Part NumbersCloningUsed ModulesCompatibility

Module Part Numbers Explained: Hardware vs Software vs Supplier Numbers, and Which Ones Must Match

Auto Module Lab Technical Team·ALOA-MAL Certified · 15+ Years ECU + Key ProgrammingJuly 28, 2026·16 min read

Why a module with the same part number can still be wrong

Two control modules can be physically identical, plug into the same connector, and still be completely incompatible. That is not a quirk of one brand. It is a direct consequence of how modern control units are built: a piece of hardware and a piece of software that are versioned separately, sold separately, and updated on separate schedules.

The person who discovers this the hard way is usually somebody who did everything right by the standard used-parts logic. They matched the number on the label. They matched the model year. They bought from a car that looked identical in the yard. Then the unit went in and the vehicle would not run, or ran badly, or logged a page of configuration faults that were not there before.

Donor mismatch is the single most common reason a used-module job cannot proceed, and almost every case of it is decided before the part is ever bought. This guide is about the numbers themselves - what each one means, which ones have to agree, and how to read them off a listing before your money leaves your account.

This is deliberately a technical companion piece rather than a repeat of two guides we already have. If your question is about the seller, the physical condition of the unit, and how to avoid buying junk, read what to check before you buy a used module donor. If your question is the more fundamental one - why plugging in a matching used module does not simply work - the pillar on used ECU swap versus cloning covers the identity and immobilizer side of it.

The three numbers on a typical module label

Pull almost any control unit and the label will carry several strings. They are not interchangeable, and knowing which is which is most of the skill.

The hardware part number

This is the manufacturer's number for the physical assembly: the board, the processor family, the memory devices, the driver stages, the connector, the housing. It is the number that decides whether a unit can physically do the job - whether it has the right number of injector drivers, the right connector pinout, the right transceiver for the bus your car uses.

The hardware number is usually the most prominent one on the label and is often the one people quote when they say "part number." It is generally the manufacturer's own numbering scheme rather than the supplier's, and it is what a parts catalogue will return when you look up your vehicle.

The software, calibration, or program number

This identifies the program running on the hardware. Depending on brand it is called a software number, a calibration number, a flash number, a datenstand, or a program level. It changes when the manufacturer issues an update, and it can be different on two cars that rolled off the line a month apart with identical hardware.

Crucially, this number often exists in two places at once: printed on the label as it left the factory, and stored in the module's memory as it currently is. Those two can disagree, because a dealer reflash updates the memory and nobody peels the label off and prints a new one. The number in memory is the truth. The number on the label is history. That single fact explains a large share of confused listings.

The supplier or manufacturing number

This belongs to the company that actually built the box - Bosch, Continental, Denso, Delphi, Siemens, Marelli, Hitachi and the rest. It typically identifies the internal hardware platform, sometimes with a batch or date code alongside it.

The supplier number is genuinely useful, because it often reveals hardware family relationships that the manufacturer's own numbering obscures. Two units with different manufacturer part numbers can share a supplier platform, and that similarity is sometimes what makes a transfer possible. It is also where you find the giveaway that a listing is wrong: a supplier number from a different hardware generation is a hard stop no matter how well the manufacturer number matches.

There are usually further strings too - a hardware revision or index, a serial number, a manufacturing date code, and sometimes a checksum or programming stamp. The revision is the one that matters most, and it gets its own section below.

Hardware versus software: which one has to match

Here is the rule that resolves most donor questions.

For a clone, the donor's hardware must match. The software is what gets transferred.

A clone works by taking the data that makes your module yours - identity, immobilizer content, calibration, adaptation values, configuration - and writing it onto another physical unit. For that to succeed, the destination hardware has to be able to run that data. It needs the right processor family, the right memory layout, the right peripheral hardware. If it does, the software difference is irrelevant, because the donor's software is being replaced anyway.

This is why a common and initially surprising answer is correct: a donor with the "wrong" calibration is usually fine. Its calibration is not the thing being kept. What matters is that the box underneath can accept yours.

The inverse is the trap. A donor with a matching software number but different hardware is not usable, and it is exactly the kind of listing that looks convincing. Sellers frequently quote whichever number is easiest to read, and a software number quoted as if it were the part number will match a great many cars that are not compatible.

The services that live on this rule are straightforward flat-rate bench jobs. GM E38, E67, and E92 ECM cloning at $250 works on exactly this basis: the controller family is confirmed, then your data goes onto the donor. Ford PCM cloning and VIN transfer at $250 is the same idea on the Ford side. Mercedes ME9.7 and MED17 ECU cloning at $250 covers the two big Mercedes engine-controller families where donor hardware compatibility is the whole gate.

