S i4438 - C

H I G H -P E R F O R M A N C E, L O W -C U R R E N T T R A N SC E I V E R
Features
 Frequency  Excellent selectivity performance
range = 425–525 MHz 58 dB adjacent channel
 Receive sensitivity = –124 dBm 75 dB blocking at 1 MHz
 Modulation  Antenna diversity and T/R switch
(G)FSK
control
OOK  Highly configurable packet handler
 Max output power  TX and RX 64 byte FIFOs
+20 dBm  Auto frequency control (AFC)
 Low active power consumption  Automatic gain control (AGC)
14 mA RX  Low BOM
 Ultra low current powerdown  Low battery detector
modes Pin Assignments
 Temperature sensor
30 nA shutdown, 40 nA standby
 20-Pin QFN package
 Data rate = 100 bps to 500 kbps
 IEEE 802.15.4g ready
 Preamble Sense Mode
 Suitable for China regulatory (State
6 mA average Rx current at
1.2 kbps Grid)
 Fast wake and hop times
 Power supply = 1.8 to 3.8 V
Applications
 China smart meters

Description

Silicon Laboratories' Si4438 is a high-performance, low-current Patents pending

transceivers covering the sub-GHz frequency bands from 425 to
525 MHz. The Si4438 is targeted at the Chinese smart meter market and
is especially suited for electric meters. This device is footprint- and
pin-compatible with the Si446x radios, which provide industry-leading
performance for worldwide sub-GHz applications. The radios are part of
the EZRadioPRO® family, which includes a complete line of transmitters,
receivers, and transceivers covering a wide range of applications. All
parts offer outstanding sensitivity of –124 dBm while achieving extremely
low active and standby current consumption. The 58 dB adjacent channel
selectivity with 12.5 kHz channel spacing ensures robust receive
operation in harsh RF conditions. The Si4438 offers exceptional output
power of up to +20 dBm with outstanding TX efficiency. The high output
power and sensitivity results in an industry-leading link budget of 144 dB
allowing extended ranges and highly robust communication links.

Rev 1.0 Copyright © 2014 by Silicon Laboratories Si4438-C

Si4 438- C
Functional Block Diagram

Product Freq. Range Max Output TX Current RX Current

Power
Si4438 425–525 MHz +20 dBm 75 mA 13.7 mA

2 Rev 1.0
 Si4 4 38-C
TA B L E O F C O N T E N T S

Section Page
1. Electrical Specifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4
2. Functional Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .11
3. Controller Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12
3.1. Serial Peripheral Interface (SPI) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12
3.2. Fast Response Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .14
3.3. Operating Modes and Timing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .14
3.4. Application Programming Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .18
3.5. Interrupts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .18
3.6. GPIO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .18
4. Modulation and Hardware Configuration Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19
4.1. Modulation Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19
4.2. Hardware Configuration Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19
4.3. Preamble Length . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .21
5. Internal Functional Blocks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .23
5.1. RX Chain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .23
5.2. RX Modem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .23
5.3. Synthesizer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .25
5.4. Transmitter (TX) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .27
5.5. Crystal Oscillator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .29
6. Data Handling and Packet Handler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .31
6.1. RX and TX FIFOs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .31
6.2. Packet Handler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .32
7. RX Modem Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .33
8. Auxiliary Blocks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .33
8.1. Wake-up Timer and 32 kHz Clock Source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .33
8.2. Low Duty Cycle Mode (Auto RX Wake-Up) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .33
8.3. Temperature, Battery Voltage, and Auxiliary ADC . . . . . . . . . . . . . . . . . . . . . . . . . .34
8.4. Low Battery Detector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .35
8.5. Antenna Diversity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .35
8.6. Preamble Sense Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .35
9. Pin Descriptions: Si4438-C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .37
10. Ordering Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .39
11. Package Outline: Si4438 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .40
12. PCB Land Pattern: Si4438 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .41
13. Top Marking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .43
13.1. Si4438 Top Marking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .43
13.2. Top Marking Explanation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .43
Contact Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .44

Rev 1.0 3
Si4 438- C
1. Electrical Specifications

Table 1. DC Characteristics*
Parameter Symbol Test Condition Min Typ Max Unit
Supply Voltage VDD 1.8 3.3 3.8 V
Range
Power Saving Modes IShutdown RC Oscillator, Main Digital Regulator, — 30 — nA
and Low Power Digital Regulator OFF
IStandby Register values maintained and RC — 40 — nA
oscillator/WUT OFF
ISleepRC RC Oscillator/WUT ON and all register values — 740 — nA
maintained, and all other blocks OFF
ISleepXO Sleep current using an external 32 kHz crystal. — 1.7 — μA
ISensor -LBD Low battery detector ON, register values main- — 1 — μA
tained, and all other blocks OFF
IReady Crystal Oscillator and Main Digital Regulator ON, — 1.8 — mA
all other blocks OFF
Preamble Sense Ipsm Duty cycling during preamble search, — 6 — mA
Mode Current 1.2 kbps, 4 byte preamble
Ipsm Fixed 1 s wakeup interval, 50 kbps, 5 byte pream- — 10 — μA
ble
TUNE Mode Current ITune_RX RX Tune — 7.6 — mA
ITune_TX TX Tune — 7.8 — mA
RX Mode Current IRXH — 13.7 — mA
TX Mode Current ITX_+20 +20 dBm output power, class-E match, 490 MHz, — 75 — mA
(Si4438) 3.3 V
*Note: All minimum and maximum values are guaranteed across the recommended operating conditions of supply voltage
and from –40 to +85 °C unless otherwise stated. All typical values apply at VDD = 3.3 V and 25 °C unless otherwise
stated.

