Numeric Values Stored as Text in Walsin Export Files: 5 of 8 Families Hold No Native Number, and 248 Inductor Cells Lose Their Unit

Numeric values stored as text are the rule in Walsin’s exports, not the exception. We read the stored cell type of all 359,974 rows: five of the eight families hold no native number in any column, and across 3,314,636 cells only 38,214 (1.15 percent) are numbers.

---

Venatech Limited, an authorised Walsin distributor in Hong Kong, holds all 203 parametric export files published by Walsin Technology (Walsin). This post is written from the stored cell types of the eight primary export sets, read with a spreadsheet library that reports whether each cell is an integer, a float or a string, not from how a cell looks on screen. We excluded the QA test row Test123999. Every part number below is real. The question is the one a BOM import hits first: when a column looks numeric, is it?

How many cells in the Walsin exports are stored as numbers rather than text?

38,214 of 3,314,636. We counted every cell outside the part-number column in 359,974 rows across the eight families and classified each as a native number, a number-looking text string, a quantity written as text with a unit attached, other text, or empty. Native numbers are 1.15 percent of the cells, and 91.7 percent of the rows (330,081 of 359,974) contain no native number whatsoever.

Family	Rows	Cells outside the part-number column	Native-number cells	A real part from the export	Match / mismatch against a filter that expects numbers
Chip resistor	234,815	2,113,335	0	WF04P1024FTL	Mismatch — resistance, tolerance, TCR and power are all text
Thin film	93,914	751,312	0	SF12T1R00ATL	Mismatch — 1.00Ω is a string, not 1.00
MLCC	24,026	360,390	8,454	MT15N0R5C500	Partial — numbers only in Size (from 1206 up) and in defect cells
Inductor	5,410	70,330	27,705	WLCW1005Z0J1N0TB	Match on inductance, thickness, Q and SRF; mismatch on RDC and current
MOV varistor	379	3,411	0	SR180K05D	Mismatch — 18Vdc and 100A are strings
Disc capacitor	929	9,290	0	SL0AH150J060	Mismatch — 15pF is a string
RF	457	6,172	2,055	RFBLN1005040C0T	Partial — insertion loss and impedance numeric, frequency a string
Antenna	44	396	0	RFANT6050110L0T	Mismatch — 4 dBi is a string

We have not seen this counted anywhere. Five families, 330,081 rows, zero native numbers. Counting columns by name, 18 hold numbers and text together (4 in MLCC, 5 in the inductor file, 9 across the RF files) and 11 hold only numbers (4 inductor, 7 RF). Chip resistor and thin film, which are 91.3 percent of the catalogue’s rows, are text from the first column to the last.

Which five families hold no native number in any column?

Chip resistor, thin film, MOV varistor, disc capacitor and antenna. In each, every quantity is a string with its unit welded on: ±0.05%, ≦±50, 1/10W, +/-10%, 4 dBi. That is 2,877,744 cells in the five families, and not one of them is a number a spreadsheet can sort, sum or compare. A column of Resistance values sorts as text, so 10.0Ω lands before 2.00Ω.

Three consequences follow, and each one is a count the reader can repeat.

1. A range filter returns nothing or everything. A “greater than 1 watt” test on Rated Power compares a string such as 1/10W with a number. Depending on the tool, it errors, coerces or silently matches zero rows.
2. The unit travels with the value. The prefix and the unit are part of the string, so 850mA and 1A are two strings, not two magnitudes. Our post on whitespace and split values found 25 values that exist in two spellings for exactly this reason.
3. Nothing needs converting to be wrong. The text is consistent inside one file and still cannot be compared with a number in your BOM.

For the families that Walsin’s exports leave entirely as text, a cross-reference has to parse the string, strip the unit, convert the prefix and only then compare. That is the work our free cross-reference tool does before it shows you a match, and it is why a differing field is stated, not hidden behind a tick.

Why does the MLCC Size column switch from text to number at the leading zero?

