PLC Programming Outsourcing: 6 Contract Items and 5 Cost Layers

Mitsubishi Electric closes final orders for the MELSEC-Q Universal Model QCPU on 30 September 2026. From the publication date of this article — 30 July 2026 — that is roughly two months away. Repair support continues until 31 October 2033, and that long tail is exactly what makes plants relax. It should not. In an equipment upgrade, the thing that actually consumes calendar time is not procuring hardware. It is migrating a program that nobody in the building can read any more.

This article is written for the people who place those orders from an ASEAN plant: expatriate plant managers, local engineering managers, and regional engineering leads covering Thailand, Vietnam and the wider region. The framing is deliberately not “here is how ladder logic works.” It is: when you outsource PLC programming, what do you decide, who do you give it to, and — the question that determines everything else — what do you keep in-house?

The 30 September 2026 Deadline: Why This Matters Now

Mitsubishi Electric announced the discontinuation of the MELSEC-Q Universal Model QCPU in Technical News FA-D-0418-A. The notice covers 14 model numbers, from Q03UDCPU through Q26UDEHCPU. The stated reason is the one given in the notice itself: some of the components used in these products have become difficult to obtain.

Repair support runs for seven years after production ends. Read the table quickly and it looks like there is plenty of room. That is precisely why, in most plants, this notice gets filed under “we will deal with it later.”

Why “Repair Support Until 2033” Is Not a Safety Net

Repair support means one thing: a service channel remains open to fix a failed unit of the same model number. It does not mean you can buy more of them. If you want to expand a line, build up a spare-parts stock, or replicate the same configuration on a second line, repair support does nothing for you. New units are only available until the final order date.

The more damaging effect is on scheduling. If you back-calculate from the end of repair support, your project start lands somewhere in 2032. But in a PLC migration, the hardware is the fast part. The slow part is taking the program that is running today, moving it to a new CPU, and proving that it behaves the same way it did before. That is where the “we do not have the source” and “nobody knows what is inside” problems land on you — and they land at the worst possible moment, when there is no slack left in the schedule.

There is also a factor specific to overseas plants. In Japan, a service engineer may arrive the day after a failure. At a plant in Thailand or Vietnam, the same “we are still within repair support” status includes dispatch arrangements, customs clearance for the part, and visa or travel logistics for the engineer. Identical support status, very different downtime. If you are planning equipment renewal as part of a broader automation roadmap, the sequencing questions are covered in our guide to factory automation in Thailand.

Omron Is Moving the Same Direction

This is not a single-vendor story. Omron’s CS1 series is being wound down, and for the CJ2 series the end-of-production dates for certain models have been extended.

The easy misreading here is treating an extension as a reprieve. It is an extension, not a withdrawal. The dates moved by two years; they did not disappear. Plants that received the extension notice and shelved their migration plan will find, when they next open the file, that the runway is shorter than it was.

The honest way to read all of this is not as news about one CPU family. It is that a whole generation of control hardware installed through the 2000s is reaching end of life at roughly the same time — across vendors, across ASEAN, and in competition for the same limited pool of engineers.

Four Routes to a Black-Box PLC Program in an Overseas Plant

The first wall you hit in a migration project is usually not technical. It is informational: *we do not have the current program.* In overseas plants, there are four common routes to that condition.

(a) Head office shipped the machine, and everything else stayed behind

The parent company relocated a machine from a domestic plant, or procured it in Japan and installed it locally. The drawings and the project file live on a server in Japan — or did, once. The department that handled it has been reorganised out of existence. You send an enquiry and get back “we will look for it,” and the thread ends there. On site you have a paper operation manual and nothing else.

(b) The machine builder or system integrator is gone

The vendor that commissioned the line exited the market, wound up the business, or still exists but has lost the engineer who wrote the code. Control software is unusually person-dependent: one resignation can make a system effectively untransferable. Commentary on Thailand’s manufacturing sector points to a widening bottleneck in experienced technical talent, which suggests this route will produce more black boxes, not fewer, over the next several years.

