Second Sourcing Without Re-Qualifying: Which Fields Must Match and Which May Not

Second source passive components without a resubmission by knowing which field moves. We checked all 2,176 of Walsin’s automotive AEC-Q200 MLCC rows: 99.5% of positions are unique once size, dielectric, capacitance, voltage and tolerance all match. Loosen tolerance and that drops to 60.8%, with up to six candidates — and three of those six have a capacitance column that contradicts its own part number.

---

Venatech Limited, an authorised Walsin distributor in Hong Kong, writes this from Walsin Technology (Walsin)’s MLCC parametric export, 00_ALL_MLCC_parts_combined.xlsx, 24,026 rows in total. This post isolates the 2,176 rows belonging to the MT series — Walsin’s AEC-Q200 automotive-qualified MLCC — because that is the population a German tier-1 actually works from. Every part number below is real and checkable against that export.

How many automotive MLCC candidates survive when every field must match exactly?

Almost all of them resolve to one. We grouped all 2,176 MT-series rows by size, dielectric, capacitance, rated voltage and tolerance — five fields, an exact match on every one. That gives 2,165 distinct signatures, of which 2,154 (99.5%) land on exactly one Walsin part number, with a maximum of two.

Match rule	Distinct signatures	Resolve to exactly one	Maximum candidates
All 5 fields exact (size, dielectric, capacitance, voltage, tolerance)	2,165	2,154 (99.5%)	2
Tolerance dropped, 4 fields exact	1,328	808 (60.8%)	6

We have not seen this counted anywhere. Inside Walsin’s own AEC-Q200 population, an exact field-for-field match is almost always already a single answer — second sourcing within the same manufacturer’s automotive line rarely needs a judgement call, as long as every field including tolerance is held fixed.

How many candidates appear once tolerance is allowed to move?

Up to six, at one position. Dropping tolerance from the match — the scenario a buyer faces when their incumbent datasheet states a tolerance the Walsin position does not carry — drops the unique-match rate from 99.5% to 60.8%, across 1,328 distinct size, dielectric, capacitance and voltage signatures. The worst case is 0402, X7R, 68 nF, 50 Vdc: six MT-series part numbers, not two.

That gap — 99.5% down to 60.8% — is the entire argument for treating tolerance as a locked field rather than a flexible one during a cross. Widening it does not just add a few extra candidates here and there; it roughly triples the share of positions where a human has to choose.

Which fields does a PPAP typically lock, and why does that matter for a cross?

Because a Production Part Approval Process submission qualifies a part against specific stated values, and a cross that silently changes one of those values is not a substitution — it is a resubmission. We hold no qualification-status column anywhere in Walsin’s 24,026-row MLCC export; nothing in the file records whether a given field was the one a customer’s PPAP covered. That has to come from the buyer, every time.

The practical rule we work to: size, dielectric, capacitance and rated voltage are rarely negotiable, because they are the fields a schematic and a land pattern depend on. Tolerance sits in between — sometimes locked by the PPAP, sometimes left open. Series and qualification grade (MT, AEC-Q200, versus MG, without AEC-Q200 certification) are almost always locked, because they are what the PPAP was actually testing.

Can the capacitance column itself be verified as a must-match field?

Not without checking it against the part number. We found three of the eleven MT-series part numbers carrying the 823 capacitance code in their part number list a Capacitance value of 68nF instead of 82nF — MT15B823J500, MT15B823K500 and MT15B823M500. The other eight 823-coded MT rows in the same export all read 82nF, consistently, which is how we know the three are wrong rather than the other way round.

Part number	Code in part number	Capacitance column reads	Consistent with the other 8 823 rows?
MT18B823K160	823	82nF	Yes
MT21B823K160	823	82nF	Yes
MT18B823K500	823	82nF	Yes
MT15B823K100	823	82nF	Yes
MT15B823K160	823	82nF	Yes
MT15B823K250	823	82nF	Yes
MT18B823K250	823	82nF	Yes
MT21B823K101	823	82nF	Yes
MT15B823J500	823	68nF	No — mismatch
MT15B823K500	823	68nF	No — mismatch
MT15B823M500	823	68nF	No — mismatch

What does the six-candidate 68 nF group actually contain once the mislabelled rows are set aside?

Three real candidates, not six. The six-candidate position we measured above — 0402, X7R, 68 nF, 50 Vdc — is made up of exactly the three 683-coded rows (genuinely 68 nF) and the three 823-coded rows above (actually 82 nF, mislabelled). A buyer whose incumbent part is a true 68 nF MLCC and who filters this export on the Capacitance column alone will see six matches and is actually choosing between three.

Part number	Tolerance	Capacitance column	True capacitance by part-number code	Match / mismatch vs 0402, X7R, 68 nF, 50 Vdc
MT15B683J500	±5%	68nF	68nF	Match — genuine
MT15B683K500	±10%	68nF	68nF	Match — genuine
MT15B683M500	±20%	68nF	68nF	Match — genuine
MT15B823J500	±5%	68nF	82nF	Mismatch — capacitance column is wrong
MT15B823K500	±10%	68nF	82nF	Mismatch — capacitance column is wrong
MT15B823M500	±20%	68nF	82nF	Mismatch — capacitance column is wrong

We will not tell you a part is interchangeable because a spreadsheet column says so. We will show you where the column itself disagrees with the part it describes, which here is the whole finding.

