Electronic Component Cross Reference: 7 Checks Before You Accept an Equivalent

An electronic component cross reference is a claim that part B can replace part A in your design. Cross-reference tools build that claim from catalogue overlap: same package, same pinout, same polarity, equal or better voltage and current ratings. What they do not see is your operating point, your qualification requirement, the lifecycle status of the proposed part or the exact orderable code on the reel. Those four gaps are where substitutions go wrong, and seven checks close them. Below, the checks are run against a real case from our catalogue: three SOT-23 MOSFETs offered for one BSS138BK line, only one of which works in the circuit.

What a cross-reference tool checks, and what it skips

A cross-reference tool checks the parameters it can read from a datasheet header: package and pin assignment, polarity, maximum drain-source voltage, maximum current and sometimes on-resistance at one headline condition. It then grades the result. The grades vary by tool, but they map onto three levels buyers should keep separate:

• Drop-in replacement - same footprint, same pinout, electrically equal or better at every condition your design uses. No layout change, and in most quality systems no engineering re-test beyond a first-article check.
• Pin-to-pin compatible - it solders onto the same pads with the same pin functions, but at least one electrical parameter differs. It needs an engineer to confirm the difference does not matter in this circuit.
• Functional equivalent - it does the same job, possibly in another package or with other pin assignments. That means a schematic or layout change and full requalification.

The trap is that tools score against the maximum ratings table, while circuits fail on the characteristics table. A MOSFET rated 60 V and 300 mA looks equivalent to one rated 60 V and 360 mA. Whether it switches fully at the gate voltage your board actually applies is a separate line in the datasheet, and no tool knows which gate voltage your board applies.

One BSS138BK line, three offers: the case

Take a BOM line of a kind we quote regularly for industrial boards in Germany: Nexperia BSS138BK in a bidirectional I2C level shifter, where the gate sits on a 2.5 V rail from an FPGA I/O bank and the bus on the other side runs at 5 V. With the Nexperia part short, two alternatives came up in cross-reference searches: ROHM's BSS138BKT116 (same base number, different manufacturer) and PANJIT's 2N7002, of which we hold 450,000 pieces of 2N7002_R1_00001 in Germany, factory sealed with date code 22+. All three are 60 V N-channel parts in SOT-23 with the same gate, source and drain pinout. A tool rates all three as matches.

Here is what the manufacturers' own documents say. Nexperia figures come from the BSS138BK datasheet, ROHM figures from ROHM's BSS138BK product page, and PANJIT figures from PANJIT's 2N7002K datasheet, the closest variant PANJIT publishes.

The 2N7002 is the instructive failure. Its threshold can sit anywhere up to 2.5 V, and threshold is defined at just 250 µA of drain current. A gate held at 2.5 V can therefore leave a worst-case part barely conducting, and I2C edges on the 5 V side stall or never pull low. On a bench with one sample it may well work, because a typical part sits near the middle of the range. The fifth board off the line, built with a unit at the top of the range, is the one that fails. At a 5 V or 10 V gate the 2N7002 is a perfectly good substitute, which is exactly why "bss138 vs 2n7002" produces contradictory answers online: both camps are right for their own gate voltage.

ROHM's part passes electrically, and on paper it even beats the Nexperia part. It fails on a check no electrical comparison catches: ROHM lists BSS138BKT116 as Not Recommended for New Designs. For a board that will be built for years, that turns today's shortage fix into tomorrow's NRND problem.

Seven checks that catch a false match

Run these seven checks on every proposed cross, in this order. The first two eliminate most bad matches in minutes, and the last five decide whether a match that works electrically can be released to production.