(c) The project file is password-protected

Read-out is restricted by a password, or protection is applied block by block so that parts of the logic cannot be opened. If the unlock key was never handed over at acceptance, the plant that legally owns the machine cannot inspect its own control logic. This is not an accident. It is an omission in the original purchase agreement.

(d) Modifications drifted, and the file no longer matches the machine

Capacity increases, fixes for intermittent faults, a new product variant. Each change was made on the machine, and not every change made it back to the file server. The only authoritative version is the one running inside the PLC; the copy on your server is several generations behind. Plants usually discover this at exactly the wrong time — when they upload the running program for a migration and diff it against what they thought they had.

What the four routes share is this: while the machine runs, nobody has a problem. The problem appears in only three situations — when you upgrade, when it breaks, and when you need more capacity. So it sits untouched for years and surfaces at the moment you have the least schedule margin available.

There Are Three Kinds of “PLC Programming” Orders

“We would like someone to develop a PLC program for us” sounds like one request. Broken down, it is three requests with fundamentally different estimating characteristics. Mixing them inside a single competitive tender is the single largest cause of downstream disputes.

Type 1: New development (new machine, new line)

Built from a specification, starting from nothing. Once the I/O count, the operating sequence and the safety requirements are settled, the engineering hours are relatively predictable. The flip side: if you ask for a lump-sum price while the specification is still moving, each bidder will assume a different scope, and the numbers you receive will not be comparable.

Type 2: Modification and feature addition

Changing existing sequence control. The item most often left out of the estimate is that reading comes before writing. Someone has to decode the existing ladder logic and understand what moves when, and that decoding effort cannot be measured until the file is opened. No comments, comments only in Japanese, no naming convention — each of these multiplies the effort. When a quotation for a modification job says “to be re-quoted after investigation,” that is usually not evasion. It is the honest answer.

Type 3: Migration and replacement (legacy equipment upgrade)

The CPU model changes, so the work is convert, then correct the differences, then re-verify every function. This is a different animal from both Type 1 and Type 2. Even when a conversion tool carries most of the code across, the time concentrates in the places it does not handle cleanly: vendor-specific instructions, timer resolution, index modification, and communication module parameters. And because re-verification requires the line to stop, coordinating with the production plan usually becomes the critical path of the whole project.

The most common tender failure looks like this: Company A quotes Type 3 with re-verification included, Company B quotes something closer to Type 1 scope, and the buyer compares the two numbers side by side. The cheaper bid wins, and verification and on-site adjustment arrive later as variation orders. Writing one line at the top of your RFQ — *this project is Type 1 / Type 2 / Type 3, in these proportions* — prevents most of this.

Read the Cost in Five Layers

Quotations for control panel design, control panel fabrication and PLC software development become comparable once you split them into five layers.

These rates come from a published cost breakdown for control panels by kts-lab. They are one example from the Japanese domestic market, not a statement of market rate. They vary with company size, project difficulty and how busy the shop is, and they cannot be substituted for local rates in Thailand or Vietnam. What is worth taking from the table is not the numbers. It is the habit of splitting any quotation into five layers before comparing it.

Define “Investment” Before You Argue About Payback

When the discussion inside your company turns to payback period or return on investment, the first thing to settle is the denominator. For this article, we define it as follows.

Investment = the sum of engineering cost across Layers 1 to 5 + materials (PLC hardware, devices, cable, enclosures) + freight and incidental costs

Every rate in the table above is a man-day rate for labour; none of them include materials. Confusing the two produces the familiar situation where each line item looks cheap and the total does not reconcile.

Two errors show up repeatedly. They are different in nature, so keep them separate:

• Error 1, omission: dropping Layer 5 out of the denominator. Deferring installation and commissioning as “to be decided later” makes the investment look smaller and the payback look faster. The moment travel and accommodation land, the calculation collapses.
• Error 2, double counting: counting Layer 4 and Layer 5 spend as a benefit. Money spent on FAT and SAT belongs in the denominator. Adding “we tested thoroughly, so commissioning was shorter, and that saving is a benefit” to the numerator counts the same money twice.

