To qualify replacement capacitor candidates you need six steps: an incoming dimensional
check, a reflow profile run against the published limits, a board-flex check, an electrical
measurement set, an environmental stress sequence, and a record that survives an audit.
Walsin Technology (Walsin) publishes enough numbers to write every pass criterion before the
first reel arrives.
Profile and preheat figures throughout from the
Walsin MLCC Application Guide.
What does a qualification plan for a crossed capacitor have to prove?
Three things, in order: that the part physically fits, that your process does not damage it,
and that the circuit still behaves. A cross-reference result gives you a specification match
and nothing more. The plan below turns that match into evidence, or into a documented reason
to stop.
A specification match is a candidate for your evaluation, not an approved substitution. That
is the framing every honest cross tool uses, including Walsin’s own, whose instructions tell
users to double-check the thickness and tolerance of the suggested part. Walsin being direct
about that is the reason a plan like this exists at all.
Venatech Limited, an authorised Walsin distributor in Hong Kong, holds an extract of 24,026
Walsin MLCC part numbers. Every pass criterion below comes from a published Walsin figure or
a measured field in that extract, so QA can check each one against a document.
Step 1: what do you measure before the part reaches the board?
Length, width, thickness and termination, on parts from the actual reel, against the
datasheet band for that exact part number — not against the case size. The dimensional
tolerance band differs by series, and the published thickness is not always fixed by the part
number stem.
Take one incumbent specification — 0603, X7R, 100 nF, ±10%, 50 Vdc — and a design with a
0.95 mm height limit at that position. Our extract returns thirteen Walsin parts on those
five fields, across six series. Five of the candidates fail the height limit at worst case
before any test is run:
| Walsin part number | Series | Published L × W × T band | Worst-case thickness | AEC-Q200 | Match / mismatch vs 0.95 mm limit |
|---|---|---|---|---|---|
0603B104K500 |
General Purpose | 1.60 / 0.80 / 0.80 +0.15/−0.10 mm | 0.95 mm | No | Match — at the limit |
0603B104J500 |
General Purpose | 1.60 / 0.80 / 0.80 +0.15/−0.10 mm | 0.95 mm | No | Match on height, tolerance ±5% |
MT18B104K500 |
Automotive MT | 1.60 / 0.80 / 0.80 +0.15/−0.10 mm | 0.95 mm | Yes | Match — at the limit |
MG18B104K500 |
Automotive MG | 1.60 / 0.80 / 0.80 +0.15/−0.10 mm | 0.95 mm | No | Match on height, no AEC-Q200 |
ST18B104K500 |
Automotive soft term ST | 1.60 / 0.80 / 0.80 ±0.30 mm | 1.10 mm | Yes | Mismatch — 0.15 mm over |
SG18B104K500 |
Soft termination SG | 1.60 / 0.80 / 0.80 ±0.30 mm | 1.10 mm | No | Mismatch — 0.15 mm over |
SH18B104K500 |
Soft termination SH | 1.60 / 0.80 / 0.80 ±0.30 mm | 1.10 mm | No | Mismatch — 0.15 mm over |
Eight of the thirteen carry the +0.15/−0.10 mm band; five carry ±0.30 mm. Same nominal
0.80 mm body, two different worst cases, 0.95 mm against 1.10 mm.
Unique to Venatech’s data. 385 of the 23,601 distinct Walsin MLCC part numbers in our
extract carry more than one published thickness, separated by a thickness symbol column
rather than by the part number stem.0805B104K500publishes 0.80 ± 0.10 mm with thickness
symbol B and 1.25 ± 0.10 mm with thickness symbol D on identical electrical fields. That is
why step 1 measures the reel and not the part number.
Record on the first-article sheet: order code including the CT packaging suffix, lot and
date code, measured L, W and T on the agreed sample, the datasheet band being measured
against, and the thickness symbol printed on the reel label.
Step 2: which reflow limits does the profile have to sit inside?
