*Industry news | October 12, 2026* Rockwell Automation has set October 31, 2026 as the discontinuation date for the FLEX I/O ControlNet adapter, and there is no successor product named. The affected part numbers are 1794-ACN15, the standard ControlNet adapter, and 1794-ACNR15, the redundant-media version. That gives a plant roughly three weeks from today to place a last-time buy. After that, the adapter comes off the shelf in Rockwell's catalog, and the only route to a replacement is whatever distributors and independent suppliers still hold in stock. A stopped line does not care how clever the migration plan is. It cares that the adapter on the shelf fits. So the October 31 date is not an engineering footnote, it is a purchase order deadline, and it arrives in three weeks. Read the rest of the picture before you decide how much stock to buy. The ControlNet adapter is one part in a wider FLEX I/O retirement cycle, and the lead times on what remains tell you how much slack you actually have.   The date that matters is October 31 October 31 is the last day the 1794-ACN15 and 1794-ACNR15 can be ordered from Rockwell. The product page carries lifecycle status End of Life with that discontinuation date, and the wording is direct enough to leave no room for interpretation: "Rockwell Automation announces that as of October 31, 2026, the Flex ControlNet Adapter will be discontinued and no longer available for sale. Customers are encouraged to remove references to the affected product(s)." That last sentence is the one worth reading twice. Rockwell is telling you to scrub the part number from your drawings, your bills of material and your maintenance procedures. It is not telling you what to put in its place. The replacement category on the product page is "Engineering Replacement." That is a specific and unhelpful phrase for anyone who owns this hardware. A functional replacement drops in and does the same job. An engineering replacement means the job can still be done, but the design changes, and the people who do the design work are not the ones holding the maintenance budget. No successor part number is listed. That is the detail that should drive your purchasing decision this month. When a vendor names a replacement, you can plan around it. You read the data sheet, you check the dimensions, you order one for the bench and you find out what breaks. When the replacement field is blank, you are the one writing the migration plan, and you are doing it without a vendor telling you which way to go. For a plant that keeps legacy ControlNet running because it works and because the capital was approved in a different decade, this is the moment where the platform question lands on the desk. Either you buy shelf stock to extend the life of what you have, or you accept that the ControlNet footprint is now on a countdown that you have not defined. The countdown is real either way. Rockwell has stopped selling. Every adapter in the field is now one failure away from being a problem you cannot buy your way out of at the usual channel. What is going out of production, and what is not The retirement does not touch every FLEX I/O adapter at once, and the gaps between the dates matter more than the dates themselves. Here is the model list, with what each one means for the shelf. 1794-ACN15, the FLEX I/O ControlNet adapter, non-redundant media. Lifecycle status: End of Life. Discontinued October 31, 2026. Replacement category: Engineering Replacement. No successor named. 1794-ACNR15, the FLEX I/O ControlNet adapter with redundant media. Lifecycle status: End of Life. Discontinued October 31, 2026. Replacement category: Engineering Replacement. No successor named. 1794-ACNR15XT, the Flex XT ControlNet adapter for extended temperature environments. Lifecycle status: End of Life. Discontinued December 31, 2027. Replacement product named: 5094-AENTRXT. 1794-AENT, the FLEX EtherNet/IP adapter with 10BaseT. Lifecycle status: End of Life. Discontinued September 30, 2028. Replacement category: Functional Replacement. Replacement product named: 1794-AENTR. 1794-AENTR, the FLEX I/O dual port EtherNet/IP adapter. Lifecycle status: Active Mature, which means it is still current. Rockwell lists an estimated lead time of 5 days. The two ControlNet adapters go first, and they go together. The XT variant, the one built for hot, cold and dirty conditions, gets an extra fourteen months and a named destination. The EtherNet/IP adapter gets almost two more years and a clean functional replacement. That pattern is not random. The platform is not being killed in one stroke. It is being retired from one network protocol at a time, and ControlNet is at the front of the queue. If you keep both a standard ControlNet adapter and an XT unit in the same plant, which is common on sites with a mix of climate-controlled panels and outdoor or washdown enclosures, the two are not on the same clock. Do not let one date blur into the other during a budget conversation. The XT users have a named successor and more time. The standard ControlNet users have neither. The EtherNet/IP side is the part of the answer that is still comfortable. The 1794-AENTR is Active Mature and Rockwell posts a 5 day lead time. That is not a part that is about to strand you. It is a part you can order through normal channels and get in the same week. The reason that matters here is that the EtherNet/IP adapter is the natural landing spot if the plant moves off ControlNet. It exists, it ships fast and it is not on a discontinuation list. That combination gives you an option, and the option costs less sleep than the blank field next to the ControlNet part. Start with what a FLEX I/O adapter actually does. One adapter connects up to eight FLEX I/O modules to the network. On the ControlNet side, that traffic runs on ControlNet V1.5. The adapter is the single point where a rack of I/O either talks to the controller or goes dark. One adapter, up to eight modules, one network connection. When that adapter fails, you do not lose one input point. You lose the whole rack. That is the number to hold onto when you are deciding how many to buy. The exposure per adapter is not the cost of the adapter. It is everything wired through it. Why a 73-day lead time makes this worse than it looks The 1794-ACN15 product page lists an estimated lead time of 73 days. Seven hundred and thirty hours of waiting, and that is Rockwell's own estimate on a part that is being discontinued in three weeks. Sit with that for a moment. The last date to order is roughly three weeks away. The lead time to receive is roughly ten and a half weeks. Those two numbers do not line up, and that mismatch is the whole story. An order placed on the last day of October does not arrive in October. It arrives somewhere in the first half of January, if the estimate holds. That is the trap in a last-time buy. People hear "last order date" and picture a firm deadline they can work backward from. In practice, a last-time buy is a commitment to wait, and you place it before you know what the next twelve months of failures will look like. For a maintenance planner, this changes the shape of the decision. You are not choosing between buying now and buying later at the same price and the same availability. You are choosing between securing stock at a known lead time and gambling that nothing in the rack fails while the channel is closing behind you. A 73 day lead time is also a signal about the supply chain behind the part. Long lead times on end-of-life hardware usually mean the part is built to order, or drawn from shrinking inventory, or subject to allocation. None of those conditions improve after the discontinuation date. They get worse, then they get scarce, then the only stock left is what is already sitting on shelves. The EtherNet/IP adapter's 5 day lead time sits right next to that 73 day figure on the same catalog, and the contrast is the point. One is a part you can order this afternoon. The other is a part you commit to months in advance to receive it at all. The practical reading is straightforward. Availability after October 31 is not a question of price. It is a question of whether anyone still holds the part. Distributor stock and independent suppliers are the channel, and that channel thins every month. If you have a critical rack, the lead time is the argument for acting this month rather than in the first week of November. The difference between ordering on October 24 and ordering on November 3 is not three percent of a purchase order. It is the difference between Rockwell's supply chain and whatever is left in the secondary market. If you need help locating what is still available, discontinued Allen-Bradley parts is a reasonable place to start a search, and it is worth doing it before the last sale date passes rather than after. The two realistic paths: keep the platform, or move to EtherNet/IP Once the adapter is gone from the catalog, the plant has two honest options. Everything else is a variation on one of them. The first path is to keep the platform. You buy enough adapters to carry the ControlNet racks through the horizon you are willing to plan for, you store them properly, and you keep running. This path does not require a capital project, a controller change or a shutdown window. It requires a purchase order and a shelf. The cost of this path is the shelf itself. Adapters that sit for years draw no power and cost nothing to hold, but they tie up working capital, they have to be stored somewhere dry and stable, and someone has to keep track of them. The benefit is that a failure on a Tuesday afternoon becomes a fifteen minute swap instead of a production event. The second path is to move to EtherNet/IP. This is where the plant stops treating ControlNet as a permanent fixture and starts treating it as a project. The 1794-AENTR exists, it is Active Mature, it ships in about 5 days, and it is the adapter the FLEX I/O platform is heading toward. Moving means more than changing the adapter. The controller side has to speak the same protocol, the network layout changes, the addressing changes, and the people who maintain the system have to learn a different set of tools. That is why this path is a project and not a purchase order. It is worth being clear about the size of the gap. ControlNet V1.5 and EtherNet/IP are different networks with different