Add all five layers of engineering cost into the denominator, then add materials and freight on top of them. Do not let anything already in the denominator reappear in the numerator. That single rule keeps an internal business case coherent.

Which Layers Actually Change When You Build in Thailand

“Is it cheaper to build in Thailand?” deserves a layer-by-layer answer rather than a yes or no.

• Layers 1 to 3 (design and software): almost entirely engineering hours, so the cost is a function of who spends how many hours. These layers are exposed to local rates.
• Layer 4 (fabrication and FAT): fabrication hours are exposed to local rates. Materials, which we deliberately placed outside the five layers, are a different story — imported components carry global prices plus import charges, so they can end up similar to, or higher than, Japan.
• Layer 5 (site installation and SAT): the largest differential. Sending people from Japan turns travel, accommodation and standby time into cost. For a company with engineers already in-country, the same work is domestic travel.

So the accurate framing is not “everything gets cheaper in Thailand.” It is that Layer 5, and part of Layers 1 to 3, change — while materials largely do not.

Labour Savings Alone Will Not Pay for the Project

Thailand’s minimum wage is set by province. As of July 2026 it ranges from THB 337 to THB 400 per day; there is no single nationwide rate. Note that these are daily figures, not hourly. Taking the top of the range and converting: THB 400 ÷ 8 hours = THB 50 per hour. Even if you eliminated one person’s work entirely — 2,400 hours a year, at 8 hours × 300 days — that is THB 50 × 2,400 = THB 120,000. Checked the other way, THB 400 × 300 days = THB 120,000. The two agree.

Real total employment cost is higher once social security, overtime and overheads are included, so treat this as a floor rather than an estimate. But the order of magnitude is clear enough: you cannot justify a PLC migration or an automation project on displaced labour cost alone. The benefits that actually carry a project are reduced defect escapes, shorter changeover time, fewer unplanned stoppages, and the emergency-response cost you avoid by not waiting until the equipment fails.

What the Incentive Schemes Actually Give Back

Thailand offers investment incentives, but folding them into a quotation before checking the conditions is risky.

• BOI “Smart and Sustainable Industry” measures: the corporate income tax exemption is 50% as the baseline. It reaches 100% only where automation or robotics is introduced into the production line and at least 30% of the value of the upgraded machinery is sourced from Thailand’s domestic automation industry. If a document tells you the exemption is simply 100%, check the conditions attached.
• Royal Decree No. 802 (200% deduction): it covers qualifying digital expenditure, but the cap is THB 300,000 per accounting period, purchases must be from a depa-registered vendor, and SME criteria apply (paid-up capital not exceeding THB 5 million and annual revenue not exceeding THB 30 million). Critically, it cannot be used by businesses already enjoying corporate income tax exemption under BOI or EEC schemes. Japanese-affiliated plants in Thailand are frequently under BOI promotion, which puts this deduction out of reach more often than not.

For context on the investment climate: the BOI reported USD 43.6 billion in investment applications for the first half of 2026. Competition for capital equipment and for engineers is unlikely to ease in the near term.

Six Items to Put in Writing Before You Place the Order

This is the core of the article. Whether outsourced PLC programming succeeds has less to do with the supplier’s technical ability than with a prior question: did you write down what stays with you? The following six items belong in the RFQ or the purchase specification, not in a verbal understanding.

Items 1 to 4: Separate Ownership from the Right to Use

“We get the source” and “we may freely modify it” are two different statements. You can hold the file and still be in a grey zone if copyright remains with the supplier and no modification licence was granted. Full assignment of copyright is not always negotiable — but a single sentence stating that “the customer, and any third party designated by the customer, may copy and modify the software for the purposes of maintenance, modification and relocation” is usually enough to make day-to-day operations work.

