Walsin RF Insertion Loss: 457 of 457 Parts Publish One, Under 5 Different Column Names

Every one of the 457 parts in Walsin’s RF export publishes a walsin rf insertion loss figure. That is the good news and the end of it. We counted 5 different column names carrying that one specification, and 407 of the 457 state it as a bare number with no max, no typical and no test condition.

---

Venatech Limited, an authorised Walsin distributor in Hong Kong, parsed all 15 files in Walsin Technology (Walsin)’s RF product export — 457 part numbers across 8 distinct functions — and counted every cell that carries an insertion-loss value. Every part number below is real and present in Walsin’s own export. We hold zero datasheets for the RF line, so nothing here is checked against a PDF; for specifications you must go to Walsin’s own site, and every figure in this post is a figure in the export, not a verified device characteristic.

How many of Walsin’s 457 RF parts publish an insertion loss figure?

All 457. We checked every row of every file and found 590 non-empty insertion-loss cells across 457 part numbers — more cells than parts, because some functions publish the figure once per band. Zero parts leave the field blank. For a line with no datasheet coverage, that is better than it sounds and worse than it looks: the 590 cells are not in one format and do not sit under one heading.

Function	Export file	Column name holding insertion loss	Parts	How the value is written	Match / mismatch against one sortable loss column
Filter	Walsin_Filter.xlsx	Insertion Loss (dB)	235	221 bare numbers, 14 free text	Mismatch — 14 cells bury a temperature range inside the value
Balun	Walsin_Balun.xlsx	Insertion Loss (dB)	72	72 bare numbers, zero text	Match on format, mismatch on meaning — no max or typical stated on any of the 72
Coupler	Walsin_Coupler.xlsx	Insertion Loss (dB)	17	4 bare numbers, 13 multi-line strings	Mismatch — 13 cells hold three or four band-and-loss pairs each
Antenna switch	Walsin_Antenna_Switch.xlsx	Insertion Loss(dB)	9	9 text strings, all ending “max.”	Mismatch — different header, and no numeric value anywhere
Multiplexer	Walsin_Multiplexer.xlsx	Insertion Loss(dB) for Frequency 1, 2, 3	116	110 numeric, 6 text	Mismatch — three columns for one specification
BAW filter	Walsin_BAW_Filter.xlsx	Insertion Loss(dB) for Frequency 1, 2	5	5 text strings ending “max”	Mismatch — “max” without a full stop here, “max.” in the switch file
SAW filter	Walsin_SAW_Filter.xlsx	Insertion Loss(dB) for Frequency 1, 2	3	3 text at band 1, - on all 3 at band 2	Mismatch — a hyphen reads as data and means nothing
Antenna switch module	Walsin_Antenna_Switch_Module.xlsx	Insertion Loss(dB)	0	file returns zero rows	Mismatch — the header exists, the shelf does not

Read the right-hand column before the left one. Seven of the eight functions carry a real population of parts; the eighth is a 12-column header with nothing under it. We have not seen this counted anywhere.

Why does one specification carry five different column names?

Because the eight function files were built as eight separate schemas, not as one catalogue. We extracted every header string across all 15 files and found 38 distinct column names, of which exactly 5 carry insertion loss: Insertion Loss (dB) with a space before the bracket on 324 parts, Insertion Loss(dB) without the space on 9, and Insertion Loss(dB) for Frequency 1, for Frequency 2 and for Frequency 3 on 124 between them. 324 plus 9 plus 124 is 457, which is the whole line.

The space before the bracket is the one that bites. A lookup written against one spelling returns nothing from the other four files and raises no error, because the column is absent rather than empty. That is the same failure mode as the chip resistor part numbers carrying an embedded space — see why Walsin part numbers do not paste into your BOM. Series is the only column name present in all 203 of Walsin’s export files; insertion loss is not consistent inside one family of 457 parts.

Can you sort the Walsin RF catalogue by insertion loss?

Not honestly, and here is the measurement. 407 of the 457 publish insertion loss only as a bare number — 1.2, 0.65, 2.5 — with no unit, no max or typical qualifier and no test condition in the cell. 50 of the 457 publish at least one free-text value, which no numeric sort will order. 24 cells contain an embedded line break, so a single cell holds several band-and-loss pairs stacked vertically. 3 cells read -, which is a null written as a character and will sort as text ahead of every number.