cable plants, different topology rules and different diagnostics. A rack does not migrate by swapping one card. The rack has to be rehosted on a network that the rest of the control system may or may not already speak. For some plants, that work is already underway. Any site that has been modernizing its control layer for the past decade probably has an EtherNet/IP backbone somewhere, and the FLEX racks are the last ControlNet island still connected. For those sites, the discontinuation is a schedule nudge, not a crisis. For other plants, the FLEX I/O ControlNet racks are the backbone. They were installed as the smart, flexible option at the time, and the rest of the plant grew up around them. For those sites, the second path is a genuine capital conversation, and the first path is the one that keeps the lights on while that conversation happens. The practical answer for most operations is a blend. Buy shelf stock for the adapters that carry the racks you cannot afford to stop, and start the engineering work on moving the racks that are either less critical or already near a planned outage. The one thing the blend is not, is free. Every month you spend deciding is a month the last-time-buy window is shorter, and the price and availability of whatever is left in the secondhand channel is a moving target. If you are weighing both paths, Rockwell Automation spares is a useful reference point for what is still being sourced and supplied in the aftermarket, so you can price the stock side of the decision against the project side. What an unplanned adapter failure actually costs the operation This is where people want a number, and this is where a responsible article refuses to give one. Downtime cost depends entirely on the plant, and any single figure quoted here would be invented, wrong for your site and useless as a budgeting input. Here is how to build your own number instead. Start with your line rate or your product throughput, whatever unit you already track output in. Multiply by the value of that output. That gives you the cost of an idle hour. Now add the labor that stands around while the line is down. Add the restart cost, which for some processes is real scrap and for others is hours of warm-up and re-qualification. Add overtime if you recover the lost volume with extra shifts. The result is your cost per stopped hour. That is the number that belongs in the decision, and it is a number only you can calculate. Then do the arithmetic that matters. If the adapter is on the shelf, the failure costs you a swap: call the electrician, power down the rack, change the card, power up, verify. A couple of hours on the wide end, often less. If the adapter is not on the shelf, the clock keeps running. You call your suppliers, you find out nobody has immediate stock, you wait. Now add the freight cost of an expedited shipment, and the premium that any scarce part commands when the seller knows you need it today. That gap between a swap and a wait is the entire justification for holding spares. It is also the number that makes a shelf of adapters look cheap. The 73 day lead time on the 1794-ACN15 turns a normal supply order into a production event. A part that is normally a next-week purchase becomes, after October 31, a part that might not be purchasable at all at the price or speed you are used to. There is a second cost that is easy to forget. A rack that has lost its adapter takes its up to eight I/O modules offline with it. If those modules feed a safety interlock, a batch recipe or a quality measurement, a "two hour electrical fix" can turn into a batch rejection or a quality hold that outlasts the repair by a wide margin. That is why the spares question is never really about the adapter. It is about the rack, and the rack is about whatever the plant does with the signals. If you have never built a downtime cost figure for your own site, do it now, while the part is still purchasable. It is much easier to justify a stock purchase with a real number on a spreadsheet than with a feeling that spare parts are probably a good idea. For plants keeping older platforms alive, legacy PLC hardware is often the practical source for the parts the original vendor no longer sells, and knowing your source before a failure is part of the same calculation. A spares audit you can run this week This is a walk-the-floor exercise and it does not need a project code. The goal is a single honest number: how many 1794-ACN15 and 1794-ACNR15 adapters your site actually depends on. Start at the panel. List every FLEX I/O rack that has a ControlNet adapter. For each one, write down the panel name, the process it serves, and whether the process is critical or can wait. Do not skip the awkward ones in remote corners of the plant; those are the racks most likely to be forgotten during planning and most painful to reach during a failure. Now count the adapters. Note whether each one is a 1794-ACN15 or a 1794-ACNR15. If your site uses redundant media anywhere, the ACNR15 count is the one to watch, because both versions share the same October 31 discontinuation and neither has a named successor. Next, check your own storeroom before you check anyone else's. Some sites are sitting on adapters they bought years ago and never recorded. A shelf count is free information, and it can change your whole buying plan in an afternoon. Then compare what you need against what you hold. The gap between those two numbers is what you are buying this month. As you go, note the failures. If an adapter has failed in the last couple of years, write down why. An adapter does not fail in isolation. Heat, vibration, dirty power and loose backplane connections all take a toll, and a rack with a history of failures will keep failing after the catalog closes. Racks with clean histories are the better candidates for the longer migration timeline. Two more things go on the audit sheet. First, list the XT racks separately. The 1794-ACNR15XT runs to December 31, 2027 with a named replacement in the 5094-AENTRXT, so it is a different line item with a different urgency. Mixing it into the standard ControlNet count will either overstate or understate your exposure. Second, write down who owns the decision. Every last-time-buy that stalls does so because the maintenance department wants the stock, the finance department wants the justification, and nobody wrote down whose signature is on the purchase order. Fix that before the last week of October. Once the audit sheet is done, the buying decision is arithmetic, not opinion. You know how many racks you carry, how many adapters you hold, and how many months of coverage you want. Multiply the gap by the unit cost, present it against your own downtime cost figure, and the conversation with finance runs itself. Do the audit this week even if the answer is "we already have enough." The audit is cheap. Finding out you were short is expensive. If your audit turns up gaps you need filled fast, industrial automation spares is the kind of source to have on standby before the deadline, not after it. The rest of the FLEX I/O lifecycle picture Zoom out, because the ControlNet adapter is not the only date on the calendar, and a plant that only reacts to this one will be doing the same exercise again in a year. Here is the sequence, in the order it arrives. October 31, 2026: 1794-ACN15 and 1794-ACNR15, the FLEX I/O ControlNet adapters, discontinued. Engineering Replacement, no successor named. December 31, 2027: 1794-ACNR15XT, the Flex XT ControlNet adapter, discontinued. Replacement product 5094-AENTRXT named. September 30, 2028: 1794-AENT, the FLEX EtherNet/IP adapter with 10BaseT, discontinued. Functional Replacement, replacement product 1794-AENTR named. Still current: 1794-AENTR, the FLEX I/O dual port EtherNet/IP adapter, Active Mature, estimated lead time 5 days. No discontinuation date listed. Read that as one continuous retirement of the FLEX I/O adapter family, protocol by protocol. ControlNet first, in two waves, with the standard adapters leading and the XT variant following fourteen months later. Then the oldest EtherNet/IP adapter, the 10BaseT version, with a functional replacement already named and a target two years out. The destination is consistent. Every retired adapter points, directly or eventually, at the EtherNet/IP side of the platform. The 5094-AENTRXT is the named replacement for the XT ControlNet adapter. The 1794-AENTR is the named replacement for the 1794-AENT. The 1794-AENTR itself remains Active Mature, shipping in about 5 days, with no retirement date announced. The pattern tells you what the platform wants to be. FLEX I/O stays. ControlNet goes. EtherNet/IP is where the racks are expected to land. For a maintenance manager, that pattern is more useful than any single part number. It means the migration plan you write today, whether it finishes next year or in five years, has one obvious direction. It also means the 1794-AENTR is the part you should understand now, because it is the adapters' future, and it is the one part in this whole family that is still easy to buy. Be precise about what Rockwell has and has not announced, because imprecision here costs money. What is documented is the sales discontinuation and the dates above. Rockwell has not announced an end of repair, an end of support, or a cut-off for spare parts through its own service channels. Do not let anyone in your next planning meeting claim otherwise. What the discontinuation does mean is concrete enough. The part is no longer sold. Availability becomes a question of existing stock. And the farther you get from October 31, the thinner that stock gets and the more the secondary market sets the price. That is the whole reason to treat the date as a purchase order deadline. It is not the day the part dies. It is the day the part stops being an order and starts being a search. Three weeks is not much runway, but it is enough to run the audit, count the racks, and place the order while the part is still a part number in a catalog rather than an inquiry to a supplier. The plants that will come out of this cleanly are the ones that spend the next week on the floor with a clipboard, not the ones that will spend the first week of November on the phone. The calendar is set. The successor field is blank. The only variable left is how many you decide to put on the shelf.