Walsin’s published MLCC profile, measured on a thermocoupled board carrying the candidate
part, not read off the oven recipe. Ramp up no faster than 3 °C/s. Soak 150–200 °C for
60–120 seconds. Spend 60–150 seconds above 217 °C and at least 30 seconds above 245 °C. Peak
at 260 °C. Ramp down no faster than 6 °C/s.
| Profile stage | Walsin limit | Pass criterion for the report |
|---|---|---|
| Ramp up | 3 °C/s maximum | Measured maximum gradient at the part location |
| Soak, 150–200 °C | 60–120 s | Measured dwell inside the window |
| Above 217 °C | 60–150 s | Measured time above liquidus |
| Above 245 °C | 30 s minimum | Measured time, at the coolest instrumented part |
| Peak | 260 °C | Measured maximum, at the hottest instrumented part |
| Ramp down | 6 °C/s maximum | Measured maximum cooling gradient |
| Start to peak | 480 s maximum | Total elapsed time |
| Reflow passes | Two recommended | Number of oven passes the part will see in production |
Two passes are Walsin’s recommended limit. A double-sided board that reflows both sides has
already spent both on the parts placed first, so plan any rework as iron work rather than a
third oven pass.
Preheat has its own limit and it moves with case size. ΔT between board and solder is held
to 190 °C at 1206 or less and tightens to 130 °C at 1210 or more. If the cross moves the
position across that boundary, the process window changed even though the schematic did not.
Step 3: how do you prove the part survives the board and not just the oven?
Run a board-level mechanical sequence: depanel, connector insertion, in-circuit test probe
force and any assembly screw-down, then inspect and measure. Ceramic cracks from board
strain, and board strain happens after the oven, at steps a thermal profile never tests.
This is where termination construction stops being a paper field. Walsin’s soft-termination
MLCC datasheets are drawn against a 5 mm board bend: SH in Ag-poly, SG in Cu-poly, ST as the
AEC-Q200 automotive line, and VT adding VW80808. A candidate without that construction has to
prove the flex margin another way.
Practical sequence to hand to QA:
- Strain-gauge the candidate position if it sits near a panel edge, connector or boss.
- Depanel using the production method, not a hand snap.
- Run connector insertion and in-circuit test with production fixtures and forces.
- Measure capacitance and insulation resistance before and after the sequence.
- Cross-section a sample from the worst-case position.
Step 4: what electrical measurements belong in the report?
Capacitance, dissipation factor and insulation resistance at the datasheet conditions,
before and after every stress step, plus the in-circuit measurement the part actually exists
to provide. Measure at the datasheet frequency and bias, because a Class II part measured
under the wrong conditions reads wrong twice.
Two conditions catch people out. First, DC bias: a Class II dielectric loses capacitance
under applied voltage, so a bench reading at 0 V does not describe the part in circuit.
Second, the tolerance field. MG18B104J500 and MG18B104K500 write tolerance as -0.05 and
-0.1 rather than as a percentage, so read the datasheet before setting a test limit.
Measure the function too. A decoupling position means measuring the rail; a filter position
means measuring the response; a timing position means measuring the period across the
operating temperature range.
Step 5: what environmental stress does the plan need?
The stress set the part’s grade implies, applied to assembled boards rather than loose parts.
Temperature cycling across the product’s specified range, damp heat at the product’s rated
humidity, and a thermal shock sequence if the incumbent was qualified to an automotive
standard.
Grade is the deciding input, and within one manufacturer it is not uniform. Walsin’s
automotive MLCC splits into MT and ST, both qualified to AEC-Q200, and MG, named in the
catalogue “Automotive Caps Without AEC-Q200 Certification” and holding 788 parts in our
extract. An MG candidate is a real, orderable part that does not bring that certification.
Step 6: what has to be recorded for the result to be auditable?
Enough that someone who was not present can repeat it. That means the orderable part number,
lot and date code, the measured dimensions, the measured profile trace, the test conditions,
the pass criteria and their source document, the sample size, and the disposition with a name
against it.
Traceability is where a distributor helps directly. Venatech supplies from authorised
channels in original manufacturer packaging, with full lot and date code traceability, and a
Certificate of Conformity is available on request. Attach the CoC so the lot tested is the
lot identified.
One field deserves its own line on the form: the order code. Walsin’s MLCC parametric export
omits the CT packaging suffix, so the catalogue key and the orderable code differ by two
characters. 0603B104K500 is ordered as 0603B104K500CT. Record both.
What does the whole plan look like on one page?
