Retrofit or replace? How to decide what to do with an ageing machine

The machine still makes parts, but the drive is obsolete, the control has no spares, the operator panel is held together with tape and nobody on site can modify the program. The retrofit quote is a fifth of a new machine. It looks like an easy decision, and often it is — but it is easy in both directions, and the expensive mistakes come from deciding on price alone.

The question is not "which is cheaper" but "what am I actually buying, and for how many more years?" Here is how we assess it before recommending either.

Test 1 — Is the mechanical base still sound?

This is the test that settles most cases, and it is the one most often skipped, because it is the only one you cannot answer from a datasheet.

A retrofit replaces controls, drives, sensors and safety systems. It does not restore worn guideways, a ground-out spindle bore, a cracked casting, lost geometric accuracy or a structure that was never rigid enough for the tolerance you now need. If the mechanical base is sound, new controls can make the machine behave like a much newer one. If it is not, you have spent real money on a machine that still cannot hold the tolerance — and you have made it harder to justify replacement next year.

Judge it on evidence, not appearance: geometric checks against the original acceptance test, measured backlash and positioning repeatability, spindle run-out and vibration, the condition of ways and screws, structural inspection. A day of measurement before the decision is the cheapest money in the whole project.

Test 2 — Does the machine still match the product?

Machines get replaced because the product moved, not because the machine failed. Check the current and next-three-year requirement against what the machine can do:

• Cycle time against the takt you now need. A machine that was comfortable at 600 units a day may be the bottleneck at 900, and no retrofit changes its basic kinematics.
• Part envelope, weight and variant range, including the variants in development.
• Tolerance and capability. Will it hold the Cpk your customer now asks for, or the one they will ask for at the next PPAP?
• Changeover time, if your batch sizes have fallen.

Where the answer is "no" on capability or cycle time, a retrofit is a deferral, not a solution.

Test 3 — What is the machine actually costing you now?

Put a number on the status quo before comparing options. Over the last twelve months:

• Unplanned downtime hours, and the output lost — valued properly if the machine is a bottleneck, at near zero if it is not.
• Maintenance spend, including the premium paid for obsolete parts and the cost of holding spares nobody else stocks.
• Scrap and rework attributable to the machine.
• Energy, if it runs old drives and a fixed-speed hydraulic pack.
• Quality escapes and the containment they triggered.

Many machines that "must be replaced" turn out to cost very little when this is totalled, and many that look fine turn out to be consuming a maintenance engineer's week every month. The number decides how much either option can justify.

Test 4 — What is the obsolescence and spares risk?

An obsolete control is not an emergency by itself. The real question is what happens on the day it fails:

• Can you still buy the CPU, drives, servo motors and HMI, new or reliably refurbished? At what lead time?
• Do you hold critical spares, and has anybody tested that the held spare actually works?
• Do you have the program, the source, the parameters and the backups — and has a restore been tested on the actual machine?
• Is there anyone within reach who can service it, or one retired engineer who knows it?

Score the consequence: a machine that is down for eight weeks waiting for a refurbished drive from an overseas broker is a production risk long before it fails. That risk, not the age on the nameplate, is what justifies a control retrofit.

Test 5 — Safety and compliance

Older machines commonly have emergency stops wired through the standard PLC rather than a safety circuit, no monitored stop, no energy isolation and lock-off provision, and guarding added after the fact. If you are modifying the machine anyway, bringing the safety system up to current practice is far cheaper as part of that work than as a separate project later — and it is usually what a customer audit asks about first. Our guarding guide covers what this involves.

What a retrofit normally includes

Indicative cost and time

Planning bands only; the machine decides the number.

The time comparison is often more decisive than the cost. A retrofit is typically weeks of planned downtime; a new special purpose machine is commonly 4–8 months from order to production, and a standard machine can be longer if it is imported. If you need capacity this quarter, that gap may settle the decision by itself.

The risks buyers underestimate

• Undocumented logic. Decades of undocumented changes live in the old program. Reverse-engineering it is the largest and most variable cost in most control retrofits — and the commonest reason a retrofit overruns.
• Interface surprises. Non-standard feedback devices, proprietary buses and sensors nobody makes an equivalent of. Found during commissioning if the survey was superficial.
• Scope creep during downtime. Once the machine is stripped, the mechanical condition is visible and additional work appears. Budget a contingency rather than pretending it will not.
• Losing the only person who knows it. If one engineer holds the knowledge, the retrofit is also the opportunity — and the deadline — to document it properly.
• Retrofitting a machine you will replace anyway. If the product is changing in two years, spend the minimum to keep it safe and running, and put the capital into the replacement.

What to ask a retrofit supplier before you order

The quality of the survey predicts the quality of the retrofit. Before accepting a quote, ask for answers in writing:

• What did you measure on the machine, and what were the readings? A quote produced without a site visit is a price, not an assessment.
• How will the existing program logic be captured, and what happens to the estimate if it turns out to be undocumented?
• Which makes and models of control, drives and safety devices, and what is their stated support horizon? A retrofit onto a range already near end-of-life buys you very few years.
• What exactly is the acceptance test? Name the cycle time, the accuracy, the capability study and the conditions under which they will be demonstrated, before and after.
• What is the downtime plan — what is prepared off the machine, what must happen with the machine stopped, and what is the fallback if something is found during strip-down?
• What documentation is handed over: electrical drawings as built, program source, parameter backups, spares list with part numbers, and operator and maintenance training?
• Who supports it afterwards, in what response time, and can your own team modify the program without calling them?

The last two decide whether you have solved the obsolescence problem or simply moved it onto a newer platform with a new single point of dependence.

A simple decision summary

We survey existing machines and lines, measure what is actually worn, and quote both paths honestly — including the cases where our recommendation is to spend less than you expected. Send us photos, the nameplate and the electrical drawings and we will tell you which side of this decision your machine falls on.