Inductor Tolerance Code vs Tolerance Column: 55 of 4,443 Walsin Part Numbers Disagree With Themselves, Counted for Slovakia

An inductor tolerance code and the Tolerance column should agree, and in 55 of 4,443 Walsin part numbers they do not. We read the letter in 13 series: 4,388 agree, 55 contradict, and 52 of those sit in three runs. A Slovak buyer ordering by letter and one ordering by column may get different parts.

For a buyer in Slovakia the order code is the first thing goods-in reads, and the parametric column is the first thing the selector reads. Venatech Limited, an authorised Walsin distributor in Hong Kong, holds the inductor parametric export published by Walsin Technology (Walsin): 5,411 rows, of which Test123999 is a QA test row we exclude, leaving 5,410. This post is written from Walsin_Inductor_ALL.xlsx and compares two fields of the same row, the letter inside Part Number and the string in Tolerance. We hold no Walsin inductor datasheet, so we do not say which side is correct. Every part number below is real.

Does the tolerance letter in a Walsin inductor part number always match the Tolerance column?

No. We located the tolerance letter at a fixed position in 4,443 part numbers across 13 series, checked that one letter is used consistently, and compared it with the Tolerance string on the same row. 4,388 rows agree (98.76 percent) and 55 contradict the column (1.24 percent). The other 967 of the 5,410 rows were not tested: 23 carry a bare zero for tolerance, and the rest have no letter we could read at a fixed position.

We have not seen this counted anywhere. The Tolerance column has only 12 distinct values, which we covered in inductor tolerance codes, absolute and percentage; this post is the cross-check between that column and the order code.

Which letter means which tolerance in the 4,388 rows that agree?

Ten letters, each mapping to one value in the overwhelming majority of rows. We built the map from the data, not from memory, and counted the rows that fit it. B, D, G, H, J and S have zero exceptions across 3,158 rows; C, K, M and N each have at least one.

The map is the file’s own. We are not asserting that it matches any standard; with no inductor datasheet on file, the order-code table for your exact series is the document to request.

Where do the 55 disagreements sit?

In four places, and the biggest two are whole runs rather than stray cells. CSME and CSMEA together account for 26 rows, the WFI2520 run for 26, and three isolated rows sit in WLCM, MCI and MCI-TM.

Take CSME. In its 0540 and 0645 codes the letter M maps to plus or minus 20 percent on 28 rows and N to plus or minus 30 percent on 32. In the 0418D code, all 13 rows are reversed: 11 M parts read 30 percent and 2 N parts read 20 percent. CSMEA repeats it exactly, so one inverted run appears twice.

What does the WFI2520 run show that a count does not?

That a letter can be redundant. WFI2012 publishes both J and K, and they differ: 28 J rows read plus or minus 5 percent and 28 K rows read plus or minus 10 percent. In WFI2520 the file lists 26 J rows and 26 K rows, and every K row also reads plus or minus 5 percent. We compared nine columns of each pair, and the 26 K parts are identical to their 26 J twins in size, thickness, inductance, unit, Q, SRF, DCR, rated current and tolerance. WFI2520FS47NJ and WFI2520FS47NK are one set of numbers under two orderable names.

Whether the K row should read 10 percent with different data, or the letter is a duplicate listing, is a question for Walsin. For the wider pattern of identical rows, see chip inductor second source.

Which side of a disagreement is right?

We cannot say, and none of the export’s 14 columns does either. A letter that disagrees with a column is a statement that one of two cells is wrong; nothing in 14 columns says which. The 5,410 rows publish no operating temperature, no qualification status and no lifecycle field, and we hold no inductor datasheet to arbitrate.

What we can say is how to bound the risk. If you order by the letter, 55 of 4,443 parts, 1.24 percent, may carry the other tolerance. If you filter by the column, the same 55 may carry the other letter. A design whose circuit tolerates plus or minus 30 percent on a power inductor is unaffected by a CSME swap between 20 and 30; one that was designed on plus or minus 5 percent is not unaffected by a K that might be 10.

What should a Slovak buyer check before ordering by tolerance?

Three checks, and 52 of the 55 disagreements fall to the first of them.

1. Read the letter and the column together. In 55 of 4,443 rows they disagree, and 52 of those 55 sit in only three runs: CSME and CSMEA size 4040, and WFI2520. We checked every row in Walsin_Inductor_ALL.xlsx for this.
2. Never assume K means 10 percent without the series. In WFI2520, 26 K parts publish 5 percent.
3. Request the Walsin order-code table for your exact series. We hold 28 Walsin datasheets, 451 pages, and none covers an inductor.

Our free cross-reference tool shows the published tolerance string beside the original and states differences instead of hiding them. 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 range sits on the inductor page.

The short version

We read the tolerance letter in 4,443 Walsin inductor part numbers across 13 series and compared it with the Tolerance column. 4,388 agree and 55 contradict, 1.24 percent. B, D, G, H, J and S have no exceptions in 3,158 rows. CSME and CSMEA size 4040 are inverted on 26 rows, WFI2520 repeats K at 5 percent on 26 rows that are identical to their J twins, and three single C rows disagree in WLCM, MCI and MCI-TM. We will not tell you which cell is wrong. We will show you both. Confirm against both datasheets and qualify in your own circuit.

Send us the inductor line off your BOM with the tolerance you designed to, and we will show the letter and the column side by side for every Walsin candidate. And a question back: when your goods-in check reads the order code, who reads the column?

Frequently asked questions