Item 2 needs more than “deliver with passwords removed.” Suppliers often retain block-level protection to safeguard their own libraries, and that can be legitimate. What you need in the acceptance checklist is which blocks are protected and who holds the key for each. A key held only in the supplier’s head is not a key.

Item 4 looks redundant once you have Items 2 and 3, but it is worth stating explicitly. When you eventually need to move to a different supplier, a broadly drafted confidentiality clause — “no disclosure to third parties” — can block you from handing the files to the new one. Carve out the exception at the beginning.

Item 5: Comment Language Is a Problem Unique to Overseas Plants

This item never comes up in a purely domestic Japanese order. Programs specified by a Japanese parent company are, naturally, commented in Japanese. But the people watching that line every day are Thai or Vietnamese maintenance technicians who cannot read those comments. So they call the Japanese expatriate engineer instead — and every night-shift and weekend stoppage funnels through one person.

Three things to decide at order time:

• The comment language in the program (Japanese only / with English / with Thai or Vietnamese)
• The language of the I/O list and sequence charts
• The display language of the HMI, and whether runtime language switching is required

Asking for translation afterwards is a separate order, and therefore a separate cost. Worse, translation commissioned in isolation is done by someone who does not understand the logic, so the accuracy of the result drops. Including it in the original scope is always the cheapest path.

For regional organisations, there is a second-order effect worth noting. Documentation written in English rather than Japanese is the version that a Vietnamese or Indonesian site can reuse when the same machine type is deployed there. Specifying English as the baseline documentation language is not only a maintenance decision; it is what makes a design portable inside your own group.

Item 6: Define the Contact Point by Response Time, Not by Company Name

“Call us if anything happens” is not an agreement. What belongs in the contract is the role or department to contact, the coverage hours (and whether they are Thailand time, Vietnam time or Japan time), the time to first response, the call-out fee and travel treatment when someone must come to site, and what changes once the warranty period ends. Only when all of that is written down can your team answer the question that matters during a stoppage: *who comes, at what cost, and within how many hours?*

Assumptions That Break When Japanese Equipment Arrives in ASEAN

Relocating a machine that ran fine in Japan, or issuing a Japanese specification unchanged to a local supplier, is a reliable way to stall at the installation stage.

Power: A Transformer Changes Voltage, Not Frequency

Japanese plants commonly distribute at 200 V class on the machine side. In Thailand, three-phase supply is 380 to 400 V at 50 Hz. The most frequent misunderstanding we encounter is the belief that “a transformer will sort out the frequency too.” It will not. A transformer changes voltage. Frequency passes straight through.

Note that Japan itself is split: 50 Hz in the east, 60 Hz in the west. Equipment from western Japan (60 Hz) moved to Thailand (50 Hz) will see induction motors started direct-on-line run slower than before; equipment from eastern Japan (50 Hz) sees no frequency change at all, though the voltage side still has to be reconciled. Where the frequency does change, That means different flow and airflow from pumps and fans, different conveyor speed, and a different cycle time. Inverter-driven axes can absorb much of this. Machines whose timing depends on mechanical relationships may need gear ratios or sprockets reviewed — and that review belongs in the design phase, not on the installation floor.

Three-Phase Three-Wire vs. Three-Phase Four-Wire

Against the delta configuration (three-phase three-wire) common in Japan, four-wire distribution with a neutral is the norm in Thailand. That changes where the single-phase control supply is derived from and how earthing is arranged. Panel drawings imported unchanged frequently assume a neutral that is not there, or ignore one that is.

Standards: IEC 60204-1 and IEC 61439 Do Different Jobs

Neither standard sits above the other. They answer different questions.

Thailand’s TIS (มอก.) standards adopt the IEC 61439 family, and only TISI (the Thai Industrial Standards Institute) can issue TIS certification. When a supplier says “we build to IEC,” that does not automatically mean the assembly is TIS certified. Establish which of the two your facility requirement actually calls for before you place the order, because retrofitting compliance evidence after the panel is built is expensive.