1. Guaranteed performance at your operating point. Find the conditions your circuit imposes: gate or supply voltage, load current, temperature. Then confirm the candidate has a guaranteed limit at those conditions, not just a headline rating. No spec at your condition counts as no guarantee.
2. Min and max values, never typical. ROHM publishes typical on-resistance on its product page, Nexperia's datasheet leads with maximums. Put ROHM's 0.58 Ω typical next to Nexperia's 2.2 Ω maximum and the ROHM part looks four times better. Typical against typical (0.58 vs 1.1 Ω) it is about twice as good. Compare like with like, and design to the limit.
3. Qualification grade. If the original carries AEC-Q101 or AEC-Q100, the cross must carry it too, on the exact ordering code. For automotive customers a change of qualified source usually triggers a PPAP resubmission as well. Our AEC-Q100 vs AEC-Q101 guide covers which grade applies to which part type.
4. Lifecycle status at the manufacturer. Check the manufacturer's own product page on the day you decide. Aggregator and distributor records can lag a status change by months, and a cross that is already NRND only buys time.
5. The full orderable code. Suffixes carry packaging, temperature grade, plating and sometimes qualification. BSS138BKT116 is a 3,000-piece reel with a 3,000-piece minimum order. The same base number with a different suffix can be a different product. Our part number suffix decoder walks through the patterns. RoHS and REACH declarations also attach to the orderable code, not the base number.
6. Robustness and thermal limits. ESD rating, power dissipation and thermal resistance decide field returns, not first power-up. Here the crosses rate 2 kV against the original's 1.5 kV, which is fine. The reverse would need a look at how exposed the pin is to connectors.
7. Marking and incoming inspection. Nexperia marks BSS138BK as %SB, where % is a site code. A cross arrives with a different top marking, and an incoming inspector or AOI program set up for the original will flag it. Update the inspection record at the same time as the approved vendor list, or a legitimate part gets quarantined as a suspected counterfeit.

The gotcha: the same part number from two manufacturers is two parts

A shared base number such as BSS138 means both parts descend from the same industry-standard device. It does not mean they share a datasheet. In our case the two "BSS138BK" parts differ in ESD rating (1.5 kV vs 2 kV), current rating (360 vs 400 mA), in whether the product page lists AEC-Q101, and in lifecycle status. Buyers treat a matching number as the safest possible cross, and it is the one that most often skips engineering review. Treat a same-number part from another manufacturer exactly like any other pin-to-pin candidate: run all seven checks, and record the result.

Who approves a cross, and what goes in the file

An engineer approves the cross; purchasing only proposes it. In most European quality systems built on ISO 9001 or IATF 16949, a component change that is not drop-in needs engineering sign-off, and an EMS building to a customer's BOM needs the customer's written deviation before it fits anything not listed. The approval file should hold five things so that the next buyer, auditor or field-repair team can trust it without redoing the work:

• the original and the approved cross as full orderable codes, with manufacturer names;
• the datasheet revision of each part that was compared;
• the comparison against the seven checks, with the operating conditions written down (in our case: "gate 2.5 V, bus 5 V, pull-up current under 3 mA");
• the scope of the approval: this assembly only, or every design using the original;
• the approver's name and date, plus any first-article test results.

The scope line matters more than it looks. The 2N7002 would be approvable as a BSS138BK cross on a relay driver with a 10 V gate on the same board, and wrong on the level shifter next to it. Approvals written as "BSS138BK = 2N7002" without conditions are how a good cross on one line becomes a field failure on another, and why maintenance teams searching for "approved cross-reference parts" often find lists nobody can vouch for.

When a cross is the wrong answer

A cross is the right answer when the original is unavailable for longer than your build schedule can absorb and a candidate passes all seven checks. It is the wrong answer in three situations. First, when the original is still in production and the shortage is temporary: genuine original stock from the independent channel, with traceability, avoids requalification entirely. Second, when the product is automotive, medical or aerospace and the qualification cost of any change exceeds the price premium on original parts. Third, when the only candidates are functional equivalents that need a layout change, at which point you are doing a redesign and should plan it as one. Our guides on second sourcing and sourcing obsolete components cover those two paths.

Most real BOM lines land in between: the original for the build that ships this quarter, and a checked cross approved in parallel for the builds after that. If you have a line where the original is short and the crosses on offer do not add up, send us the part number and the conditions it runs at. GlobX will quote the original where we can find traceable stock in Europe and name the alternates we would put through the seven checks, so your engineer starts from a shortlist rather than a search result.