The honest statement about the 407 is not that the figure is wrong. It is that you cannot tell what it is. On the 9 antenna switches and the 5 BAW filters the value is explicitly a maximum, because the string says so. On the 72 baluns and the 221 numeric filters nothing says so, and a maximum and a typical figure for the same device are not the same number. We will not tell you the 407 bare numbers are maxima. We will tell you the export does not say.

Five worked examples from the 2400–2500 MHz shelf, where the comparison matters most:

Part number	Function	Case size in the Size column	Insertion loss as published	What the cell actually contains	Match / mismatch for a like-for-like loss comparison
RFBPF1608050A0T	Band pass filter	1608	0.5	a bare number, no condition	Mismatch — unqualified; could be typical or maximum
RFBLN1005040A57N3T	Balun	1005	0.55	a bare number, no condition	Mismatch — same format, different function, unbalanced to balanced path
TFCPL0605B2450T	Coupler	605	0.32	a bare number, with 14.6 in Coupling in BW	Mismatch — through-path loss, not a filter’s passband loss
RFCPL1810B2450T	Coupler	1608	0.74	a bare number	Mismatch — 2.3 times the 605 part, and the part number says 1810 while Size says 1608
AMDIP2012050LP7B1U	Diplexer	2012	0.5 and 0.8	two columns, one per band	Mismatch — no single figure exists for this part

Note the fourth row. RFCPL1810B2450T carries 1810 in its own part number and 1608 in the Size column. We read both and we are not going to reconcile them for you from an export we cannot check against a datasheet.

Which RF functions publish insertion loss per frequency instead of once?

Three of the eight: multiplexer, BAW filter and SAW filter, covering 124 of the 457 parts. The other five functions publish one figure for the whole part. That split is not arbitrary — a diplexer has two signal paths and a single loss figure for it would be meaningless — but it does mean the field you must read changes with the function you are buying.

The per-frequency columns are unevenly filled, and the gaps are informative:

• Multiplexer Insertion Loss(dB) for Frequency 1: 116 of 116.
• Multiplexer for Frequency 2: 116 of 116.
• Multiplexer for Frequency 3: 13 of 116.
• BAW for Frequency 1: 5 of 5. BAW for Frequency 2: 1 of 5.
• SAW for Frequency 1: 3 of 3. SAW for Frequency 2: 3 of 3, and all three read -.

The SAW row is the one worth a sentence. SF11091582GC2T, SF11092350B4004T and SF11092595B4108T all carry a hyphen in the second-frequency loss column. A script that tests for a non-empty cell counts that as published data. It is not data. Walsin’s whole SAW shelf is three part numbers, all at case size 1109, so this is not a large population — but it is the entire line and the entire line is written this way. For how the four RF functions divide up between them, see Walsin RF: balun, filter, coupler and multiplexer.

What does the multiplexer export get wrong about its own three-band parts?

It under-counts them by two. 13 of the 116 multiplexer rows publish a third frequency band and a third insertion-loss figure, but the Multiplexer Type column names only 11 of them Triplexer. The two extra rows, AMDIP1606L96A8Q1C and AMDIP1606L96B8Q1C, are both typed Diplexer and both carry three bands — 2400–2500 MHz, 4900–5100 MHz and 5150–5950 MHz — with three separate insertion-loss values of 0.7, 1.3 and 1.0 dB. We checked all 116 rows and these are the only two where the type label and the populated band count disagree. We have not seen this counted anywhere.

That matters to a selector, not to a theorist. Filter the file on Multiplexer Type == "Triplexer" and you get 11 three-band parts and miss two. Filter on a populated Frequency 3 and you get 13, two of which say they do something else. Neither filter is wrong about the data. The data is wrong about itself.

This is the same file that is offset by one column from Pin Description onward, with Category empty on all 116 rows. We reproduced that offset before relying on anything to its right: on AMDIP2004L157B1U the Pin Description column holds 50, the impedance columns hold attenuation text, and Pin Description.1 holds LTCC. The insertion-loss columns sit to the left of the offset and are unaffected, which is the only reason the counts above stand. For how the band strings themselves are written, see choosing an RF chip filter by band.

Two more spelling defects in the same family, both reproduced from scratch. The filter file’s Filter Type column writes band pass two ways — Band Pass Filter with a double space on 110 parts and Band Pass Filter on 2 — and writes Low Pass filter with a lower-case f on 94 while High Pass Filter takes a capital on 3. The balun file’s Phase Difference column uses three different degree characters across 72 rows: the standard degree sign on 51, a ring operator on 16 and a ring above on 5. None is a performance problem. All of them break a sort.