Sign-Off by a Licensed Engineer

In Thailand, certain electrical design and inspection work — particularly anything touching the plant’s incoming supply and distribution — requires sign-off by an engineer licensed by the Council of Engineers (สภาวิศวกร). This goes beyond the standalone control panel for one machine; it becomes relevant as soon as you modify existing switchgear or distribution. Whether your supplier can provide that sign-off in-house or has to subcontract it is a question to settle while you are still comparing quotations, not after the panel is on the truck.

Component Sourcing: The Same Part Number Is Not the Same Lead Time

A part number selected in Japan may or may not be obtainable locally. Distributor coverage, stock and lead time are entirely separate matters. Substituting locally available components during design does more than remove waiting time at installation: it means future spare-parts procurement can be completed in-country, which is an operational benefit that lasts for the life of the machine. This is a design decision you can only make during an upgrade.

IEC 61131-3 Does Not Guarantee Portability

“If we keep the program compliant with the international standard, we can switch vendors later.” Unfortunately, that expectation does not survive contact with practice.

IEC 61131-3 is the international standard for PLC programming languages. It historically defined five languages — LD, ST, FBD, SFC and IL. In Edition 4 (2025), IL (Instruction List) was removed from the standard, leaving four languages in the current edition. A notation used for decades can, in other words, fall out of the standard entirely.

Standard-Compliant Does Not Mean Drop-In

The standard defines language syntax and structural framework. It does not unify how each vendor implements them. In practice, migration is blocked by things like:

• Vendor-specific extended and special instructions (common in positioning, communications and motion)
• Internal representation of data types, register addressing schemes, and index modification syntax
• Timer and counter resolution and retentive behaviour
• Library and function block compatibility
• How variables are bound to the HMI

The cumulative result is that moving across platforms is effectively a rewrite. Vendor lock-in in PLC programming is a recurring theme in industry commentary, and budgeting on the assumption that “compliance means easy switching” is how migration projects overrun.

Maintenance Teams Prefer Ladder

There is a second, human constraint. Structured Text is well suited to arithmetic and data handling, but it is harder for a plant maintenance technician to read. Ladder logic has been understood as an extension of relay circuits for decades, and an electrical technician can follow a chain of contacts and coils without formal training.

A common pattern in ASEAN plants: a machine arrives with logic written predominantly in ST, and from then on the local team cannot isolate a fault during a stoppage. Every incident becomes a wait for the vendor. The problem is not ST. The problem is choosing a notation the site cannot read and not budgeting for the training that would fix it.

Three decisions to make at order time:

1. Which parts are written in ladder and which in ST (state the split explicitly)
2. Whether maintenance training is inside the scope of the order or a separate line item
3. How many people are trained, and in which language the material is prepared

Training cost is easy to cut from a quotation because it is invisible. What you cut, you pay back later as downtime.

Things You Can Only Add During an Upgrade

The panel door is open, and the wiring is accessible, at only two kinds of moment in a machine’s life: at commissioning, and at a major upgrade or migration. On a line that cannot stop, “while you are in there, could you also…” is not a request that gets approved. That is why what you choose to build in during an upgrade has an outsized effect on the following decade.

Build the Data Tap Now, Even If You Are Not Ready to Use It

If you are considering production monitoring or IIoT — or think you might consider it in two years — there is real value in installing the tap now even with no application behind it.

Concretely: an additional Ethernet port, a tidied and documented address map for the devices representing machine state (running, stopped, fault, production count), a reserved area readable from outside, and a communication module if one is needed. During an upgrade these are absorbed as incremental hours. After the line is running, adding them means stopping production and opening the panel all over again. What you eventually do with the data is covered in our article on factory IoT and production monitoring.

Tighten the Remote Maintenance Path While You Have the Chance

Most upgrades either introduce vendor remote maintenance or inherit an existing arrangement. Design this deliberately. Ransomware campaigns targeting factory VPNs have been reported publicly, and an always-on vendor VPN is one of the easiest entry points an attacker can find in an industrial network.

