Cross-referencing is the discipline of establishing whether one part can replace another. Done well it keeps obsolete machines running for decades; done carelessly it swaps a working part for one that fits but does not function. The difference lies in how rigorously the comparison is made.
What must match
- Function — the job the part performs, not its shape.
- Ratings — voltage, current, pressure, torque, load and duty cycle.
- Materials and sealing — compatibility with the medium and the environment.
- Interfaces — mounting dimensions, shaft or bore, flange, thread and electrical connection.
- Performance characteristics — response time, accuracy, backlash, flow coefficient.
The order to work in
- Original manufacturer supersession. The maker may publish a direct replacement; always check first, it is the lowest-risk answer.
- Same family, next generation. Often dimensionally identical with improved ratings.
- Different maker, same standard. Where a part conforms to a recognised standard, an equivalent from another brand is usually acceptable.
- Functional equivalent. The most work and the most risk — requires documented comparison and often testing.
Why it goes wrong
- Matching dimensions and ignoring ratings.
- Assuming a suffix is decorative — it usually encodes coil voltage, contact configuration or version.
- Accepting a cross-reference from a database without verifying it against the manufacturer.
- Failing to record the substitution, so the next person repeats the exercise.
Keeping the record
Every qualified cross-reference should be written down: original reference, replacement reference, what was compared, what differs, who approved it and when. That record turns a one-off decision into an asset — and it is the difference between a plant that learns and one that keeps solving the same problem.
The equivalence confidence ladder
Not every cross-reference deserves the same level of trust, and treating them as equal is how a plant ends up running production on a guess. Classify each equivalence you establish, and record the class with it. The class tells the next person how much verification to do before they rely on it.
| Level | Evidence behind it | Safe for | Verification before use |
|---|---|---|---|
| A — Manufacturer confirmed | Written statement from the original maker, or a supersession notice naming the replacement | Direct substitution in production equipment | Confirm the notice covers your variant and date range |
| B — Same maker, different catalogue | Both numbers appear in the maker’s own catalogue or datasheet with identical specification tables | Direct substitution | Check interface dimensions, not just the specification table |
| C — Dimensional and functional match | Measured against a reference sample; function and ratings compared item by item | Non-safety, non-critical applications | First-article inspection plus a monitored trial |
| D — Catalogue similarity | The numbers look alike and the descriptions match | Emergency hold-over only, with a documented risk acceptance | Full dimensional and functional verification |
| E — Marketplace claim | A listing or broker states the parts are interchangeable | Nothing. This is a lead, not an equivalence | Everything |
A record that says “equivalent” without a level is worse than no record, because it transfers false confidence to whoever reads it next. A record that says “Level C, measured 2024-03 by J. Tan, trial running on Line 3” tells the next person exactly what remains to be done.
Cross-referencing a part you cannot measure
The ladder above assumes you have something physical to compare. Often you do not — the part failed, went out with the scrap, or is buried inside an assembly that cannot be opened without a shutdown.
Three routes still produce a defensible answer:
- Work from the interface side. The mating parts outlive the consumable. Measure the bore, the shaft, the flange bolt pattern, the port thread, the terminal spacing. A part that matches every interface and every rating is a strong candidate even without the original in hand.
- Work from the drawing of the assembly. Assembly drawings survive in machine documentation far more often than component drawings. A position number plus a bill of materials entry frequently identifies the component uniquely.
- Work from a sister machine. If another unit of the same model still carries the original part, that part is your reference. Photograph every marking on it before it is disturbed.
When none of these is available, the honest output is not an equivalence — it is a shortlist of candidates, each with the specific measurements still required to confirm it.
A worked cross-reference, end to end
The process is easier to follow with a concrete example. A pressure transmitter fails on a 2009 packaging line; the maker discontinued it in 2018; the enquiry arrives with only a photograph of the nameplate.
- Establish the true identity. The nameplate gives a maker, a type code and a serial. The type code decodes to a family, a measuring range and an output signal. Range and output are the two attributes that must not change.
- Check the supersession chain first. The maker’s own catalogue shows a direct successor two generations later. That is a Level A equivalence, and it ends the search: the successor is bought from a distributor.
- Only if that fails, widen out. Look for equivalents from other makers with the same process connection, range, output and wetted materials. This is Level C at best and needs a first-article check.
- Decide what to buy. One successor unit for the failed position, plus a decision on whether to standardise the fleet on the successor or hold the old type as long as spares exist.
Note the order: original maker first, because it is the only route that can reach Level A. Searching the wider market first costs more and produces weaker evidence.
Questions we are asked about cross-referencing
Can a part with a higher rating replace a lower-rated one? Usually yes for a passive part — a higher voltage or pressure class rarely hurts. It is not automatically true for anything with a control function, where a different rating often means a different response characteristic. Compare the characteristic curve, not the headline number.
Is a superseded number always safe? Almost always, but check two things: whether the successor changed its interface, and whether it is itself already superseded. A three-generation chain is worth resolving to the current number before ordering.
How long should a cross-reference record be kept? As long as the equipment runs. The cost of the record is one line in a database; the cost of losing it is repeating the whole exercise during a breakdown.
References we have recently quoted in this category
Need a specific reference? Send the exact number to [email protected] with quantity and target date — we reply with sourcing options, equivalent/replacement candidates, condition and lead time.
Related reading
- Obsolete PLC, Sensor & Motor Parts: A Cross-Reference Workflow
- MRO Spare Parts Management
- Spare Parts Management: Principles and Practice