch lost motion, and a filter used to mask a mechanical fault keeps working right up to the point the fault gets worse.

Fix the mechanism. Replace the worn gearbox, preload the rack, change the screw. This is a different budget, and it belongs in the quotation as its own line rather than as a contingency.

State the tuning result the way it will be used later. Gains, filters and profile limits have to be restorable after a drive is swapped at two in the morning, and they live in the same parameter file as the encoder offset.

What Do You Gain From New Drives That Is Not Motion?

A condition monitor, at no extra hardware cost, on every axis you convert. A servo drive computes torque and following error every control cycle because it needs them to work, and once the controller can read them, the machine is measuring itself on every move it makes.

We already build that way, so the value is not theoretical. On the GMP filling and sealing platform the capping torque curve is logged per container rather than sampled. On the 12-station rotary medical assembly machine, running a 15-second cycle, every press-fit station carries a strain-gauge load cell and records its full force-displacement curve against the part serial number, so a fault is attributed to the station that caused it rather than found at final inspection. How that generalises is on our inline dimensional measurement page.

On a retrofitted axis the same signal points at the old mechanics rather than at the product. The torque needed to make an identical move on an identical load measures what the mechanism is doing, and a trend that climbs over months is a guide, a screw or a belt telling you something quietly. Following error does the same job from the other side: a jam grows the error until the drive faults the machine with a code and a number, rather than in the silence an open-loop axis fails in.

Two things separate data from disk usage. Somebody has to look, so agree at scoping who reads the trend and what they do when it moves. And the alarm structure gets rebuilt during a retrofit anyway, which is the moment when tagging each alarm with a loss category costs almost nothing, so downtime arrives already sorted; the method is in our note on measuring OEE in production. Doing it two years later means touching a program that is by then in service.

One caution comes attached. A machine that ran for fifteen years with no network port suddenly has an EtherNet/IP switch, a data path to a plant server and an engineer who would like to dial in from home. Keep the controller, drives and I/O on their own segment with no route outward, put anything that talks upward on a second interface, and decide the remote access policy before commissioning, as set out in our guide to IEC 62443 for industrial control systems.

What Holds the Load When the New Drive Loses Power?

On a vertical axis, whatever is mechanical, because the new drive holds nothing at all once power is removed.

This matters more on a retrofit than on a new build, because the old machine may have held its vertical axis in a way the replacement does not reproduce. A self-locking worm, a low-lead screw or a brake integral to the old motor all disappear the moment somebody specifies a more efficient transmission or a motor without a brake option.

Read three figures off the new motor’s datasheet before assuming the brake covers it: static holding torque against the gravity load with the heaviest tool fitted, permitted braking energy and number of emergency stops, and engagement delay, because the gap between the drive dropping torque and the brake gripping has a distance attached to it. Whether that brake is a machine component or a rated safety function is an output of the risk assessment, and the full treatment of holding torque, counterbalance and power-loss testing is on the servo lift and positioning page.

One delivered reference for how the stop itself behaves. On our EV battery dismantling line the emergency stop implements Safe Torque Off as the primary stopping method, removing power from the robot servo drives within 10 ms so the machine stops under mechanical friction and gravity rather than under controlled deceleration. That is correct behaviour there. On an axis pointing up, the same sentence describes a descent, which is why the power-loss case gets tested rather than assumed.

What Will a Retrofit Not Fix, and What Will It Expose?

The findings below are the kind that turn into change orders, and they are cheaper to find with a dial indicator during the survey than with a commissioning engineer during the shutdown.

The last row is the one with a safety consequence attached, and it is why stopping performance has to be measured again rather than inherited: a new servo drive on old mechanics changes run-down time in ways no datasheet predicts, and the safety distances for scanners and light curtains are calculated from ISO 13855 approach speeds of 2,000 mm/s up to 500 mm and 1,600 mm/s beyond, plus stopping performance measured on the built machine. If run-down gets longer, guard locking may become necessary where a plain interlock was sufficient. That work, and the safety function register behind it, is on our machine safety and CE marking page.

Two obligations sit above all of this and belong in the kickoff meeting rather than in the closing report. The European Commission states that the Machinery Directive applies to products placed on the EU market for the first time, or where existing machinery is modified to such an extent that it becomes de facto new machinery. A modernisation that crosses that line puts the party doing it in the position of placing new machinery on the market, with the conformity work that follows, and where the line sits is a judgement made per project rather than a number. Separately, a control-system retrofit is commonly the action that triggers a re-assessment, and a design documented against the 2015 edition of ISO 13849-1 will need its performance level calculations restated when that happens.

In a regulated plant there is a third. The mechanical process usually did not change, so the performance qualification argument rests on comparability with historical data, which works only if the original validation package was retained. How the impact assessment and test scope get agreed with a quality unit is on our computer system validation page.