Four points to specify in the order:

• Open the connection only when needed (no permanent tunnel — the just-in-time access model)
• Restrict the path so it reaches only that machine (least privilege; not a route onto the whole plant network)
• Time-box every session (closed when work ends, or automatically expiring)
• Keep an audit log of who connected, when, to what, and what they changed

Exceptions created “just for maintenance” have a way of staying open for years. The wider network design principles are set out in our guide to OT security for factories.

Re-examine the Safety Circuit

The older the machine, the more its safety circuit reflects the thinking of its installation year: emergency stop category, redundancy in the safety relay, door interlock architecture, reset logic. If you are replacing the CPU anyway, evaluating the safety circuit at the same time is simply efficient. It matters even more if you expect to introduce robots into processes where people work close to the machine — designing the safety requirement in from the start is far cheaper than adding it later. The safety standard for collaborative applications was revised as ISO 10218 in 2025, and we cover what changed in our article on implementing collaborative robots.

Four Types of Supplier, and What Each Is Good For

Suppliers for outsourced PLC software development fall into four broad types. None of them is inherently better than the others; each fits certain projects and certain in-house capabilities.

Judge on Five Axes Before You Look at Price

Line the candidates up on these five axes first. Price comes after.

Every axis where a candidate scores poorly converts directly into future risk. The useful part of the exercise is that weakness on one axis can be compensated on another: if continuity is doubtful, tighten the source-delivery and documentation requirements accordingly, so that the next supplier can pick the work up. General criteria for evaluating engineering and software partners in the region are discussed in our guide to choosing a system development company in Thailand.

One more structural question: single supplier or split. Awarding hardware (Layers 1 and 4) and software (Layers 2 and 3) to different companies can reduce unit rates, but when a fault appears you lose days arguing whether it originated in design or fabrication. There is a further trap: if you have not also decided which of the two carries Layer 5 — installation and commissioning on site — then the first problem that appears on the factory floor stalls while each party argues it falls in the other’s scope. If you split, name in the purchase order who carries Layer 5 and which party owns overall integration responsibility. Without that, you own it by default.

Five Failure Patterns We Keep Seeing

Here is the same material reframed as the failures that actually occur. All five are preventable on paper, before the order is placed.

Pitfall 1: Accepting the Machine With No Source and No Documentation

Commissioning is late and the only thing that matters is starting production. Under that pressure, source code and documentation quietly drop out of the acceptance criteria. The machine runs, so nobody feels the loss. The loss arrives three years later, when the engineer who ran the project has moved on and the supplier relationship has lapsed. One sentence in the acceptance protocol prevents it: “acceptance is conditional on receipt of the complete project file set and the I/O list.”

Pitfall 2: Comparing Bids That Mix the Three Order Types

We covered why Type 1, Type 2 and Type 3 estimate differently. In practice all three often coexist in one project — you upgrade part of an existing line and add a new function on top. If the RFQ asks for a lump sum without separating them, one bidder will include the modification scope and another will not, and both will look like valid numbers. The only reliable countermeasure is to require subtotals broken out by Type 1, Type 2 and Type 3 in the RFQ itself.

Pitfall 3: Leaving Site Installation and Commissioning Out of the Estimate

Layer 5 gets excluded as “to be decided later,” and the capital request goes through the internal approval process without it. Reality then adds flights, accommodation, standby time, interpreters, local support labour, and extra nights when commissioning runs long. Most of the reasons commissioning runs long are not the machine at all — they are utilities and production scheduling. Always include Layer 5, even as an allowance, and agree the daily rate for overrun before work starts.

Pitfall 4: Ordering to a Japanese Specification and Stopping at Installation

Voltage, frequency, wiring configuration, standards, component availability — all covered above. Drawings are issued unchanged to a Japanese specification, and the missing neutral is discovered after the panel lands in Thailand; or a selected component is unavailable locally and adds a month of waiting. Bringing one local engineer into the design review, once, catches most of these before they cost anything.