Because the number-ness of a case-size cell follows its first digit, with no exception. In 00_ALL_MLCC_parts_combined.xlsx, all 16,885 rows whose size starts with a zero are stored as text, and all 7,140 rows whose size starts with 1 or 2 are stored as integers. We checked all 24,026 rows. The one remaining row is a defect: its Size cell holds a series name and a tab character, for the part 1808B102M252.

Size	Rows	Stored as	A real part	Match / mismatch against a lookup typed as a number
01R5	608	text	01R5N0R8B160	Mismatch — contains a letter, cannot be a number
0201	2,544	text	RF03N0R4D500	Mismatch — typed as 201, the lookup misses all 2,544
0402	3,991	text	MT15N0R5C500	Mismatch — typed as 402, misses all 3,991
0603	4,730	text	SG18B102K500	Mismatch — typed as 603, misses all 4,730
0805	4,912	text	FT21B103M500	Mismatch — typed as 805, misses all 4,912
1206	3,520	integer	TT31X225K100	Match — typed as 1206 finds it; typed as text, it misses
2220	327	integer	2220B223J202	Match — same condition

Nine integer sizes (1111, 1206, 1210, 1808, 1812, 1825, 2211, 2220, 2225) and seven text sizes (the first five rows above plus 0306 and 0505) make up the column. A buyer whose BOM holds 0402 as a number has lost the zero and will never match. A buyer whose BOM holds 1206 as text will never match an integer. The same split runs through the RF export: the 11 parts at 0605 are text, 437 other case-size cells are integers and 9 antenna-switch cells are 1.1x0.7-style strings. Our post on the five spellings of 0201 covers what the text strings say; this is what type they are stored in.

Which Walsin cells are bare numbers with no unit?

248, all in the inductor export. In Walsin_Inductor_ALL.xlsx, RDC and Rated Current are text columns: 0.045Ω, 20mΩ, 1.36A, 850mA. But 177 RDC cells and 71 Rated Current cells are native numbers with no unit, sitting in columns where their neighbours carry Ω and mΩ, or A and mA. We found them by reading the stored type of every cell in both columns.

Part	Series	Column	Cell holds	Neighbour in the same series	Match / mismatch
MCI0603TM0N6BHBP	MCI-TM	RDC	0.07, a float	all 172 MCI-TM rows are bare	Mismatch — unit absent on the whole series
WLPN202012N1R0PB	WLPN	RDC	0.07, a float	all 5 WLPN rows are bare	Mismatch — same figure as the MCI-TM row, no unit on either
WLCW2012HQJ2N5PB	WLCW	RDC	20mΩ, text	341 rows across the file use mΩ	Match — unit stated
WLCM0603Z0J47NTB	WLCM	Rated Current	100, an integer	396 WLCM rows say mA, 122 say A	Mismatch — mA or A is not stated
WLCM0603Z0B0N3TB	WLCM	Rated Current	850mA, text	the same series, same size	Match — unit stated

The bare cells fall in whole series, not at random: MCI-TM and WLPN account for all 177 RDC cells, and WLCM, WLBD and WLBD(HC) for all 71 current cells (41, 16 and 14). The SRF(min) column states no unit in its header or its cells on 5,410 rows. Tolerance holds a bare 0 on 23 rows, a defect we have already published.

The native number is the dangerous one. A text cell announces its unit; a bare number invites you to assume one.

What should a BOM import test before it trusts a numeric column?

Three tests, each cheap. We ran all three across the 359,974 rows to see what they would catch. The first test, “does this column hold one type?”, flags 18 columns by name: 4 in the MLCC file, 5 in the inductor file and 9 across the RF files. The other two tests are below.

1. Test the stored type, not the display. Count integers, floats and strings per column. A column with more than one type, such as Size in the MLCC file or RDC in the inductor file, needs a rule before it joins to anything.
2. Compare as text with the leading zero kept. Store case sizes as text in your own sheet. 0402 as a number is 402 and matches 0 of 3,991 rows.
3. Never infer a unit from a bare number. If a cell has no unit, read the header, and if the header has none either, ask. We stop at an unstated unit and so should your import.