When Is a Servo and Drive Retrofit the Wrong Project?

This is the section that decides whether the rest of the page is worth trusting, and it names the cases we would rather decline.

The mechanism is at the end of its life. Where the guides are notched, the screws have measurable lead error and the frame has been repaired more than once, new electronics buy a control system for a machine that needs a mechanism. The retrofit will run, and every symptom in the table above will arrive at once.

The constraint is not the drive. Where the machine misses rate because of a manual load step, a dwell the process genuinely needs or an upstream starve, the axis was never the bottleneck, and faster drives on the wrong axis produce a machine that waits more precisely.

Nothing is asking the machine to change. The platform is still supported, spares are obtainable, no new variant is coming, no data or compliance requirement has appeared. Hold spares and spend the money on whatever is constraining the line. We would rather say that than quote it.

Only one axis is the problem and it has two ends. A clamp, a gate or an ejector is a cylinder with reed switches, and it does not improve by becoming electric.

A standard replacement covers the job for less. Where a proven machine or a packaged actuator does what the retrofit is being asked to do at a lower price, the useful answer is to say so, which costs you a conversation rather than a commitment.

Our own exclusions, stated so nobody spends a month finding out. We do not build production welding cells. We are not a notified body and we do not issue CE certificates, though we support the conformity work, including LVD and CE testing and Ministry of Manpower lifting certification where the machine includes lifting equipment.

What Belongs in the Survey and the Acceptance Plan?

A drive retrofit is priced from a survey rather than from a description, and what the survey has to establish is set out below.

On the machine. Nameplate data for every motor, drive and gearbox, including the ones not in scope. Feedback type per axis. Cable routes and what flexes. Panel space, incoming supply and enclosure heat load. The existing safety circuit and what it actually does, which is where the undocumented work lives. Whatever schematic, listing or manual still exists, and a note of what does not. And the machine’s real sequence, watched running, since the drawing and the machine have usually diverged.

On the plan. The downtime window, split into what can be prefabricated off the machine and what can only be done with it stopped. The fallback if the cutover overruns, decided before the panel is opened. And who owns the boundary between the new motion hardware and whatever is not being replaced.

On acceptance. Ask for these in writing before the work starts, because a test in the plan is cheap and one invented at handover is an argument.

• Repeatability at the tool point, with the approach direction and the load stated, not at a convenient mid-stroke position.
• Stopping performance measured on the retrofitted machine, with the safety distances re-confirmed against it.
• Power-loss behaviour on every vertical axis, at worst-case load, warm as well as cold.
• A torque and following-error baseline recorded at handover, so the condition trend has a reference point.
• A soak test long enough to turn the consumables into maintenance intervals.
• Parameter files, program archive and as-built schematics handed over as deliverables, with a named owner.

We do not publish a price or a lead time for this work, because the survey sets both. Design, build and testing happen at our Woodlands Link facility, which is the practical argument for a local builder on retrofit work: the adjustments that follow a cutover take hours when the people who did it are in the same industrial estate. Where the retrofit is part of tying old equipment into a new cell, that scope is on our machine tending automation page.

Frequently Asked Questions

Can we fit new servo motors to the existing gearboxes and mounts?

Sometimes directly, more often with an adaptor plate and a new coupling, and the check is mechanical before it is electrical. A modern motor of the same continuous torque is frequently a different physical part: shorter, lighter, with a different shaft diameter, keyway, flange pilot and bolt circle, and a lower rotor inertia. Rotor inertia is an item often skipped, because the ratio between the reflected load inertia and the rotor is what decides how tuneable the axis is, and a physically smaller motor on an unchanged load raises that ratio. Send the motor nameplates, the gearbox ratio and a drawing of the mount, and that question is answerable without a site visit.

Do we have to change the PLC to change the drives?

Not always, and the two decisions are worth keeping apart. Drives can be replaced under a controller that stays, provided the controller can still talk to them: the network the new drives speak, the spare capacity in the existing program, and whether the platform is still supported all decide it. The argument for doing both at once is that the panel is already open, the machine is already down, and the work of finding out what the old circuits do has already been paid for. The argument against is scope, because a controller migration carries its own program, its own testing and its own risk.

Will a servo retrofit make the machine faster?

Not on its own, and speed is rarely the business case. What a drive retrofit buys first is adjustability: a cam-driven axis is not adjustable, so a new product variant becomes a machining job, while a servo with an electronic cam profile turns it into a recipe change. What it buys second is data, because torque and following error become available to the controller on every move. Rate is a property of the mechanism, the process and the cycle structure, so on a machine already at the limit of its screws, guides and dwell times, new drives will show you that limit rather than move it.