If your plan involves relocating equipment to a third country, check the destination’s import rules early as well. In Vietnam, for example, new rules governing the import of used machinery (Circular 30/2025/TT-BKHCN) took effect on 1 January 2026, setting technical criteria that determine whether a given machine may be imported at all. Because eligibility turns on the age and condition of the equipment, any upgrade plan that assumes a cross-border relocation needs that confirmation at the start, not after the machine is crated.

Pitfall 5: No Named Contact Point, Then a Resignation

During commissioning the engineers on both sides talk directly; if something happens, you call a mobile number. That relationship stays personal for a few years, and the day either person leaves the company, the channel disappears. This is route (b) from the beginning of this article, and the uncomfortable truth is that it is regenerated with every new order unless someone stops it. Put a department name and a response time in the contract, so the relationship is company-to-company rather than person-to-person.

How to Run It: Five Stages From Inventory to Handover

Finally, the practical sequence, in five stages. The point worth emphasising is that the first stage requires no supplier at all — you can complete it entirely in-house.

Start With the Equipment Register

This is the highest-return activity in the entire process. As long as this register is vague, nobody can quote you a price — no matter who you ask. Conversely, once the table exists, every supplier you approach returns a comparable number.

The more fields you can only fill with “unknown” or “unverified,” the harder the upgrade will be. Do not try to eliminate the blanks first. The objective at this stage is simply to make visible where the blanks are. Then work through the equipment in order of stoppage impact, and your budget allocation stops being arbitrary.

A Realistic 90-Day Path

Stages 1 to 3 can be completed in about 90 days without stopping production. Stages 4 and 5 are the execution phase that follows.

The step that consumes the most time in days 1 to 30 is uploading from the running machines and diffing against the files you hold. This is where black-box routes (c) — password protection — and (d) — version drift — surface. Surfacing them early is the entire purpose of the 90 days.

One caveat on timing: these 90 days take you to a supplier decision — they are not a plan for beating the final order date. From this article’s publication date of 30 July 2026 there are only 62 days until 30 September, so running the full 90-day sequence would overshoot it. Where a deadline applies, separate the CPU procurement from the program migration. Use the day 1 to 30 asset register purely to pin down which model numbers you need, and secure the hardware before the order window closes. Migration and re-validation can follow. Get the parts securable first, then work on the contents.

Summary

Outsourcing PLC programming looks like a technical decision. It is actually a decision about what you keep. The source code. The password keys. The right to modify. Documentation in a language your maintenance team can read. And a named party to call when the line stops. If those five things sit on your side of the table, you can choose your supplier at the next upgrade and at every modification after that. If they do not, you have exactly one option, and it was decided years ago.

The timing pressure is real: final orders for the MELSEC-Q Universal Model QCPU close on 30 September 2026, and Omron’s CJ2 deadlines follow at the end of March 2027 and the end of March 2028. But you do not need to have decided whether to upgrade in order to start. Building the equipment register and finding out where the blanks are is available to you today, at no cost beyond a few people’s time.

At TOMAS TECH we build factory IT and FA/automation systems for manufacturers across Thailand and the wider ASEAN region. If it would help to talk through how to start the equipment inventory, or how to establish the condition of an existing program before you commit to anything, we are happy to have that conversation at the pre-decision stage — including the case where you have not yet decided whether an upgrade is needed at all. You can reach us through our contact page.

Frequently Asked Questions

How much does it cost to outsource PLC programming?

There is no single figure, because the cost depends on the order type (new development, modification or migration) and on the I/O count. What does help is reading the quotation as five layers. As a reference point, a published cost breakdown from a Japanese control panel builder gives approximately JPY 40,000 per man-day for design, JPY 40,000 for programming, JPY 30,000 for fabrication, JPY 40,000 for in-house testing and JPY 40,000 for site work. Note that these five rates do not map one-to-one onto the five layers: fabrication and in-house testing both sit in Layer 4, and Layer 3 has no separately published rate. That is one example, not a market rate, and it should not be used as a proxy for local rates in Thailand or Vietnam. Note also that all of these are labour rates: materials such as the PLC hardware and field devices are not included.