Six steps, each with a named input document and a pass criterion. None of it requires
equipment a contract manufacturer does not already have. The only outside inputs are the
datasheet band, the reflow limits and the grade.
| Step | What it tests | Source of the criterion | Evidence produced |
|---|---|---|---|
| 1 | Dimensional fit, L, W, T, termination | Part datasheet, published band | First-article dimensional sheet |
| 2 | Process thermal window | Walsin MLCC application guide | Thermocoupled profile trace |
| 3 | Board strain and flex survival | Board strain standard, plus datasheet bend figure | Before/after electrical, cross-section |
| 4 | Electrical behaviour, bench and in-circuit | Part datasheet test conditions | Measurement set at stated conditions |
| 5 | Environmental stress to grade | Product qualification standard | Stress report with sample size |
| 6 | Traceability and disposition | Lot and date code, CoC | Signed qualification record |
Run steps 1 and 2 before committing to a build. They are cheap, they use published numbers,
and between them they eliminate most candidates that were never going to work.
FAQ
How do you qualify a replacement capacitor from a cross-reference result?
Six steps: measure dimensions against the specific datasheet band, verify the reflow profile on a thermocoupled board, run a board-flex sequence, take electrical measurements at datasheet conditions before and after stress, apply the environmental stress the grade implies, and record lot codes with the disposition.
Can I skip the dimensional check if the case size matches?
No. The dimensional tolerance band differs by series. In the 0603 X7R 100 nF 50 V group, eight Walsin parts publish +0.15/−0.10 mm and five publish ±0.30 mm, giving worst-case thicknesses of 0.95 mm and 1.10 mm from the same 0.80 mm nominal body.
What reflow limits should a qualification profile be checked against?
Walsin’s published MLCC profile: 3 °C/s maximum ramp up, 150–200 °C soak for 60–120 seconds, 60–150 seconds above 217 °C, at least 30 seconds above 245 °C, 260 °C peak, 6 °C/s maximum ramp down, and 480 seconds maximum to peak. Two reflow passes are the recommended limit.
Does the preheat limit change if the crossed part is a different case size?
Yes. Walsin holds ΔT between board and solder to 190 °C at 1206 or less and to 130 °C at 1210 or more. A cross that moves the position across that boundary has changed the process window even though the schematic is unchanged, so the profile has to be re-verified.
Why does the plan measure parts from the reel rather than trusting the part number?
Because 385 of the 23,601 distinct Walsin MLCC part numbers in Venatech’s extract carry more than one published thickness, resolved by a thickness symbol column. 0805B104K500 appears at 0.80 mm and at 1.25 mm on identical electrical fields. Measurement is faster than argument.
Does a soft-termination candidate need a different test in this plan?
The same tests, with the board-flex step treated as the decisive one. Walsin’s SH, SG and ST soft-termination datasheets are drawn against a 5 mm board bend. If the incumbent was a soft-termination part, a candidate without that construction has to demonstrate the flex margin on your board.
What does a Certificate of Conformity add to a qualification record?
It ties the lot you tested to the lot identified in the record. Venatech supplies from authorised channels in original manufacturer packaging with full lot and date code traceability, and a Certificate of Conformity is available on request. Attach it to the qualification file rather than filing it separately.
Is a specification match enough to release the part to production?
No. A specification match is a candidate for your evaluation, not an approved substitution, and no distributor or manufacturer can assess how a part behaves in your circuit. Complete the plan, record the evidence, and let your own change-control process release it.
Next step
Send the BOM line you are crossing and the mechanical constraint at that position. Venatech
Limited, an authorised Walsin distributor in Hong Kong, will return the candidate Walsin part
numbers with the published dimensional band, worst-case thickness, thickness symbol,
termination construction, qualification grade and orderable code, so step 1 is already filled
in when QA receives the plan.
Send your BOM to Venatech · Browse the
Walsin MLCC range · Read
MLCC reflow profile and mounting and
Flex cracking and board bend: why soft termination exists.
Cross it yourself. The cross-reference tool takes a single part number or a whole BOM column and returns the Walsin candidates with every field that differs stated plainly — qualification stays with you. Ready stock shows what is held in Hong Kong today, with an RFQ against any line.