October 13,2026

  On Wednesday, Rockwell Automation stops selling new PowerFlex 70 drives. Not all of them, but the ones most plants actually have bolted into a cabinet. If you have a line running on a PowerFlex 70 and you have not yet decided what happens when the next one fails, that decision has now been forced on you. It is two days out. I have watched three of these transitions in fifteen years. The pattern does not change. The vendor announces, everyone nods, nobody budgets, and eighteen months later a maintenance tech is on a phone at 2 a.m. trying to find a rebuilt unit from a broker who will not answer the provenance question. That is the bill I am trying to keep you from paying. This is not a technical piece. You know what a drive does. This is about what the calendar costs, which parts you can still buy and when, and what you should put on a shelf before the dates arrive.   What exactly ends on September 30, and what does not   Read the announcement carefully, because the headline oversells it and the detail is where your budget lives. Rockwell's own PowerFlex 70 product page states: "Rockwell Automation announces the PowerFlex 70 AC Drive will be discontinued and no longer available for sale as of September 30, 2026. End of life status and discontinuation date only apply to the IP20/NEMA UL Type 1, IP54/NEMA UL Type 12, and Flange versions. The NEMA 4X versions remain in Active Mature status. To support your transition, we recommend modernizing to the PowerFlex 750 family of AC drives." That is the vendor's language, not mine. Two things matter in it. First, the NEMA 4X versions stay in Active Mature. If your washdown or outdoor installations run 4X enclosures, you keep a sales channel for now. Active Mature is not the same as Active, and it is not a promise of forever, but it is not a stop date either. Do not lump your 4X units in with the rest of the fleet when you build the audit. You will overspend on parts you can still order normally. Second, the stop applies to specific enclosure families, and Rockwell points users to its Product Lifecycle Status search tool to check individual catalog numbers. That is not a formality. A plant with forty PowerFlex 70 nameplates on a spreadsheet thinks it has forty problems. In practice the mix of frames, enclosures and option cards means you have maybe six or eight distinct catalog numbers that actually matter, and some of them are in a different lifecycle status than the rest. Pull the numbers. Read them one at a time. Do not trust the family name. The other half of this news cycle is Mitsubishi Electric, and there the deadlines are longer but the details are sharper. Two Sales and Service bulletins, both issued April 2026 and revised in August 2026, both citing difficulty obtaining components, cover the HF-KN series AC servo motors (Bulletin No. 26-07AE) and the MR-JN series AC servo amplifiers (Bulletin No. 26-06AE). The HF-KN has been on the market 17 years, since 2009. That is a long run for a servo motor, and it means the installed base across packaging, food processing and light assembly lines is deep. The bulletin sets a made-to-order transition on March 31, 2027, order acceptance through March 31, 2028, last shipment and production end on March 31, 2029, and repair acceptance and parts supply through March 31, 2036. A third bulletin, No. 26-09AE, covers the MR-EM340GF Simple Motion board and the SW1DND-EMSDK-B MELSOFT EM Software Development Kit. Same dates. More on that one shortly, because the replacement part is the awkward one.   The calendar, in the order it will hit you   Here is the sequence. No table, just the dates in the order they land, because that is how you build a purchase order schedule. · September 30, 2026: new PowerFlex 70 sales end for IP20/NEMA UL Type 1, IP54/NEMA UL Type 12, and Flange versions. Two days from now. If you need a new unit in one of those enclosures, this is the last full business week and it is already ending. · December 2026: the FX3G series PLC reaches its discontinuation date, following a June 2025 Mitsubishi announcement that set a last order date of June 2026 with the FX5UJ as the recommended successor. If you missed that order date, you are already in the secondary market for FX3G hardware. Note this one, because it is the case study for the whole article: the notice came eighteen months ahead, the last order date passed quietly in June, and the discontinuation date arrives in December with nobody's attention on it. · March 31, 2027: HF-KN servo motors, MR-JN servo amplifiers and the MR-EM340GF motion board and its SDK all move to made-to-order. This is the first real pinch point. Made-to-order means lead times stretch and the vendor is building to demand, not to stock. If you have a project commissioning late 2027, that date is on your critical path whether your project manager knows it or not. · April 2027: CE and UKCA marking removal takes effect on orders received on or after that date, for the same three Mitsubishi product groups. This is a different kind of deadline, and I will give it its own section below. · March 31, 2028: order acceptance closes on HF-KN, MR-JN and the MR-EM340GF. After this date you cannot place an order at all. This is the hard buy date. Everything before it is planning. After it, you are buying from someone else's shelf. · March 31, 2029: last shipment, and production ends, for HF-KN, MR-JN and the MR-EM340GF. Whatever was ordered by March 2028 ships by this date and then the factory stops. · March 31, 2036: repair acceptance and parts supply end for the same Mitsubishi product groups. Seven years of repair window after production ends. That is generous compared to what most drive lines get, and it is the single most important line in this article for anyone panicking about a 2029 date. Read that list once more and notice what is missing. Rockwell has published a sales end for the PowerFlex 70. Rockwell has not, on the material I can find, published an end of repair, end of support or spare parts cut-off for the same drive. I am not going to invent one for them, and you should be suspicious of any supplier who tells you a Rockwell repair window that the vendor has not stated. The absence of a date is not a runway. It is an absence of a date. Plan accordingly.   Why April 2027 is not like the other dates   Every other date on that calendar is about volume: can I order it, will it ship, can it be repaired. April 2027 is about paperwork, and paperwork is the deadline that catches European and UK plants off guard. Both the HF-KN and MR-JN bulletins state, in Mitsubishi Electric's own words: "The CE marking will be removed from the [series] because it does not comply with the EU Cyber Resilience Act (CRA). In conjunction with the removal of the CE marking, the UKCA marking will also be removed. The orders received by Mitsubishi Electric on or after April 2027 will be supplied without the CE marking and UKCA marking." I am reporting that as the vendor's statement. I am not going to argue the legal reading of it, because that is for your compliance people, not for a spares buyer at 6 a.m. on a Monday. What I will tell you is the practical consequence, and it is a procurement consequence, not a technical one. If your plant is in the EU or the UK and your machinery documentation references CE or UKCA marked components, an order placed in April 2027 arrives with a rating plate that looks different from the one in your existing drawings. That is a conversation with your quality manager, your notified body if you have one, and whatever customer audit sits over your head. It is not a conversation you want to have for the first time the week you need the part. The hazard gets worse, and the bulletins say so plainly: both the former and the new versions may be in distribution at the same time depending on stock status. That means a distributor with old marked stock and a distributor with new unmarked stock can both quote you the same catalog number on the same day, and the marking on the plate may differ. If you need a specific marking, you have to specify it on the purchase order and you have to confirm it at goods-in. Assume nothing from a part number alone. For a plant in the Middle East or the Americas, this matters less directly, though it matters if you export machinery into Europe or if your OEM builds to European documentation. And it matters for resale value on anything you later sell into the EU. Know which of your three audiences applies before you decide how much of the April 2027 date is your problem. The MR-EM340GF has one more wrinkle. Mitsubishi states that the recommended replacements, the motion control software SWM-G or the motion control board MR-EM441G, are not compatible with the MR-EM340GF and differ in specification. That is the vendor's language, and it is about as clear as a vendor gets. There is no drop-in. The replacement SDK is SW1DND-EMMSDK-B, and the bulletin references Mitsubishi document No. 26-21E, titled "Change in Packaging Labels and Rating Plates due to Enforcement of EU Cyber Resilience Act." If you run an MR-EM340GF, the part is not a stock buy. It is a migration project with its own engineering hours, and it belongs in the capital plan, not the consumables budget.   What this costs, and what it does not   Let me separate the fear from the bill, because the two are not the same number. The fear is that September 30 means your PowerFlex 70 fleet is orphaned and every failure is a line-down event. That is not what the announcement says. Sales end is not support end. Rockwell stops selling new units in three enclosure families. It has not announced a repair cut-off. Your existing drives keep running, and the installed base is large enough that the third-party repair and remanufacturing market will serve it for years. For legacy drive spares, the practical question is not whether parts exist. It is who has them and what condition they are in. The fear on the Mitsubishi side is the same, and the bill is smaller than you think. The repair window runs to March 31, 2036. That is seven years past production end. Most plants will have migrated off HF-KN and MR-JN long before that, because the replacement lines exist and are documented: HF-KN053(B) to HK-KT053W(B), HF-KN13(B) to HK-KT13W(B), HF-KN23(B) to HK-KT23W(B), HF-KN43(B) to HK-KT43W(B), HF-KN73(B) to HK-KT7M3W(B) on the motor side, and MR-JN-10A to MR-J5-10A, MR-JN-10A1 to MR-J5-10A, MR-JN-20A to MR-J5-20A, MR-JN-20A1 to MR-J5-20A, MR-JN-40A to MR-J5-40A on the amplifier side. If your architecture runs an MR-JN-10A1 or MR-JN-20A1 on a 1-phase 100 V AC input, note that Mitsubishi requires the MR-CM08K1 simple converter in combination. That is a line item people miss when they quote the swap. The Mitsubishi bulletins do add one honest caveat worth reading twice: repairs may not be possible once component inventory is exhausted. So the repair window is a maximum, not a guarantee. Anyone telling you 2036 is a certainty has not read past the date line. That leaves the real variable, and it is not the vendor. It is the secondary market. Once orders close, every part you buy comes off someone's shelf. Condition and provenance become the deciding factors, and those do not appear on a data sheet. A remanufactured MR-JN that has been tested on a load bank with a documented repair record is a different asset than a unit pulled from a decommissioned panel and wiped down. Both can be quoted at the same price. Only one belongs on a critical line. The way you control that is by knowing your supplier's testing procedure before the emergency, by keeping serial numbers on your own units so you can trace what came back from repair, and by deciding in advance which of your lines can tolerate a used unit and which cannot. For a plant, servo drive parts and drive spares bought from the secondary market are a legitimate, budget-saving route, provided the supplier tests, documents and warrants. The moment a supplier cannot describe their incoming inspection, the price stops being the point.   