Will I automatically receive the PLC source code?

No. If it is not in the contract, delivering executables only is not a breach. Specify four things when you place the order: whether project files are delivered and in what format; how passwords and protection are handled and who holds the unlock keys; whether copyright is assigned or a licence covering modification is granted; and whether you may have a third party modify the program. For machines already delivered, there is often still room to negotiate these terms — a maintenance contract renewal is a natural moment to raise them.

Should control panel design and fabrication go to the same company?

Either arrangement can work, but if you split them, name the party responsible for overall integration before work starts. Awarding design (Layer 1) and fabrication (Layer 4) separately can lower unit rates, yet when a fault appears you spend days determining whether it originated in design or in build. On overseas projects that argument crosses borders and time zones, which makes it slower still. If you are working with a supplier for the first time, awarding the whole package to one party is usually faster overall.

When should we upgrade the PLC on ageing equipment?

Three inputs. First, the manufacturer’s end-of-production and repair-support dates — for the MELSEC-Q Universal Model QCPU, final orders close on 30 September 2026 and repair support ends on 31 October 2033. Second, the production impact if that machine stops. Third, and in practice the most important: whether you actually hold the program and the documentation. For equipment where you do not, the duration from project start to completion cannot be estimated at all. Work backwards from what is missing, not from the repair-support date.

Is HMI and touch panel screen development quoted separately?

In most cases it is estimated as separate effort from the PLC software. In the framework used in this article it is Layer 3. The effort scales with the number of screens, whether multiple languages are required, how far alarm and trend displays are developed, and whether recipe management is included. For plants in ASEAN we recommend putting display language switching (English, Thai, Vietnamese, Japanese as applicable) into the requirements from the start. Adding it later often forces screen layouts to be reworked, which costs more than doing it once.

Does building the control panel in Thailand reduce cost?

The answer differs by layer. Site installation and commissioning (Layer 5) is where the difference is largest, because travel, accommodation and standby time for engineers flown in from Japan largely disappear. Design and software (Layers 1 to 3) are also labour-driven and therefore exposed to local rates. Materials for panel fabrication — a cost we deliberately placed outside the five layers — behave differently: imported components carry global prices plus import charges, so they may be similar to Japan or higher. Rather than “everything gets cheaper,” the accurate expectation is that some layers change and others do not. Confirm separately which parts of the scope require TIS or IEC 61439 compliance and sign-off by a Council of Engineers licensed engineer, as that work has to be planned in from the beginning.

References

1. Mitsubishi Electric Technical News FA-D-0418-A (discontinuation of MELSEC-Q Universal Model QCPU)
2. Mitsubishi Electric: MELSEC-Q discontinued product list
3. FA equipment discontinuation quick reference (MELSEC-Q)
4. Omron: latest information on discontinued products and recommended replacements
5. Takashima Denki: Omron PLC (CS1/CJ series) — extension and withdrawal of end-of-production dates
6. Control panel cost breakdown (example of man-day rates)
7. Supply voltages by country (overseas specification, seigyo-sekkei.com)
8. JETRO: Overview of Thai Industrial Standards (TIS) and obtaining certification marks
9. Electrical panel standards: IEC and มอก. (TIS) in Thailand
10. Council of Engineers Thailand (สภาวิศวกร)
11. Product standard for industrial control panels: IEC 60204-1 or IEC 61439-1 (GT Engineering)
12. An Overview of IEC 61131-3 (control.com)
13. Systemic Challenges in PLC Programming and Maintenance: Vendor Lock-In
14. Ransomware targets factory VPNs as Secomea urges OT remote access overhaul (VIR)
15. Why Thailand’s Manufacturing Sector Is Facing a Talent Bottleneck
16. Thailand Secures $43.6bn 1H 2026 Investment Surge (BOI OSOS)
17. Keep up with IEC control cabinet standards (Siemens)
18. Vietnam: New Import Rules on Used Machinery (Circular 30/2025/TT-BKHCN, effective 1 January 2026)