RF Transceivers Test Data In Production

SXS046 vs AD9361: Measured Performance Comparison

Published: November 16, 2026 · Updated: November 16, 2026 · Category: RF Transceivers · Reading time: ~9 min

Datasheet parameters tell you what a transceiver is designed to do. Measured data tells you what silicon actually does. Shenxin Tech publishes a full-parameter comparison of the SXS046 against the AD9361 in the SXS046 / AD9361 Comparison Report (PDF): the two devices were characterized on a 93K automated test system at three case temperatures — 25°C, −55°C and 125°C — supplemented by transmitter measurements on a manual test board. This article walks through the headline results and explains how to read them.

One convention matters before any number: the report lists the SXS046 as minimum / average / maximum across the automated-test sample set, next to the measured value of the AD9361 reference device. The SXS046 average is therefore the right column to compare against the AD9361 figure. Blank cells and "/" in the report mean "not measured under that condition", not zero.

Receiver Results at Room Temperature (Tc = 25°C)

Selected rows from the report's room-temperature table. Gain, noise and linearity figures below are at maximum RX gain unless the condition column says otherwise.

ParameterSXS046 (93K average)AD9361 (measured)UnitConditions
Maximum gain (relative)70.99971.404dB@ 800 MHz
Maximum gain (relative)69.94268.932dB@ 2300 MHz (RX1A, RX2A)
Maximum gain (relative)61.35961.148dB@ 5500 MHz (RX1A, RX2A)
Gain step0.9720.978dB—
RSSI range94.19596dB—
Noise figure (NF)2.1961.925dBMaximum RX gain
Out-of-band IIP3-17.711-17.22dBmMaximum RX gain
Out-of-band IIP247.94448.483dBmMaximum RX gain
DC offset-85.612-84.165dBmRX front-end input, digital-domain measurement
Image rejection75.87379.592dBMinimum RX gain
RX1 to RX2 isolation (A/C ports)86.41887.977dBMinimum RX gain
Modulation accuracy (EVM)-41.536not reporteddBSXS046 on manual board, 19.2 MHz reference clock

A second measurement block in the same table (40 MHz reference clock conditions) reads:

ParameterSXS046 (93K average)AD9361 (measured)Unit
Noise figure (NF)3.2303.621dB
Input IP3 (IIP3)-13.003-11.821dBm
Input IP2 (IIP2)50.74354.418dBm
LO leakage-80.842-80.143dBm
Image rejection79.42580.24dB
Modulation accuracy (EVM)-40.11-39.90dB

Read honestly, the two devices trade small advantages: the AD9361 reference measures slightly better on noise figure in the first block and on image rejection, while the SXS046 averages better on noise figure and IIP3 in the second block and on gain at 2300 MHz and 5500 MHz. Nothing in the tables supports a claim of identical performance — what they support is that the SXS046 sits in the same performance envelope, parameter by parameter, across the full operating range. The complete tables, including the −55°C and 125°C runs, are in the comparison report (PDF).

Transmitter Bench Results (Manual Test Board)

The report's second section measures the TX channel on a manual test board: one AD9361 reference unit (column "ADI") against three SXS046 samples (85#, 90#, 105#). Output power across the band:

Frequency*AD9361 (ADI)SXS046 85#SXS046 90#SXS046 105#
4502.731.851.881.90
8502.141.581.681.32
9501.801.161.160.92
1450-0.02-0.65-0.60-0.75
2400-1.09-1.71-2.04-2.39
3300-0.82-1.59-1.75-2.31
5200-4.61-4.88-4.66-5.32
5800-7.93-6.86-6.98-7.69

Output power in dBm. *The source report lists frequencies without stating the unit; values are reproduced exactly as published. The same section tabulates LO leakage, image rejection and second/third harmonic suppression on the same frequency grid: harmonic suppression is comparable between the devices, while image rejection on the bench board favors the ADI reference unit at several frequencies — one more reason the report publishes the raw tables instead of a summary verdict.