Hardware revisions and why direction matters

Within a single hardware part number there are usually revisions - sometimes called an index, a version, an edition, or just a letter or digit appended to the number. A revision is a change to the physical unit that was not significant enough to earn a new part number: a component substitution, a board layout tweak, a memory device swapped for a newer equivalent, a supplier change.

Revisions are where a lot of the real-world friction lives, and the asymmetry is the thing to remember:

A later hardware revision will sometimes accept an earlier calibration. The reverse frequently fails.

The reason is ordinary engineering practice. Later hardware is generally designed to remain backward-compatible so that existing programs keep working and the factory does not have to change its process. Earlier hardware was designed before the later software existed and knows nothing about it. If the newer calibration expects a memory region, a peripheral, or a capability the older board does not have, it will not run - and depending on the platform it may not even be accepted for writing.

That asymmetry has a practical consequence when you shop: if you cannot get an exact revision match, a slightly later donor is usually the safer bet than a slightly earlier one. It is not a guarantee, and on some platforms revision matching is strict in both directions, but as a default it is the right way to lean.

Supersessions versus different families

Two things look similar on a listing and are not remotely the same.

A superseded part number is usually fine. Manufacturers replace part numbers all the time, most often because a component in the assembly changed, a supplier changed, or a running production improvement was made. A supersession is an explicit statement by the manufacturer that the new number replaces the old one for service purposes. If your number supersedes to the donor's number, or both supersede to a common later number, you are generally looking at a compatible unit.

A different family is not fine, no matter how close the number looks. Numbers in a series are not a continuum. A single digit can be the difference between a controller with a different processor, different memory, or a different bus architecture. Manufacturers assign numbers for administrative convenience, not to signal compatibility, and adjacency in a catalogue means nothing about interchangeability.

The honest way to tell them apart is not to eyeball the numbers. It is to check the supersession chain in a parts catalogue and to confirm the supplier hardware platform. When those two agree, you have real evidence. When you only have the manufacturer number and a hunch, you have a coin flip.

Region, market, and option variants

The same vehicle built for different markets frequently carries different modules, and the differences are not cosmetic.

  • Emissions and fuel calibration vary by market. A controller built for one regulatory region carries a calibration developed for that region's fuel, aftertreatment hardware, and test cycle. Standardized on-board diagnostics have applied to light vehicles sold in the United States since the Environmental Protection Agency required OBD-II beginning with the 1996 model year, but the calibration behind that standard interface is still market-specific.
  • Lighting and body configuration vary by market, which matters enormously on body control modules and lighting modules.
  • Option content changes module configuration even within one market. Tow packages, upgraded audio, driver assistance, keyless entry, remote start, transmission variants - all of it shows up as configuration data, and sometimes as a different part number entirely.
  • Transmission and drivetrain pairing is one of the strictest. A controller calibrated for one transmission or final drive is not a candidate for a vehicle with another.

The safety-related side of that list is not optional content. NHTSA administers the federal motor vehicle safety standards that govern how restraint, braking, and lighting systems must behave, and organisations such as the Insurance Institute for Highway Safety publish the testing that drives much of that hardware. A module whose configuration does not match the safety hardware actually fitted to a vehicle is not a cosmetic mismatch - it is a system that may not behave the way it was certified to behave.

This is a large part of why module electronics have grown so much. NHTSA has documented that a mainstream vehicle now typically carries well over 50 networked control units, and Car and Driver has reported that electronics and software account for roughly 40% of the value of a new vehicle. More content means more variants, and more variants means more ways for a donor to be almost right.

The comparison table: which numbers must match

Number on the label What it identifies Must it match for a clone? What goes wrong if it does not
Hardware part number The physical assembly - board, processor, memory, drivers, connector Yes, or a documented supersession of it Write refused, or a unit that accepts data and then will not run
Hardware revision / index A running change within one hardware number Preferably; later donor is safer than earlier Data written but faults on start, or write rejected outright
Software / calibration number The program currently running on the module No - this is what gets transferred Nothing; the donor calibration is replaced by yours
Supplier / manufacturer number The builder's own hardware platform identifier Yes in practice - it is the best family check A hidden family mismatch that the part number did not reveal
Market or region variant The regulatory market the unit was built for Yes for calibration-critical modules Emissions, lighting, or diagnostic behaviour that does not match the car
Option or configuration code The optional equipment the unit was configured for Not for the clone itself, but it must be reconfigured Missing or non-functional features, configuration faults
Serial / date code The individual unit and when it was built No Nothing - informational only

What actually happens when the numbers do not match

There are three distinct failure outcomes, and they are worth knowing because they have very different consequences for your wallet.

1. The write is refused

The best outcome. The programming tool reads the destination unit, compares what it finds against what is being written, and declines. Nothing has changed on the donor, nothing is damaged, and you have learned the donor is wrong for the cost of the shipping.