The decisions this forces this quarter   You have a short list and a long list. Do the short list in the next ninety days, before the March 2027 date starts pulling lead times. 1. Run a spares shelf audit against actual installed units. Walk the floor, pull every nameplate, and record catalog number, serial number, enclosure type and date of installation. This is the step everyone skips and it is the step that saves the most money. You will find drives on the shelf that belong to lines you decommissioned in 2019, and you will find lines running hardware that nobody has a spare for. 2. Build a clean catalog number list, one line per distinct number, with quantity installed and quantity on the shelf. On the PowerFlex 70 side, sort the list by enclosure family and check each number against Rockwell's Product Lifecycle Status tool rather than assuming the family status applies. On the Mitsubishi side, separate the three groups: HF-KN, MR-JN, MR-EM340GF. 3. Set a stock buy date tied to the real deadline, not the scary one. For Rockwell that date is now, for the IP20, IP54 and Flange units you still run. For Mitsubishi the hard date is March 31, 2028 for order acceptance, and the made-to-order transition on March 31, 2027 is the date by which lead times start hurting you. Anything you intend to buy new, buy before March 2027 or accept the lead time. 4. Start the migration case for the capital plan, not for the emergency. A migration case is a one-page document: how many units, what the replacement part number is, what changes in the panel, what the engineering hours look like, and what the plant loses if it does not do it. For MR-EM340GF that document is mandatory because the replacement is explicitly not compatible. For PowerFlex 70, Rockwell recommends the PowerFlex 750 family, so the case is a modernization, and modernizations go in next year's budget, not this month's emergency account. 5. Brief your quality or compliance contact on the April 2027 marking change, before it becomes a goods-in problem. It is a fifteen-minute conversation today and a stranded shipment later. That list is small on purpose. You are not going to migrate forty drives this quarter, and you should not try. You are going to know what you have, buy the parts that go away first, and put the rest on a documented track.   Where a spare parts supplier fits   I am not going to pretend a parts supplier solves a lifecycle problem. We do not. What a good supplier does is narrower and more useful: it holds stock on the numbers you are about to lose, it can tell you which units it has tested and how, and it does not disappear during the week you actually need the part. What you should ask any supplier quoting you on discontinued hardware, and ask before you need them: Do they test, and on what? A load test on a drive and a run test on a servo are minimums. Ask for the procedure. Can they trace provenance on a used or remanufactured unit? If they cannot describe where a part came from, treat the unit as unknown condition. Do they hold buffer stock on the numbers you care about, or do they source to order? Sourcing to order means your lead time equals their lead time, which is exactly the problem you were trying to solve. Will they confirm the marking on the rating plate for Mitsubishi parts after April 2027? If you need CE or UKCA, this is the question that separates a supplier who understands the change from one who has not read the bulletin. The honest answer for most plants is a mixed model. Critical lines get new or vendor-repaired units. Non-critical lines and low-frequency spares come from tested secondary stock. Automation spares sourced that way, with documented testing, beat an unplanned line-down every time. Get that mix decided before the calendar decides it for you.   FAQ   Does September 30, 2026 mean my PowerFlex 70 drives stop working or stop being supported?   No. It means Rockwell stops selling new units in the IP20/NEMA UL Type 1, IP54/NEMA UL Type 12 and Flange families. Your installed drives keep running. Rockwell's published material documents the sales end and points users to its Product Lifecycle Status tool for individual catalog numbers. It does not publish a repair or support cut-off for this drive, and you should not accept a supplier's claim that it does.   My plant runs NEMA 4X PowerFlex 70 units. Am I affected?   Not by this date. Rockwell states the NEMA 4X versions remain in Active Mature status. Active Mature is not the same as Active, so keep it on your watch list, but you are not in the September 30 stop.   What is the actual last date I can buy HF-KN motors and MR-JN amplifiers?   Per the Mitsubishi bulletins, order acceptance runs through March 31, 2028, and last shipment and production end is March 31, 2029. The made-to-order transition happens earlier, on March 31, 2027, so treat March 2027 as the date lead times start working against you.   Is the April 2027 CE marking change a reason to stop using these parts?   That is a decision for your compliance function, not for the parts buyer. Mitsubishi Electric states that CE marking will be removed because the series does not comply with the EU Cyber Resilience Act, that UKCA marking will also be removed, and that orders received on or after April 2027 will be supplied without either marking. What it means for your machinery documentation depends on your market and your customer requirements.   Can I buy the replacement part and drop it in?   On the motor and amplifier side, the replacement lines are documented (HF-KN to HK-KT, MR-JN to MR-J5), and there are combination notes, including the MR-CM08K1 simple converter for certain 1-phase 100 V AC configurations. On the MR-EM340GF, Mitsubishi states plainly that the recommended replacements are not compatible and differ in specification. That one is an engineering project, not a part swap.   How long should I plan on the repair route?   Mitsubishi bulletins set repair acceptance and parts supply through March 31, 2036 for HF-KN, MR-JN and the MR-EM340GF, with the caution that repairs may not be possible once component inventory is exhausted. Plan the repair route as a bridge, not as a permanent strategy.   The FX3G situation is already past its last order date. What does that tell me?   It tells you the notices are accurate and the dates are real. Mitsubishi announced the FX3G discontinuation in June 2025, set a December 2026 discontinuation date and a June 2026 last order date, with the FX5UJ as successor. The last order date passed. If you were not watching, you are now buying FX3G hardware from stock. That is exactly the position you do not want to be in with HF-KN and MR-JN in 2028.   Should I buy a lifetime supply of spares now?   Rarely. Shelf stock costs money, ties up capital and can degrade. Buy what your failure history says you consume over the made-to-order window plus a buffer, not a lifetime. The exception is a part with no compatible replacement, like the MR-EM340GF, where the case for holding stock is stronger because the alternative is a redesign. SEO fields - do not paste this block into the website Meta Title PowerFlex 70 Sales End September 30: Legacy Deadlines Meta Description Rockwell PowerFlex 70 sales stop September 30, 2026, and Mitsubishi HF-KN and MR-JN run out in 2029. The deadlines, the CE marking change and what to stock now. URL Slug powerflex-70-sales-end-legacy-automation-deadlines-2026 Category news Publishing steps: 1) paste Meta Title + Meta Description into the Yoast SEO fields, 2) set the URL slug, 3) pick the category, 4) request indexing in Search Console. Site: tztechio.com

September 28,2026

Holiday Notice: Mid-Autumn Festival and National Day 2026 Dear customers, Our office and warehouse in Xiamen will be closed for two public holidays:   Mid-Autumn Festival: September 25 to September 27, 2026 National Day: October 1 to October 7, 2026 We are back at work on Thursday, October 8, 2026. September 28 to 30 are normal working days. Orders. Orders confirmed and paid by September 30 will be dispatched before the National Day closure, subject to stock. Orders received during the closures are processed in the order they arrive, starting October 8. If you need parts in October, send your enquiry before September 30 so we can confirm stock and shipping in time. Support. Email and WhatsApp messages are still received during the holiday, but replies will be slower. For urgent breakdowns, send the part number and fault details to WhatsApp +86 175 5077 6091 and we will come back to you as soon as we reopen.   Thank you for your continued business. We wish you a good week.     Xiamen Zantang Technology Co., Ltd Tel / WhatsApp: +86 175 5077 6091 Email: info@tztechautomation.com tztechio.com    

September 24,2026