Why does this matter specifically for a tier-1 AVL submission in Germany?

Because the field a PPAP locked is exactly the field this defect hides. A tier-1 engineer in Germany running an automated BOM cross against Walsin’s export, filtering on “68 nF,” would accept all six rows above as capacitance matches — and three of them are actually 82 nF, a 20.6% capacitance difference that a filter check alone would never catch. If that tier-1’s PPAP locked capacitance at 68 nF, three of the six “matches” would fail incoming inspection, not because the part is wrong, but because the export said something the part number itself contradicts.

This is the case for never automating a field-for-field cross past the export alone: Walsin’s own approval sheet, 11_Automotive_MT.pdf, is the only source that resolves which of the six is which. We hold it; we have not yet opened it page by page to confirm the correct capacitance, and we say so above rather than guess.

What should a tier-1 send us before we return automotive candidates?

Four things, in this order. First, the locked fields from the PPAP — which of size, dielectric, capacitance, voltage and tolerance the approval actually tested, because the export carries no qualification-status column at all. Second, the grade: AEC-Q200 (MT, 2,176 rows), without AEC-Q200 certification (MG, 788 rows) or soft-terminated automotive (ST, 118 rows) are three different populations, not one. Third, confirm the capacitance by part-number code as well as by column, because we have now shown the column can be wrong. Fourth, paste the whole line into the free cross reference tool, which filters by commercial or automotive grade and states every difference from your original rather than hiding it, and check what is physically in Hong Kong on ready stock, a dated snapshot as at 23 September 2026.

For the fields a cross tool asks you to check before it asks you anything else, see Thickness and tolerance: the two fields every cross tool asks you to check. For the documentation reality behind an AVL submission, see AVL submission documentation: what we hold. The full automotive MLCC range sits on the MLCC page.

The short version

Within Walsin’s own AEC-Q200 automotive MLCC population, an exact five-field match is already a single answer 99.5% of the time — second sourcing does not usually need a judgement call if tolerance stays locked. Drop tolerance from the match and that collapses to 60.8%, with up to six candidates at one position. At that worst position, three of the six are not real matches at all: the capacitance column reads 68 nF on rows whose own part-number code says 82 nF, confirmed against eight other 823-coded rows that read correctly.

Lock the fields your PPAP actually tested, send us the part number alongside the spec sheet, and we will tell you which field moved — including the one time the spreadsheet itself was wrong.

---

Frequently asked questions

<!– wp:rank-math/faq-block {"questions":[{"id":"faq-question-01","title":"How do I second source a passive component without triggering a PPAP resubmission?","content":"Keep every field the PPAP tested fixed and identify which field, if any, is genuinely flexible. Within Walsin's AEC-Q200 automotive MLCC export, matching size, dielectric, capacitance, voltage and tolerance resolves 99.5% of positions to exactly one part number, so holding tolerance fixed usually removes the judgement call entirely.","visible":true},{"id":"faq-question-02","title":"What happens if I let tolerance move during a cross?","content":"The unique-match rate drops from 99.5% to 60.8% across the same automotive MLCC population, and the worst position returns six candidates instead of one or two. Tolerance is the field most likely to be the one a PPAP actually locked, so widening it is the change most likely to need a resubmission.","visible":true},{"id":"faq-question-03","title":"Can I verify Walsin's exported capacitance column without checking the part number?","content":"Not always. We found three MT-series part numbers whose part-number code reads 823, which should decode to 82nF based on eight other 823-coded rows in the same export, but whose Capacitance column reads 68nF instead. A filter on the column alone would wrongly treat them as 68nF matches.”,”visible”:true},{“id”:”faq-question-04″,”title”:”Does Venatech claim AEC-Q200 certification for its parts?”,”content”:”No. Walsin’s own datasheets describe AEC-Q200 \”compliance,\” not certification, and Venatech makes no qualification claim of its own. Request the qualification report for the exact part number before design-in; this post does not substitute for it.”,”visible”:true},{“id”:”faq-question-05″,”title”:”What fields are usually safe to leave open during a second-source cross?”,”content”:”None, without confirming against the PPAP first. Our own export carries no column recording which field a qualification covered, so size, dielectric, capacitance, voltage, tolerance, series and grade all have to be checked against the buyer’s own approval documentation rather than assumed from the catalogue.”,”visible”:true},{“id”:”faq-question-06″,”title”:”Does Venatech have a presence in Germany?”,”content”:”No. Venatech Limited operates from Hong Kong, with Shenzhen, Singapore and Jiangmen, and a Chengdu location in preparation. A German tier-1 uses the same free cross reference tool and the same dated Hong Kong stock snapshot as any other market; nothing about the process changes by country.”,”visible”:true},{“id”:”faq-question-07″,”title”:”Why does automotive lead time make this urgent now?”,”content”:”The trade press has reported MLCC lead times of 26 to 32 weeks on automotive-grade series through 2026, with EV and AI server demand pulling on the same manufacturing capacity. A second source identified correctly on the first pass avoids losing weeks to a resubmission caused by a field nobody checked.”,”visible”:true}],”listStyle”:””,”titleWrapper”:”h3″,”titleCssClasses”:””,”contentCssClasses”:””,”sizeSlug”:””} –>

---