Most competent bench processes are designed to produce this outcome rather than a heroic attempt. A refusal is a safety feature, not a failure of tooling.

2. It flashes, then throws faults

The frustrating middle case. The data is accepted, the module boots, and it communicates - but the system does not behave. You get configuration faults, communication faults with modules that were previously happy, features that do not work, or intermittent behaviour that comes and goes with temperature.

This is what a hardware revision mismatch or a subtle family difference typically produces. The processor runs the code well enough to start, but a peripheral is not where the code expects it, or a calibration references a capability the board does not have. These cases eat diagnostic time because the fault list points everywhere except the real cause.

3. The unit is bricked

The worst case, and the reason bench work is done on a bench. A write that is interrupted, mismatched at a low level, or performed on a unit that cannot support it can leave a module that no longer boots and no longer answers on the bus - no communication, no recovery through the diagnostic connector.

Some bricked units are recoverable at the chip level. Some are not. Either way the donor is now a project rather than a part.

Two things reduce this risk enormously, and both are why in-vehicle attempts go wrong more often than bench attempts. First, stable power. A regulated bench supply does not sag the way a car with a tired battery does when a cooling fan cycles mid-write. Second, verification before the write, which means reading and archiving the destination unit's original content first so there is something to fall back to.

"I turn away more donors on the numbers than I do on condition. A customer sends me a beautiful clean unit, no corrosion, perfect connector - and it is one hardware revision back from what his car needs. If he had sent me a photo of the label before he clicked buy, that would have been a thirty-second conversation instead of a wasted purchase and two weeks of downtime." — Independent powertrain module technician, 19+ years (anonymized)

How to read a donor listing before you buy

Most listings show you enough to decide, if you know what you are looking at. Work through this in order.

  1. Demand a photo of the actual label, not a stock image. If the seller will not photograph the unit they are selling, the listing tells you nothing. This is the single highest-value question you can ask.
  2. Read every string on the label, not just the biggest one. Write down the manufacturer part number, the supplier number, the revision or index, and any software or calibration number.
  3. Identify which is which before comparing. A software number matched against your hardware number is not a match, it is a coincidence.
  4. Check the supersession chain. Confirm whether your number and the donor's number supersede to a common later number. Same chain is strong evidence. Different chains is a stop.
  5. Confirm the source vehicle. Year, model, engine, transmission, drive type, and market. On body and lighting modules, option content too.
  6. Compare revisions and lean later. If an exact revision match is not available, a later revision is generally the safer choice than an earlier one.
  7. Ask what the seller knows about its history. Was it removed from a running vehicle? Was it a warranty return? Has it been reflashed or tuned? A unit pulled from a running car is worth more than one from a shelf with no story.
  8. Confirm the return policy in writing before you pay. Even a careful buy can be wrong, and a donor you can send back is a donor with a floor under its cost.

Questions worth asking a seller verbatim: was this removed from a complete vehicle; does it power up and communicate; what is the full string on the supplier label; is there a revision letter or index after the part number; and can you photograph the connector pins.

The reason to be this careful is that these are not $30 parts. Vehicle valuation services such as Kelley Blue Book and J.D. Power publish used-vehicle pricing that shows how quickly electronics-heavy repairs erode the economics of an older vehicle, and Consumer Reports reliability survey work consistently places in-car electronics among the most-reported trouble spots across brands. A wrong donor is not just a wasted purchase; on an older car it can be the thing that pushes a repairable vehicle into the not-worth-fixing column.

When we can confirm compatibility from the numbers, and when we cannot

Being straight about this matters more than sounding capable.

On many platforms, compatibility can be confirmed from the numbers before you buy. Where the hardware family, the supersession behaviour, and the revision rules are well documented and consistent, sending a photo of your original module label and a photo of the donor listing label is enough for a yes or a no. That conversation costs nothing and it happens before your money is committed. It is the single most useful thing this article can convince you to do.

Some cases genuinely cannot be answered from a photograph. They include modules where the label understates what is inside, units that have been previously reflashed or modified so the label no longer describes the software, low-volume or grey-market variants without a clear supersession history, and units where the only way to know what is on the board is to power it and read it.

Those cases go to a bench evaluation. Bench evaluation at $150 per hour is the honest answer to an unknown: the unit is powered on a regulated supply, its actual content is read where it can be read, and you get a factual report on what it is and what can be done with it - rather than a guess dressed up as a quote.

If the module in question is a BMW, the decision often folds into a broader repair-or-replace question, and our guide to BMW module repair versus replacement, module by module covers which BMW modules are usually worth repairing before you go donor hunting at all. On older BMW chassis the related question is whether the engine controller and the security system can simply be brought back into agreement, which is the ground covered in the BMW DME EWS delete bench programming guide - relevant because on BMW EWS cars the controller and the security module carry matched identity data, and a donor with the wrong hardware cannot hold that data correctly no matter what is written to it.