  *Industry news | September 14, 2026* The date that matters to you is not 11 December 2027. It is 11 September 2026, three days ago. On that Friday, the reporting duties under the EU Cyber Resilience Act started to apply, and they apply to the equipment already bolted into your cabinets, not to the equipment you might buy next year. If a controller, HMI or drive on your floor has a data connection and someone actively exploits a vulnerability in it, the company that made it now has 24 hours to raise an early warning and 72 hours to file a full notification with an EU platform. Nobody sent you a letter about it. Your spare parts shelf and your capital plan were both built around the December 2027 date, because that is the date everyone quoted, and that date is still real for the CE marking side of things. It just is not the date that governs reporting. Those two dates now sit three months and roughly fifteen months apart, and the gap between them is where a plant gets caught out. Here is the operational version. Reporting started. Your installed base is inside it. Your patch path for most of that installed base probably does not exist. Those three sentences together are the whole story, and the rest of this piece is about what a maintenance or plant manager can actually do with them.   What changed on September 11   The Cyber Resilience Act is Regulation (EU) 2024/2847. It entered into force on 10 December 2024. Article 71(2) is the article that sets out when the pieces switch on, and it does so in stages rather than all at once. Chapter IV, the part about notified bodies, has applied since 11 June 2026. Article 14, the reporting duty, applies from 11 September 2026. The main conformity regime, meaning the Annex I essential requirements, the technical documentation and the CE marking, applies from 11 December 2027. On the Friday just gone, what became live is the obligation on manufacturers to report. From 11 September 2026, a manufacturer must report actively exploited vulnerabilities and severe incidents affecting the security of products with digital elements. This is the Article 14 reporting duty. In plain terms: if a connected product the manufacturer sells is being exploited in the wild, the manufacturer has to tell the authorities, and it has to do it fast. Fast means measurable. Once the manufacturer becomes aware, it files an early warning within 24 hours, a full notification within 72 hours, and a final report no later than 14 days after a corrective measure is available for an actively exploited vulnerability, or within one month of the 72-hour notification for a severe incident. Reporting is done once, through the CRA Single Reporting Platform, or SRP, which was established by ENISA under Article 16 and was operational on 11 September 2026. The notification goes to the CSIRT of the manufacturer's main establishment and gets shared with ENISA and the other national CSIRTs. ENISA published the list of CSIRT coordinators covering all 27 member states on 4 September 2026, which tells you the plumbing was being finished right up to the deadline. There is one detail circulating that gets misread, and it matters because it separates a public-relations problem from a legal one. Commission Delegated Regulation (EU) 2026/881, adopted on 11 December 2025, lets a receiving CSIRT delay onward dissemination on cybersecurity grounds. Read that carefully. It restricts who sees a report. It does not change when it is filed. Nothing in that delegated regulation pauses the manufacturer's own filing clock. If you are talking to a vendor who suggests the clock might be paused, the clock is not paused. Then there is Article 14(8), which adds a separate duty to inform impacted users. If the manufacturer does not do that in time, the notified CSIRTs may inform users themselves where they judge it proportionate and necessary. That is the part to keep in your back pocket during a warranty conversation, because it is the regulation's own answer to the question of who eventually tells the plant. If the maker stays quiet too long, the regulator can go around them.   Why legacy equipment is caught   The assumption I hear most often, and I have heard it in almost those words from people who should know better, is that anything sold before December 2027 is grandfathered and therefore irrelevant to this regulation. That is half true, and the half that is false is the expensive half. Article 69(2) grandfathers products placed on the market before 11 December 2027 unless they undergo a substantial modification after that date. So an untouched machine, sitting there doing its job with the original firmware and no significant changes, is outside the CE marking and Annex I essential requirements. Fine. For an unmodified line you are not going to get dragged into a conformity project, and nobody is going to ask you to re-document a panel that was built in 2011. Article 69(3) overrides that specifically for reporting. The Article 14 duty applies to all in-scope products with digital elements placed on the market before 11 December 2027, with no age cutoff and no carve-out for old equipment. Sit with the phrase "no age cutoff" for a second. It does not say five years, it does not say ten years, and it does not say "still supported." It says all in-scope products placed on the market before that date. Now combine that with the scope rule. Products with digital elements whose intended or reasonably foreseeable use includes a direct or indirect logical or physical data connection to a device or network are in scope. Your plant runs Siemens, Rockwell, Mitsubishi and a long tail of other brands, mostly as controllers, HMIs, drives and I/O blocks. The ones that sit on Ethernet or on any fieldbus that reaches the plant network, or that anyone can plausibly connect a laptop to, tick that box. There are explicit exclusions, and they are narrow and specific: medical devices, vehicle type approval, civil aviation products, marine equipment, products developed exclusively for national security or defence, and identical spare parts. A packaging line controller is not on that list. So here is what that means for a plant running controllers and HMIs installed between 2005 and 2020. You are not the obligated party. The manufacturer is. But the equipment you own is the subject of the obligation, and it is subject to it without any age limit. A thin-client panel PC commissioned in 2008 is as much a candidate for a reported exploited vulnerability as a unit shipped last month, as far as the reporting duty is concerned. The item on your floor has not changed. What changed is that the company that made it now has a legal clock attached to what happens to it. There is a practical consequence buried in that, and it is the one I would want my own team to understand. In a reporting regime with a 24-hour clock, the manufacturer's willingness and ability to know what is happening in old equipment becomes a compliance asset. Vendors who cannot see into a decade-old build are not off the hook. They have to report what they know. But the flow of information runs through their support and security channels, and if your plant is not in that stream, you will be the last to know.   The part that should worry a plant manager more than the reporting   Reporting is somebody else's deadline. The patch question is yours, and it is worse. Think about the sequence. A vulnerability in an out-of-support controller gets actively exploited. The manufacturer becomes aware, and it starts a 24-hour clock and then a 72-hour clock. Under Article 14 it has to say what it knows. What it does not have to do is fix the thing. The Commission has been explicit that a manufacturer may genuinely be unable to investigate a vulnerability in a legacy product, because build environments cannot be recreated, dependencies are unavailable, or the people who knew the code have left. In that case the manufacturer must still notify, but is not required by the regulation to meet the other obligations, such as vulnerability handling. The duty is to report what is known, not to reopen a decade-old build. Read that as an operations manager and the sentence has teeth. An out-of-support controller will not get a patch. It may well get reported. So there is now a defined path where the world learns that your line is exploitable, the maker files the paperwork on time, and nothing physically changes at your plant. The report is a disclosure event, not a remedy. That reframes the support period from a nice-to-have procurement line into a security date, and it shifts where the risk sits. Under Article 13(8), a manufacturer must define a support period for vulnerability handling, at least five years or the expected in-use period if shorter, starting when the product is placed on the market, and must disclose the support period end date at the point of purchase in at least month and year format. Under Article 13(9), each security update must remain available for at least 10 years after issuance, or for the remainder of the support period, whichever is longer, and security updates must be supplied free of charge, with a narrow exception for tailor-made business products. A minimum support period of five years is a floor set by regulation, and it is short. Five years from a purchase order in 2026 lands in 2031. Five years from a 2019 order landed in 2024, which is already behind us. I have panels in this plant older than some of the people who now run them. So the stated support end date, in month and year, is now the closest thing you have to an expiry label on a control asset, and it belongs in the asset register next to the model number, not in a supplier's marketing deck. The migration argument follows from that without any engineering. If the support end date is the last date you can expect a vulnerability fix, and the reporting regime reaches back before 2027 with no age cutoff, then an end-of-support date is a security date. A dated security date is a procurement problem. It is not an IT problem, because IT does not buy the controller, does not stock the spare, and does not own the line when the migration window closes. It sits on the maintenance manager's desk, and eventually on the capital committee's desk.   