What should you ask for before you commit to a Walsin RF part on insertion loss?

Three things, in this order, and the first is not negotiable. Ask for the datasheet. We hold 28 Walsin datasheets, 451 pages. Zero of them cover any RF part number.

1. Name the part number and ask whether the loss figure is maximum or typical. On 407 of 457 the export does not say, and that single word changes a margin calculation.
2. Ask for the band edges the figure applies to. The 2400–2500 MHz shelf is written six different ways in the filter file across 111 parts, and it is the only band in the whole RF line holding a filter, a coupler and a balun.
3. Cross the surrounding passives while you are there. Paste up to 500 part numbers or drop your Excel or CSV into the free cross-reference tool and read every stated difference.

Check what is physically in Hong Kong on ready stock: 1,125 part numbers as at 23 September 2026, a dated snapshot and not a live feed. The full RF range sits on the Walsin RF page.

The short version

All 457 Walsin RF part numbers publish an insertion-loss figure, under 5 column names, in 590 cells. 407 state it as a bare number with no maximum, no typical and no test condition. 50 carry free text, 24 cells hold an embedded line break, and 3 read -. 13 of 116 multiplexer rows publish a third band while only 11 are typed Triplexer. We hold zero datasheets for this line and we will not pretend otherwise: every specification claim above is a claim about a spreadsheet cell. Qualification stays with you, and on this line so does verification — our job is to get you the document that settles it.

Send us the RF line items off your bill of materials and we will tell you, part number by part number, exactly which column the loss figure came from, whether the export qualifies it as a maximum, and what we will have to ask Walsin for. And a question back: when your selector sorted that shelf by insertion loss last month, which of the five column names did it read?

---

Frequently asked questions

<!– wp:rank-math/faq-block {"questions":[{"id":"faq-question-01","title":"How many Walsin RF parts publish an insertion loss figure?","content":"All 457, in 590 non-empty cells across 15 export files and 8 functions. No part leaves the field blank. The figure appears under five different column names, and 407 of the 457 give it as a bare number with no maximum, typical or test-condition qualifier anywhere in the cell.","visible":true},{"id":"faq-question-02","title":"Why are there five insertion loss column names in one Walsin RF export?","content":"Because the eight function files use eight separate schemas, with 38 distinct column names between them. Filter, balun and coupler write Insertion Loss (dB) with a space. Antenna switch, BAW, SAW and multiplexer write Insertion Loss(dB) without one. Three of those also split the figure across per-frequency columns.”,”visible”:true},{“id”:”faq-question-03″,”title”:”Does Venatech hold datasheets for Walsin RF components?”,”content”:”No. We hold 28 Walsin datasheets, 451 pages, covering multilayer ceramic capacitors and two chip resistor series only. Zero cover any of the 457 RF part numbers. For RF specifications we point you at Walsin’s own site and flag every figure, and we will request a datasheet in writing on your behalf.”,”visible”:true},{“id”:”faq-question-04″,”title”:”Is the insertion loss figure in Walsin’s RF export a maximum or a typical value?”,”content”:”The export does not say on 407 of the 457 parts. The 9 antenna switches and the 5 BAW filters state \”max\” inside the cell text. The 72 baluns and the 221 numeric filter rows state nothing. We will not characterise an unqualified number as a maximum on your behalf.”,”visible”:true},{“id”:”faq-question-05″,”title”:”How many Walsin multiplexers cover three frequency bands?”,”content”:”Thirteen of 116 rows publish a third band and a third insertion-loss figure, but the type column names only eleven as a triplexer. Two parts typed as a diplexer carry three bands at 2400 to 2500, 4900 to 5100 and 5150 to 5950 megahertz with three separate loss values.”,”visible”:true},{“id”:”faq-question-06″,”title”:”Can I sort Walsin’s RF catalogue by insertion loss?”,”content”:”Not reliably. Fifty of 457 parts carry free text in the field, 24 cells contain an embedded line break holding several band-and-loss pairs, and 3 cells read a hyphen that sorts as text. A numeric sort silently drops or misplaces all of them before you see a result.”,”visible”:true}],”listStyle”:””,”titleWrapper”:”h3″,”titleCssClasses”:””,”contentCssClasses”:””,”sizeSlug”:””} –>

---