We hold 28 Walsin datasheets, 451 pages, and the datasheet states the unit that a bare export cell leaves out. Ask for the sheet covering your part number and it will be in your inbox. For inductor, thin film, MOV, disc, RF and antenna we hold none, so confirm those lines against Walsin’s own site. Why a column that looks like a selector can still select nothing is covered in the walsin-mlcc termination defect post, which accounts for 1,314 of the 8,454 MLCC native cells.

What is in Hong Kong is on ready stock: 1,125 part numbers as at 23 September 2026, a dated snapshot and not a live feed. The inductor range sits on the inductor page.

The short version

Five of eight Walsin families, 330,081 of 359,974 rows, hold no native number in any column. Native numbers are 38,214 of 3,314,636 cells, 1.15 percent. In the MLCC export the Size cell is text if it starts with a zero (16,885 rows) and an integer if it does not (7,140 rows), with no exception. 248 inductor cells, 177 RDC and 71 rated current, are bare numbers with no unit, in whole series. We will not tell you a part matches on a number we cannot see the unit of. We will show you the cell as stored. Qualification stays with you.

Send us the column headers and the first ten rows of your BOM and we will tell you which columns are text, which are numbers and which are both. And a question back: when your last import matched nothing at 0402, did you suspect the part, or the zero?

---

Frequently asked questions

<!– wp:rank-math/faq-block {"questions":[{"id":"faq-question-01","title":"How many cells in the Walsin exports are native numbers?","content":"38,214 of 3,314,636 cells outside the part-number column, which is 1.15 percent. Chip resistor, thin film, MOV varistor, disc capacitor and antenna hold none at all, covering 330,081 of 359,974 rows. The numbers sit in the inductor, RF and MLCC files, and 1,314 of the MLCC ones are a known data defect.","visible":true},{"id":"faq-question-02","title":"Why does my MLCC size lookup miss 0402 parts?","content":"Because the 0402 size is stored as text and the zero is part of it. In the combined MLCC export, 16,885 rows with a size starting in zero are strings and 7,140 rows from 1111 upward are integers. A lookup value typed as a number becomes 402 and matches none of the 3,991 rows at 0402.","visible":true},{"id":"faq-question-03","title":"Are the chip resistor values in the Walsin export numbers?","content":"No. Every quantity in the 234,815 chip resistor rows is a string with the unit attached, including resistance, tolerance, TCR and rated power. The 93,914 thin film rows are the same. A spreadsheet will sort them as text, so 10.0Ω sorts before 2.00Ω, and a numeric comparison needs the unit parsed off first.”,”visible”:true},{“id”:”faq-question-04″,”title”:”Which Walsin cells have a number but no unit?”,”content”:”177 inductor RDC cells and 71 inductor rated-current cells. The RDC cells are every row of the MCI-TM series (172) and the WLPN series (5). The current cells are in the WLCM, WLBD and WLBD(HC) series. Neighbouring rows state Ω, mΩ, A or mA, so the bare cells need confirming against Walsin.”,”visible”:true},{“id”:”faq-question-05″,”title”:”Should I convert the export to numbers before importing it?”,”content”:”Only after you have parsed the unit and the prefix, and only for columns that are single-type. Converting blindly turns 0402 into 402 and drops the unit from 850mA. Keep case sizes as text, parse quantities into a number plus a unit column, and keep the original string beside the result.”,”visible”:true},{“id”:”faq-question-06″,”title”:”Does the stored type differ between the export and the datasheet?”,”content”:”A datasheet is a document and has no stored cell type, but it states units and conditions that the export leaves out. We hold 28 Walsin datasheets, 451 pages, for 26 MLCC and 2 chip resistor series, and none for the inductor, thin film, MOV, disc, RF or antenna lines.”,”visible”:true}],”listStyle”:””,”titleWrapper”:”h3″,”titleCssClasses”:””,”contentCssClasses”:””,”sizeSlug”:””} –>

---