What we need from you to answer the question

To get a compatibility answer without a bench evaluation, message the lab with:

  • The VIN, and the year, model, engine, transmission, and market.
  • A clear photo of your original module label, in focus, with every string readable.
  • A clear photo of the donor label if you already have the unit, or the listing photo if you are still deciding.
  • The symptom and history - what the car is doing, what has already been tried, and whether the module has ever been reflashed or tuned.
  • Whether your original module still powers up and communicates, even if it does not run the car properly.

That last point matters more than most people expect. A module that is functionally dead in a vehicle is very often still perfectly readable on a bench, and if it reads, your vehicle keeps its own identity and configuration instead of being rebuilt from scratch on a stranger's data.

Proof of ownership is required for any work touching keys, immobilizer content, or vehicle security. That is not a formality - it is the line between a legitimate repair shop and one you should not be sending parts to.

Practical shipping notes

Modules ship to one of two addresses depending on the carrier, because a USPS PO Box cannot receive a courier delivery.

USPS: PO Box 120241, Arlington, TX 76012 UPS and FedEx: 1009 Oakwood Ln # 120241, Arlington, TX 76012

Bench work is performed at the Arlington workshop. This is a nationwide mail-in service, and return shipping is selected and paid by you at checkout, starting from $24.95, with faster tiers available.

Frequently asked questions

Do the part numbers have to match exactly for a clone? The hardware part number has to match, or be a documented supersession of yours, and the hardware revision should be as close as possible with a later revision being safer than an earlier one. The software or calibration number does not need to match at all, because that is precisely what gets replaced during the transfer.

Can I use a donor with a different software or calibration number? Yes, in most cases that is completely normal and not a problem. The donor's calibration is overwritten by your module's data during the clone, so its original program level is irrelevant as long as the underlying hardware can run yours. A hardware family or revision mismatch is the real disqualifier, not a software difference.

What is the difference between a superseded part number and a different part number? A supersession is the manufacturer explicitly stating that a new number replaces an old one for service purposes, which normally means the newer unit is a valid substitute. A different number in the same numeric series is not the same thing at all - adjacent numbers can be entirely different hardware families with different processors and memory.

What happens if I install a module with the wrong part number? There are three outcomes. It may simply refuse to accept programming, which costs you nothing but the shipping. It may flash and then produce configuration or communication faults that waste days of diagnostic time. Or a low-level mismatch can brick the unit so it no longer boots or answers on the bus.

Can you tell me if a donor is compatible before I buy it? On many platforms, yes - send a clear photo of your original module label and the donor listing label along with your VIN, and the numbers alone are often enough for a definite answer. Cases where the label no longer describes what is on the board, or where the variant history is unclear, need a bench evaluation at $150 per hour instead of a guess.

Is the number on the label always the software currently running on the module? No, and assuming so causes real confusion. The label records what left the factory, while any dealer reflash updates the memory without changing the printed label, so the two routinely disagree on a car that has had service campaigns applied. The number stored in memory is the accurate one.

Should I worry about which market or region the donor came from? Yes, particularly on engine, body, and lighting modules, because emissions calibration, lighting configuration, and safety-system behaviour are all market-specific even when the diagnostic interface is standardized. A unit from a different market can be physically identical and still be configured for equipment your vehicle does not have.

The bottom line

The numbers on a control module are three or four separate pieces of information wearing similar-looking formats, and treating them as one number is what turns a straightforward used-part repair into a two-week problem. The hardware number and the hardware revision decide whether a transfer is possible. The software number is the thing being transferred, so it does not need to match. The supplier number is your best check on hardware family when the manufacturer's own numbering hides the relationship. Supersessions are generally safe; adjacent numbers in a series are not. Later revisions accept earlier calibrations more often than the other way round.

Flat bench pricing on the clone services in this guide is public: $250 for GM E38, E67, and E92 ECM cloning, $250 for Ford PCM cloning and VIN transfer, and $250 for Mercedes ME9.7 and MED17 ECU cloning. Where the numbers cannot answer the question, bench evaluation at $150 per hour reads the unit and tells you what it actually is. Return shipping is chosen and paid at checkout from $24.95.

If you are looking at a donor listing right now, do not click buy yet. Photograph your own module label, screenshot the seller's label, send both with your VIN, and get an answer first. It is the cheapest step in the entire job and it is the one that decides whether everything after it works.

Ship your module today

Flat-rate pricing, 24-hour bench turnaround, return speed your choice at checkout. Most jobs back on your bench within a week.

More from the Lab