What the plant has to do   None of this requires you to become a compliance officer. It requires four pieces of work that a manager can assign to named people, and one of them can start this week. Build an asset list of connected control equipment, with the support period end date next to each item. The word "connected" is doing real work there. Anything on the plant network, anything on a fieldbus that bridges to it, anything an integrator or a vendor laptop plausibly touches. If you have a CMMS or an asset register that already holds the tag numbers, extend it rather than starting a parallel spreadsheet, because a second list will rot. Ask suppliers and distributors for the declared support end date in writing. For anything bought from now on, treat that date as a specification alongside I/O count and price. Nobody negotiates a controller without knowing how many digital inputs it has. From this quarter, the support end date belongs in the same conversation, in months and years, in writing, on the quote if you can get it. A verbal answer is not a support statement. Put every automation vendor's security advisory channel on someone's job description. Not a shared inbox nobody reads, and not a personal email that leaves when the person does. A named role, with a backup, and a bench check that both people can log in. Vendor advisories are how you find out that a vulnerability has moved from theoretical to actively exploited, and that is the trigger point for everything in this article. Sort the equipment into what can still be patched and what cannot. That split is not a compliance exercise. It is your migration queue. Everything inside a support period goes on the maintain list. Everything outside it goes on the migration list, ordered by how much the line costs you when it stops. That is the whole triage. You do not need a risk model to draw that line; you need the support dates from step two and a view on criticality, which you already carry in your head. Fold 11 December 2027 into the capital plan rather than treating it as an IT matter. That is the date the full conformity regime applies, and it is also the deadline that has been quoted at you for two years, which is exactly why it is the wrong date to plan around alone. If your migration queue only gets funded when the 2027 date starts to bite, you will be executing migrations under time pressure with no negotiation room on lead times or price. Funding the queue a year early is cheaper than funding it in the last quarter of 2027, and the security argument is now the reason to do it rather than a slogan. There is one more thing that belongs in a manager's head, and it is about relationships rather than process. Your route to a support statement is usually the distributor or reseller who sold you the module, or the integrator who specified it. Those people have the manufacturer relationship you do not, and they can get a written answer where your general inquiry goes nowhere. This is a reason to keep a small number of parts and support relationships in decent condition rather than chasing the cheapest quote on every single transaction. When you need a support end date in writing, a documentation response or a last-time-buy allocation, you find out quickly who actually wants your business.   Timetable   Date | What applies 11 June 2026 | Chapter IV of the regulation, on notified bodies, applies. 11 September 2026 | Article 14 reporting duties apply. Manufacturers must report actively exploited vulnerabilities and severe incidents affecting the security of products with digital elements. Once a manufacturer becomes aware | The reporting clock runs: early warning within 24 hours, full notification within 72 hours, and a final report no later than 14 days after a corrective measure is available for an actively exploited vulnerability, or within one month of the 72-hour notification for a severe incident. 11 December 2027 | The full manufacturer conformity regime applies, including Annex I essential requirements, technical documentation and CE marking. Two notes on that table. First, the reporting row in the middle is not a one-off event. It is a standing obligation that fires each time a manufacturer becomes aware of an actively exploited vulnerability or a severe incident, which is why the clock, not the date, is the thing to keep in view. Second, the two dated rows are not the same project. The September 2026 row is about disclosure and it is already running. The December 2027 row is about conformity and it is what your capital plan has to absorb.   Penalties, briefly   Article 64 sets the money. For breaching the Annex I essential requirements or the Article 13 and Article 14 obligations, administrative fines run up to EUR 15 million or, for an undertaking, up to 2.5 percent of total worldwide annual turnover for the preceding financial year, whichever is higher. Other obligations carry up to EUR 10 million or 2 percent. Supplying incorrect, incomplete or misleading information to notified bodies or market surveillance authorities carries up to EUR 5 million or 1 percent. Member States set the rules on penalties, so the practical enforcement shape will vary by country. That is the manufacturer's exposure. It is not yours, and I would not invent a number for what it costs a plant, because it does not. What it does do is explain behavior. A supplier facing a potential EUR 15 million fine, or 2.5 percent of group turnover, has a hard reason to build a real vulnerability process: intake channels, triage, an ability to say something within 24 hours. It also has a hard reason to be careful about what it writes down. Expect documentation requests around supported life, update availability and known-issue status to get more serious over the next eighteen months, and expect a vague verbal answer to be replaced by a templated written one. That is a good outcome for a buyer, on balance, because a manufacturer that has to measure something tends to manage it. The buyer's job is to ask the question in a form that produces a dated answer rather than a brochure. I would not read too much else into the penalty numbers. Nobody is being fined today, and I am not aware of any enforcement wave arriving before the conformity regime starts in 2027. The reason to care now is that the September date is live and the incentive it creates reaches into how much information you get about your own plant. The spare parts angle Identical spare parts are excluded from the scope of the regulation. That exclusion is real and it is specific, and it is relevant to what a parts supplier sells. If you buy an identical replacement module to swap into a cabinet that already has that module, the spare part itself is not a product with digital elements under this regulation. Do not turn that into a compliance answer, because the controller those parts go into is not excluded. The exclusion carves out the spare part; it does not carve out the device the part sits in. A plant cannot retire its exposure by buying one more identical module for the shelf. When that controller becomes the subject of a report because a vulnerability in it is being actively exploited, the spare part exclusion does nothing for you at all. The real answer is one of two things. Either a support period that runs long enough that you never hit the wall, or a migration plan with a date on it. There is no third option that involves a purchase order and a shelf. That said, the spares side and the migration side are not in conflict, and a manager should not treat them as such. Keeping critical systems running while migration is scheduled is the whole job, and an out-of-support line usually needs stocked spares precisely because it cannot be patched. Think it through from the operations side. If a controller is end-of-support, there is no firmware fix coming, so a failure of that controller is a hardware event you have to answer with hardware. The migration may be eighteen months out in the capital plan, and the line still has to run every shift in between. That means the last-time-buy question, the stock level and the sourcing route for that controller become more important, not less, in the period between the security date passing and the migration date arriving. A spare on the shelf is not a fix for the vulnerability. It is how you survive the exposure long enough to migrate away from it, which is a different and legitimate thing. There is a further reason to keep the parts route warm, and it is not about the regulation at all. A migration program takes years to execute in a plant of any size. In that window, brownfield lines break the way brownfield lines always break, at the worst possible moment, on the shift with the least cover. The plant that has a documented parts route and a sensible stock position moves through its migration without the two problems colliding.   This week's checklist   1. Start the connected asset list. Get tag numbers, manufacturer, model and a rough install year for every controller, HMI, drive and I/O block that reaches the plant network. Assign one owner and set a date to have the first pass done. The support end date column can be blank for now. 2. Send a written support period question to the top five vendors or distributors by installed base. Ask for the support period end date for each product family, in month and year, in writing. Note the date you asked and put a follow-up on the calendar. 3. Name the person who owns vendor security advisories, and the backup. Confirm both can actually access the channels for your main automation brands. If not, get the subscriptions set up this week. 4. From now on, treat the support end date as a specification on new purchases, alongside I/O count and price. Add it to the standard inquiry template so it stops being an optional question. 5. Run the first triage pass: mark everything inside its support period as maintain, everything outside as migrate, and rank the migrate list by what a stop costs the line. You now have a migration queue with a security rationale behind it. 6. Put 11 December 2027 in the capital plan as a dated line item with a budget number attached, not a note. The conformity regime starts that day and it is the deadline that funds your migration queue. 7. Note the September 2026 fact in whatever risk log or plant review you run. Reporting is live, it reaches back with no age cutoff, and it is the reason your support end dates now get read as security dates. None of this is paperwork for its own sake. It is the same set of facts a maintenance manager would want anyway, which is what the equipment is, how long it will be supported, what it costs to keep running, and when it gets replaced. The regulation just put dates on the columns. If you need a fast read on what is still serviceable and what should be heading for the migration queue, the controller and drive pages at tztechio.com/plc and tztechio.com/industrial-automation are a reasonable place to check part availability against your asset list, and the brand pages for Siemens and Allen-Bradley cover most of what sits in a European legacy panel. Start with the asset list. The rest follows from it. -------------------------------------------------------------------------------------------- 🏢 About TZ Tech   TZ Tech is a leading supplier of industrial automation, electrical, instrumentation, and telecommunications components. We specialize in sourcing ready-to-ship distributor stock, allowing us to offer highly competitive pricing and short lead times. Thanks to our extensive inventory, we can even source rare and discontinued parts that are hard to find elsewhere.   🛡️ Our Quality Commitment   We understand that quality is your top priority. Every component undergoes a strict screening and inspection process so you can buy with absolute confidence. For legacy or discontinued parts, we believe in complete transparency and will always provide an honest, accurate report on the product's condition. Plus, all brand-new parts come backed by a full 1-year warranty.   ✉️ Get in Touch     Have a project or a part you need? Send us your inquiry today! Our team is dedicated to providing a fast response within 6 hours (excluding weekends).  

