RF Insertion Loss by Case Size: In 72 Walsin Baluns the Larger Package Loses More in 1,283 of 1,660 Pairs

RF insertion loss by case size does not follow the rule most designers assume. We compared all 72 Walsin baluns: the larger package published the higher loss in 1,283 of the 1,660 size-differing pairs that are not tied. Walsin’s 2.4 GHz filters give four different answers.

Venatech Limited, an authorised Walsin distributor in Hong Kong, parsed all 15 files in the RF export published by Walsin Technology (Walsin): 457 part numbers across 8 functions. This post is written from the Size and insertion-loss columns of the balun and filter files. Every part number below is real. The question comes from a design review we keep hearing: “if I move up one case size, do I buy lower loss?”

Does a bigger Walsin RF balun lose less signal?

No, not in the figures Walsin publishes. We grouped all 72 baluns by the Size column and compared every pair of parts at different sizes. Of 1,841 cross-size pairs, 1,283 show the larger package with the higher published loss, 377 show it lower and 181 tie. That is 77.3 percent of the 1,660 pairs that differ, running against the common assumption.

The picture is not a clean staircase. Loss rises from 0605 to 1608 and then eases at 2012, so the rank correlation between footprint and loss is +0.58 rather than +1. But the best-case floor does not move with size at all: 0.55 dB is the lowest figure at 0605, 1005 and 1608. We have not seen this counted anywhere. Size is not buying the floor.

Does the same rule hold for 2.4 GHz filters?

Partly, and the answer depends on the filter type. We counted the 111 filters whose band is 2400 to 2500 MHz and split them by Filter Type after merging the double-space spelling. For the 68 band pass filters with a numeric loss, size and loss are unrelated: the larger package is lower in 650 pairs and higher in 711, with 81 ties, a rank correlation of +0.05.

The three band pass filters with no numeric loss hold text such as 1.50 dB max. at 25℃ and are excluded from the counts.

Why does the size and loss relationship change sign between functions?

Because the parts are not the same circuit in different boxes. A low pass filter, a band pass filter, a balun and a balanced filter spend their package volume on different things, and the export says nothing about which. Across the 108 numeric 2.4 GHz filters of all types the rank correlation is +0.40, but inside each type it runs from -0.37 to +0.42, so the pooled figure describes the product mix, not the physics.

The extremes make the point. At 2400 to 2500 MHz the lowest published filter loss is 0.37 dB on RFLPF1608060AGT, a 1608 low pass part. The highest is 4.5 dB on RFBPB2520120A4T, a 2520 balanced filter. Between them sit RBBPF1411060A8T at 1.0 dB in 1411 and RFBPF2012080A6T at 3.5 dB in 2012, both band pass. A designer who sorts the whole 2.4 GHz shelf by case size and reads off a trend is reading the filter-type mix. If you must compare, fix the function first, then the band, then ask what the loss figure means. For the function-by-function size grid, see Walsin RF case size: 17 sizes, 8 functions.

What can you not conclude from these insertion loss figures?

Less than the table suggests. We read every insertion-loss cell and 407 of the 457 RF parts publish a bare number with no maximum, typical or test condition. A bare 0.7 and a bare 1.2 may not be the same kind of figure. The counts above compare numbers as published, not performance.

Two further limits apply. Several size and type cells are thin: the 1005, 1109 and 1510 band pass and balanced rows hold one part each, so any single comparison there rests on one figure. And the export carries no board-level data: nothing about matching network, ground plane or layout, all of which move measured loss. The five column names carrying loss across the RF line are covered in Walsin RF insertion loss: 457 of 457 parts.

How do you test your own incumbent part against this?

Fix function and band, then compare the published loss at your size and at the next size down. The 2.4 GHz balun shelf is the cleanest test because all four sizes of interest sit in it. Leave the right-hand column for your own part and check it field by field.

A loss figure is one field of several. Phase difference, amplitude difference and balanced impedance decide whether a balun fits your transceiver; our balun selection guide lists them. Our free cross-reference tool states differences rather than hiding them, and up to 500 part numbers go in at once. 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 RF products page. We hold zero datasheets for the RF line; the figures above come from Walsin’s export alone. Confirm against both datasheets and qualify in your own circuit.

The short version

• 72 baluns, 1,841 cross-size pairs: the larger package is higher in 1,283, lower in 377, tied in 181.
• The lowest balun loss, 0.55 dB, appears at 0605, 1005 and 1608. Size does not move the floor.
• At 2.4 GHz, band pass filters show no size trend (+0.05); low pass filters lean the other way (-0.37).
• 407 of 457 RF loss figures are bare numbers. Treat every comparison as provisional.

Pick the function, then the band, then ask the loss question, in that order. And a question back: if the smaller package loses less, what exactly was the larger one for?

Frequently asked questions