The One Functional Difference: Baseband Filter Tune Dividers Must Be Even

The report's functional comparison concludes that the SXS046 supports pin-to-pin replacement of the AD9361 on the existing footprint, with exactly one register configuration requirement. The clock for baseband bandwidth (BBF) calibration is derived by dividing the main clock: the RX divider is set by Reg1F8[7:0] (BBF Tune Divide) and the TX divider by Reg0D6[7:0]. On the SXS046 the divider must not be an odd number — Reg1F8[0] and Reg0D6[0] must not be set to 1. The divider follows from the BBPLL frequency and the baseband bandwidth:

BBF Tune Divide = ceil( (BBPLL Freq × ln(2)) / (BBBW × 1.4 × 2 × π) )

The report gives two ways to handle it:

Method 1 — adjust and reinitialize: After initialization, read back the two registers. If a divider is odd, adjust the baseband bandwidth and reinitialize the chip; the changed bandwidth changes the division ratio to an even value.
Method 2 — force bit 0 to zero: If the initialization values can be modified, write Reg1F8[0] and Reg0D6[0] directly to 0 before initialization. The power-up manual states the same rule from the SXS046 side: 0x1F8[D0] and 0x0D6[D0] must be set to 0.

This is the concrete content behind the line used across this site — "register configuration adaptation is required". It is a bounded, checkable item, and the power-up configuration and calibration guide places it in the full initialization sequence.

What This Data Does and Does Not Establish

The comparison report establishes measured, like-for-like behavior of the two devices on the same test systems, over temperature, across the full parameter list — receiver gain, noise, linearity, isolation, transmitter power, leakage, harmonics, auxiliary converters and logic levels. For an engineer qualifying a second source, that is the evidence base for the RF side of the decision.

It does not replace board-level validation. Front-end matching, clocking and the register port still have to be verified on the target hardware, which is why Shenxin pairs the report with samples and FAE support. Start from the AD9361 alternative overview for the migration picture, and the SXS046 product page for the datasheet and sample requests.

Frequently Asked Questions

Where does this comparison data come from?
From Shenxin Tech's published SXS046 / AD9361 Comparison Report: full-parameter characterization on a 93K automated test system at 25°C, −55°C and 125°C, plus TX-channel measurements on a manual test board. The full report is a free download: Comparison Report (PDF).
Is the SXS046's performance identical to the AD9361?
No — and the report does not claim that. Measured values differ slightly in both directions depending on the parameter and test condition; the tables show the SXS046 operating in the same performance envelope as the AD9361 reference device. Judge the parameters that matter for your band and gain plan from the tables themselves.
What is the one configuration difference between SXS046 and AD9361?
The baseband filter tune divider must be even on the SXS046: Reg1F8[0] and Reg0D6[0] must not be set to 1. Either force bit 0 to 0 in the initialization values, or adjust the baseband bandwidth and reinitialize. All other configuration is handled as described in the report's functional comparison.
Why are some cells in the report blank or marked "/"?
Blank cells and "/" are reproduced from the source measurements and mean the parameter was not measured under that condition — they do not indicate zero.
Can I get samples to run my own comparison?
Yes. Samples are available with a standard lead time of approximately 6 weeks, and Shenxin FAE support can help scope a like-for-like bench comparison on your own board. Contact felix@shenxinic.com or +852 51367095.

Related Articles

AD9361 Pin-to-Pin Alternative: SXS046 RF Transceiver from Shenxin Tech — Parameter comparison, migration notes and FAQ
SXS046 Power-Up Configuration and Calibration Guide — The 16-step initialization sequence and its timing budget
AD9361 to SXS046 Hardware Migration Checklist — Package, power, timing and RF checks before samples go on a board
SXS046 Product Page — Specifications, datasheet, comparison report and manual downloads

Run Your Own SXS046 vs AD9361 Comparison

Samples, the full comparison report and direct FAE support are available now, with a standard lead time of approximately 6 weeks.

View SXS046 Request Samples