September 14,2026

*Industry news | September 7, 2026* Mitsubishi Electric has put an end date on the MELSEC-F FX3U, the micro PLC that has run a large share of the world's small machines since it launched in 2004. In Technical Bulletin FAM-A-0116-B, "Production Discontinuation of the MELSEC-F FX3U/FX3UC Series Models and Extension Models," issued in April 2026 and updated as Version B in May 2026, Mitsubishi sets out the formal wind-down: make-to-order production begins January 1, 2028, order acceptance closes March 31, 2028, production stops September 30, 2028, and repair support continues until September 30, 2035. The official bulletin PDF is public on Mitsubishi's technical news page. The schedule matters well beyond the PLC industry's usual end-of-life notices. The FX3U is not a niche controller. Since 2004 it has been the default brain inside packaging lines, labeling machines, small water-treatment skids, HVAC panels, gate and valve controls, and OEM-built machines that ship to every market on earth, and it became the flagship of the MELSEC-F micro family. When a platform with that installed base gets a final order date, the news reaches past Mitsubishi's catalog into the spares strategy of every plant that still runs one. The practical summary for buyers and maintenance engineers: normal catalog ordering of FX3U and FX3UC products continues through 2027. From January 1, 2028, Mitsubishi builds these models only to order, and the last day a new order can be accepted is March 31, 2028. Units ordered in that window will be produced in the months that follow, and new production ends September 30, 2028. Repair support then runs seven years, to September 30, 2035, with one exception: the FX3U-128ASL-M, developed jointly with Anywire Corporation, gets only about one year of repair support after production stops.   Why the FX3U phase-out is a bigger event than a typical end-of-life notice   Mitsubishi's stated reason is the one that ends most long-lived electronics: the supply chain ran down. The bulletin says production parts for the target models have been discontinued one after another, making it hard to maintain production, and that the manuals for the target models will be discontinued as well. The FX3U announcement is the latest step in a broader contraction of legacy MELSEC hardware. Through 2026 Mitsubishi set a September 30, 2026 final order deadline for the high-speed Universal Q CPUs under bulletin FA-A-0418-A and moved the MELSEC-L series to make-to-order under bulletin FA-A-0466-A. The pattern: Mitsubishi is concentrating development and production on current platforms, chiefly the MELSEC iQ-F (FX5) series for the small-machine class, giving each legacy family a defined, published end rather than an open-ended availability promise. None of this makes the FX3U a bad controller or existing machines obsolete next year. It puts a clock on new supply, and the window to buy new units and stock spares for machines that will run into the 2030s is finite and published. For anyone who maintains FX3U equipment, the time to plan is now.   The phase-out schedule at a glance   Milestone | Date | What it means in practice Make-to-order production starts | January 1, 2028 | Mitsubishi stops building FX3U/FX3UC and related extension products for stock. Orders continue to be accepted, but production runs to order. Last date for order acceptance | March 31, 2028 | Final day Mitsubishi will accept new orders for the affected models. After this date, no new orders can be placed through normal channels. Production discontinuation | September 30, 2028 | Manufacturing ends. Orders placed before March 31 are produced in the intervening months; after this date there is no new production at all. End of repair support, standard models | September 30, 2035 | Seven years after production discontinuation. Repair requests are handled within this window, subject to repair parts availability. End of repair support, FX3U-128ASL-M | About September 30, 2029 | The AS-Link master module co-developed with Anywire Corporation gets roughly one year of repair support after production stops, about six years less than the rest of the family. The arithmetic from the date of this article makes the urgency concrete. As of September 7, 2026, roughly nineteen months of order acceptance remain: about fifteen months of normal catalog ordering through the end of 2027, followed by the three-month make-to-order window from January 1 to March 31, 2028. Orders placed by March 31 will be built in the six months between the order deadline and the September 30 production stop. For repair planning, the same arithmetic gives a nine-year repair horizon from today for standard models, ending September 30, 2035. The FX3U-128ASL-M is the outlier, with support ending around September 30, 2029, roughly three years from now; any site running that module should treat it as the first item in a spares review. The precise answer to "when is the FX3U discontinued": order acceptance ends March 31, 2028, and new production ends September 30, 2028; everything after that runs on inventory, old stock, and the repair channel.   Which FX3U and FX3UC models are affected   The bulletin's coverage is broad, a sign of how deep the FX3U ecosystem ran. It covers FX3U main units across their power-supply variants, FX3UC connector-style main units, the FX2N and FX3U and FX3UC extension hardware that bolted onto those CPUs, the programming and communication accessories, and the memory cassettes many sites use for program storage.   FX3U main units   Mitsubishi groups the affected FX3U main units by power supply type; the example model numbers below come from its own replacement guidance. Affected group | Suffix types covered | Example models named in the bulletin FX3U AC-powered main units | /ES types | FX3U-32MR/ES, FX3U-64MT/ES, FX3U-80MR/ES FX3U DC-powered main units | /DS and /DSS types | Covered by the bulletin's model lists FX3U UA1 types | UA1 variants | Covered by the bulletin's model lists The /ES AC-powered family is the one most readers will recognize, since it accounts for the majority of FX3U machines wired to AC mains. The /DS and /DSS DC-powered variants cover installations that feed the controller from a 24 V DC bus, and the UA1 types are additional voltage variants Mitsubishi sold into specific markets; all are folded into the same discontinuation. The example models show the shape of the affected range. FX3U-32MR/ES is the 32-point relay-output main unit that became the default controller for small machines; FX3U-64MT/ES is its 64-point transistor-output sibling for higher-speed applications; and FX3U-80MR/ES shows the bulletin reaching into the larger main units. Picking these three as worked examples signals that the phase-out covers the mainstream of the family, not just the edges.   FX3UC main units   The FX3UC is the connector-style compact version of the family, built for machines where panel space is tight and connector wiring replaces screw terminals. The bulletin covers the DC-powered FX3UC main units across their point sizes and suffix variants. Affected group | Point sizes | Suffix types covered | Example model named in the bulletin FX3UC DC-powered main units | 16, 32, 64, and 96 point | /D, /DS-T, /DSS, and /DS-TS types | FX3UC-64MT/D FX3UC machines are common in compact OEM equipment where the connector format keeps wiring tidy and replacement fast. The FX3UC-64MT/D example anchors the replacement guidance: Mitsubishi maps it directly to the FX5UC-64MT/D, preserving the connector-style form factor in the migration path.   Extension units, cables, converters, and memory cassettes   The affected list goes beyond main units. Because the FX3U system grew by extension, the bulletin covers the extension hardware many installations carry, and it reaches back into the FX2N catalog: FX3U installations frequently mix FX2N extension units with FX3U main units on the same DIN rail, which is why the extension list includes FX2N 32-point-class I/O units among other items. Equipment class | What is affected | Notes I/O extension units and extension models | FX2N, FX3U, and FX3UC I/O extension units and extension models | Includes FX2N 32-point-class I/O units; covers the modules used to grow systems beyond the main unit's onboard I/O Programming and communication cables | FX-422CAB-150 and related cables | The cable family used to link PCs and programming tools to FX-series PLCs Signal converters | FX-USB-AW | The USB signal converter used in PC-to-PLC programming and diagnostic setups Memory cassettes | FX3U-FLROM-64L and FX3U-FLROM-1M | Mitsubishi lists no alternative models for these cassettes The memory cassette item deserves special attention. The FX3U-FLROM-64L and FX3U-FLROM-1M are the flash memory cassettes that plug into FX3U main units for program storage and ROM operation, and the bulletin states that there are no alternative models for them, a stronger statement than a usual replacement note. Sites that use cassettes as their program backup method, or that run programs from ROM mode, will need an engineering answer rather than a purchasing one, because no direct successor exists. For maintenance teams, the practical consequence is simple. The FX-422CAB-150 cable and the FX-USB-AW converter are what an engineer reaches for when a laptop has to talk to an FX3U for programming or diagnostics, and when they leave the catalog the everyday service loop changes. Buy the spares needed before the order deadline; the last-order clock applies to cables and converters just as it does to CPUs.   FX5U replacement: the bulletin's recommended migration path   Mitsubishi's recommended alternatives are the MELSEC iQ-F series: FX5U, FX5UJ, and FX5UC, a recommendation it has made consistently for years. The iQ-F line is the stated successor platform for the small-machine class, and it moves engineering onto GX Works3 software rather than the GX Works2 environment that FX3U users know. The example mappings below are the bulletin's most useful guidance for engineers planning a changeover. Existing FX3U model | Direct FX5 replacement | Alternative configuration in the bulletin FX3U-32MR/ES | FX5U-32MR/ES | FX5UJ-40MR/ES FX3U-64MT/ES | FX5U-64MT/ES | FX5UJ-60MT/ES plus FX5-8EYT/ES FX3U-80MR/ES | FX5U-80MR/ES | Not listed FX3UC-64MT/D | FX5UC-64MT/D | Not listed /MS sink-triac output variants | /MT transistor units plus external relay adapter boards | FA-TH16YSR20S relay adapter, with FA-CBL10(Y)M20 accessories in some mappings Read the table the way a design engineer would. The cleanest migrations are the direct ones: an FX3U-32MR/ES becomes an FX5U-32MR/ES with the same point count, relay output type, and model-number shape. The FX5UJ alternatives are the economy track: FX5UJ is the slim, cost-oriented entry point of the iQ-F line, and the bulletin shows a 40-point FX5UJ-40MR/ES absorbing a 32-point FX3U application. For the 64-point case, it pairs a 60-point FX5UJ-60MT/ES with an FX5-8EYT/ES eight-point extension to reach the needed I/O count. The /MS mapping is the one that surprises people. The /MS sink-triac output variants were used where outputs had to switch AC loads without relay wear. In the iQ-F world, the bulletin routes those applications to /MT transistor-output units plus external relay adapter hardware, specifically the FA-TH16YSR20S relay adapter board with FA-CBL10(Y)M20 accessories in some mappings. In plain terms, the AC-switching function moves out of the PLC onto an external relay board driven by standard transistor outputs, and specifying that change means checking load currents, switching frequency, and panel space, because the external board occupies real estate the old triac outputs did not. Three engineering realities sit behind every row of the mapping table. First, the iQ-F platform programs in GX Works3, not GX Works2; ladder written for the FX3U does not load into an FX5 as a finished program, and programs that lean on FX3U-specific modules or cassette-based ROM operation need the most rework. Second, the FX5U family adds capabilities the FX3U never had, including built-in Ethernet on the FX5U main units. Third, the FX5UC keeps the connector-style form factor for FX3UC users, so the mounting philosophy survives even where the program does not. For a Mitsubishi PLC parts buyer, migration is a project with an engineering budget, not a swap order, which is why starting before the 2028 deadline matters.   What the phase-out means for plants running FX3U today   The FX3U installed base will not disappear in 2028. Machines built on FX3U controllers will keep running for a decade or more after production stops, because PLCs are retired when the machine is retired, not when the catalog changes. What the bulletin changes is the supply equation around those machines. Consider the timeline from the plant floor. A machine bought new in 2026 has a repair-support horizon to September 30, 2035, roughly nine years from today, and an older machine has the same horizon. The bottleneck is not the repair window; it is new supply. After March 31, 2028, the only sources of new FX3U units are distributor inventory that has not sold, and after September 30, 2028, the only sources of any FX3U hardware are old stock, the used market, and the repair channel. None of those scale to cover a plant that finds a failed CPU in 2030 with no spares on the shelf. That is the trap for the unprepared. The FX3U will not fail all at once in 2028; it will fail one machine at a time through the 2030s: a power supply after a surge, an output module after a shorted actuator, a CPU after years of heat cycling. Each failure is a trivial fix while new parts are available and a scheduling problem when they are not. The phase-out converts a routine spares question into a finite procurement window that closes March 31, 2028. For OEMs the stakes are different but just as real. Any OEM still designing new machines around the FX3U in 2026 and 2027 is committing those machines to a platform whose order book closes in under two years. A machine program launched in 2027 with a five-year production run will need FX3U hardware until 2032, four years after Mitsubishi stops building it. That math requires either a large upfront spares buy or a mid-run platform change, the most expensive kind. The bulletin is effectively a design-freeze notice for OEMs: new-machine designs should move to the iQ-F line before the window closes, while existing FX3U-based products need a documented last-buy plan.   Keep running or migrate: work the decision machine by machine   The phase-out does not force a single answer; it forces a decision, machine by machine. The keep-running case is strong for equipment that is mechanically sound and far from the end of its useful life: it needs a spares plan, not a new controller. The migration case is strong for machines that are already retrofit candidates, need new connectivity, or have outgrown the FX3U. Factor | Lean keep running | Lean migrate to FX5 Machine remaining life | Five or more years of sound mechanical life | Machine already near retrofit or replacement Spares position | Clean path to stock CPU, I/O, and accessories before March 31, 2028 | No budget or appetite for a last-buy stock position Program complexity | Simple ladder, well documented, easily re-created if lost | Heavy use of FX3U-specific modules, cassettes, or AS-Link hardware Connectivity needs | Standalone machine, current communications are enough | Plant now demands Ethernet, data collection, or remote access Engineering resources | No in-house GX Works3 experience and no migration budget | Staff ready to move to GX Works3 and revalidate programs FLROM cassette use | None, or program stored only as a file backup | ROM-cassette operation is part of the design, and there is no cassette alternative The keep-running economics deserve a fair hearing. Stocking a spare CPU, a spare I/O module, and the accessories a machine needs is inexpensive relative to a migration, which carries engineering time, program conversion, revalidation, panel rewiring, documentation updates, and the risk of startup problems on a machine that was running fine before. For a healthy machine with simple control requirements, the last-buy spares route is frequently the lower-cost, lower-risk answer. The catch is that the last-buy has a deadline: March 31, 2028. The migration case wins when the machine is already a retrofit candidate. If the plant is re-engineering the panel anyway, putting in an FX5U instead of buying a last FX3U costs little extra and resets the clock: the FX5 platform is current, its production is not under an end-of-life notice, and its repair horizon starts fresh. Migration also wins on capability, since FX5U main units bring built-in Ethernet and the iQ-F line is where Mitsubishi's development effort is concentrated. A plant that standardizes on iQ-F ends up with one engineering environment, GX Works3, instead of maintaining GX Works2 skills solely for an aging fleet. The wrong answer is doing nothing until 2028 and letting the calendar make the decision. A plant that waits until January 2028 to start an audit will be placing last-buy orders in the same quarter it is trying to evaluate migrations, with order books tightening and no time for a second round of purchasing. Every machine in the fleet should carry an explicit keep-running or migrate label, with a date attached, before the make-to-order period begins.   The repair reality: a seven-year tail with limits   Mitsubishi's repair support runs to September 30, 2035 for standard FX3U and FX3UC models, seven years after production discontinuation. That commitment is the reason the phase-out is not an immediate obsolescence event. But a repair-support window is not a guarantee that every failed board can be fixed on demand. Repairs depend on the availability of repair parts, and the bulletin's own rationale, components discontinued one after another, is the constraint that governs the repair channel as well. As the years pass, repair will get slower and less certain for the oldest modules, worth factoring into spares math rather than assuming effortless coverage for seven full years. The FX3U-128ASL-M exception is the sharpest detail in the bulletin and the easiest to miss. The module, an AS-Link master unit developed jointly with Anywire Corporation, receives only about one year of repair support after production stops, placing its cutoff around September 30, 2029. A co-developed product depends on a partner's supply and service commitments, which run on a different clock than Mitsubishi's own hardware. Any site with FX3U-128ASL-M modules in service should treat them as the highest-priority spares item in the audit: one spare bought before the order deadline costs far less than a network master failure in 2031 with no repair path and no successor on the shelf. There is also a documentation angle. The bulletin states that the manuals for the target models will be discontinued. Wiring guides, programming references, and special-function module documentation should be downloaded and archived before they disappear from Mitsubishi's documentation servers; archived manuals are cheap insurance, and the one asset in this phase-out that does not expire if you grab it now.   What to do this week   The bulletin rewards early action, and most of the actions are cheap today. A practical checklist for any plant or OEM running FX3U or FX3UC hardware: 1. Pull the installed-base list. Count every FX3U and FX3UC CPU, note the exact model and suffix (which determines AC or DC power, output type, and point count), and flag machines with FX2N extension units, FLROM cassettes, or an FX3U-128ASL-M. 2. Match your models against bulletin FAM-A-0116-B. The PDF is public; confirm every module you run is in scope, including extension units and accessories that are easy to forget. 3. Label every machine keep running or migrate, with an owner and a date, before the end of 2026. 4. Build the spares list for keep-running machines now: CPUs, I/O modules, power-related spares, an FX-422CAB-150 cable, an FX-USB-AW converter, and anything else the service kit needs. Price it while normal ordering is open. 5. Treat the FX3U-128ASL-M as urgent. Its repair support ends around September 30, 2029, about six years before the rest of the family, and a spare should be on order well before March 31, 2028. 6. Resolve the FLROM question. The FX3U-FLROM-64L and FX3U-FLROM-1M have no alternative models, so cassette-based program storage is an engineering task that should start now, not during a 2028 changeover. 7. Archive the manuals for every affected model before they are discontinued, and store them where the maintenance team can find them. 8. Put the dates on the calendar: normal ordering through 2027, make-to-order January 1 to March 31, 2028, production stop September 30, 2028. Set internal procurement deadlines at least a quarter ahead of each date. 9. For OEMs, freeze new-machine designs on FX5 and give existing FX3U-based products a documented last-buy plan with quantities and a date. 10. Run one migration pilot. Convert a simple machine's program to GX Works3 on an FX5U or FX5UJ and validate the bulletin's mapping against real I/O counts and panel constraints. A pilot done in 2027 makes the 2028 deadline a non-event.   The bottom line for buyers   Mitsubishi has given the market something it does not always get: a precise, published end date with years of warning, and a coherent replacement path in the iQ-F series. The takeaways are clean. New FX3U and FX3UC hardware remains available through normal ordering for the rest of 2026 and all of 2027. The make-to-order window from January 1 to March 31, 2028 is the final opportunity to place orders. Production ends September 30, 2028, and repair support continues to September 30, 2035, with the FX3U-128ASL-M exception ending around September 30, 2029. For plants that act this year, the phase-out is manageable: spares can be stocked, manuals archived, keep-running machines secured, and migration candidates moved to FX5U or FX5UJ replacements on an engineering schedule rather than an emergency one. For plants that wait, the same phase-out becomes a scramble of last-minute orders in a three-month window, rising reliance on old stock, and machines whose repair path narrows year by year. The FX3U earned its place by being dependable, inexpensive, and everywhere. The phase-out does not change that record; it changes the procurement rules, and the new rules have a deadline. The work starts with an audit and a calendar. The PLC spares categories and industrial automation parts sections at tztechio.com track this transition, and the Mitsubishi page is a practical starting point for locating FX3U stock and FX5 alternatives while both remain available through normal supply channels. URL Slug: mitsubishi-fx3u-fx3uc-phase-out-news ---------------------------------------------------------------------------------------------- 🏢 About TZ Tech   TZ Tech is a leading supplier of industrial automation, electrical, instrumentation, and telecommunications components. We specialize in sourcing ready-to-ship distributor stock, allowing us to offer highly competitive pricing and short lead times. Thanks to our extensive inventory, we can even source rare and discontinued parts that are hard to find elsewhere.   🛡️ Our Quality Commitment   We understand that quality is your top priority. Every component undergoes a strict screening and inspection process so you can buy with absolute confidence. For legacy or discontinued parts, we believe in complete transparency and will always provide an honest, accurate report on the product's condition. Plus, all brand-new parts come backed by a full 1-year warranty.   ✉️ Get in Touch     Have a project or a part you need? Send us your inquiry today! Our team is dedicated to providing a fast response within 6 hours (excluding weekends).  

September 10,2026