4 Rev 1.0
 Si4 4 38-C

Table 2. Synthesizer AC Electrical Characteristics1

Parameter Symbol Test Condition Min Typ Max Unit
Synthesizer Frequency FSYN 425 — 525 MHz
Range (Si4438)
Synthesizer Frequency FRES-525 425–525 MHz — 14.3 — Hz
Resolution2
Synthesizer Settling Time tLOCK Measured from exiting Ready mode with — 50 — μs
XOSC running to any frequency.
Including VCO Calibration.
Phase Noise L(fM) F = 10 kHz, 460 MHz — –109 — dBc/Hz
F = 100 kHz, 460 MHz — –111 — dBc/Hz
F = 1 MHz, 460 MHz — –131 — dBc/Hz
F = 10 MHz, 460 MHz — –141 — dBc/Hz

Notes:
1. All minimum and maximum values are guaranteed across the recommended operating conditions of supply voltage and
from –40 to +85 °C unless otherwise stated. All typical values apply at VDD = 3.3 V and 25 °C unless otherwise stated.
2. Default API setting for modulation deviation resolution is double the typical value specified.

Rev 1.0 5
Si4 438- C
Table 3. Receiver AC Electrical Characteristics1
Parameter Symbol Test Condition Min Typ Max Unit
RX Frequency FRX 425 — 525 MHz
Range (Si4438)
RX Sensitivity2 PRX_0.5 (BER < 0.1%) — –124 — dBm
(500 bps, GFSK, BT = 0.5,
f = 250Hz)2
PRX_40 (BER < 0.1%) — –108 — dBm
(40 kbps, GFSK, BT = 0.5,
f = 20 kHz)2
PRX_100 (BER < 0.1%) — –104 — dBm
(100 kbps, GFSK, BT = 0.5,
f = 50 kHz)1
PRX_9.6 (BER < 0.1%) — –114 — dBm
(9.6 kbps, GFSK, BT = 0.5,
f = 4.8 kHz)2
PRX_OOK (BER < 0.1%, 4.8 kbps, 350 kHz BW, — –108 — dBm
OOK, PN15 data)2
(BER < 0.1%, 40 kbps, 350 kHz BW, — –102 — dBm
OOK, PN15 data)2
(BER < 0.1%, 120 kbps, 350 kHz BW, — –98 — dBm
OOK, PN15 data)2
RX Channel Bandwidth BW 1.1 — 850 kHz
RSSI Resolution RESRSSI — ±0.5 — dB
1-Ch Offset Selectivity, C/I1-CH Desired Ref Signal 3 dB above sensitiv- — –60 — dB
450 MHz2 ity, BER < 0.1%. Interferer is CW, and
desired is modulated with 2.4 kbps
F = 1.2 kHz GFSK with BT = 0.5, RX
channel BW = 4.8 kHz,
channel spacing = 12.5 kHz
Blocking 1 MHz Offset2 1MBLOCK Desired Ref Signal 3 dB above sensitiv- — –77 — dB
ity, BER = 0.1%. Interferer is CW, and
Blocking 8 MHz Offset2 8MBLOCK
desired is modulated with 2.4 kbps,
— –84 — dB
F = 1.2 kHz GFSK with BT = 0.5,
RX channel BW = 4.8 kHz
Image Rejection ImREJ Rejection at the image frequency. — 40 — dB
IF = 468 kHz
Notes:
1. All minimum and maximum values are guaranteed across the recommended operating conditions of supply voltage and
from –40 to +85 °C unless otherwise stated. All typical values apply at VDD = 3.3 V and 25 °C unless otherwise stated.
2. Measured over 50000 bits using PN9 data sequence and data and clock on GPIOs. Sensitivity is expected to be better
if reading data from packet handler FIFO especially at higher data rates.

6 Rev 1.0
 Si4 4 38-C

Table 4. Transmitter AC Electrical Characteristics1

Parameter Symbol Test Condition Min Typ Max Unit
TX Frequency
FTX 425 — 525 MHz
Range
(G)FSK Data Rate2 DRFSK 0.1 — 500 kbps

OOK Data Rate2 DROOK 0.1 — 120 kbps

Modulation Deviation
f525 425–525 MHz — 750 — kHz
Range
Modulation Deviation 14.3
FRES-525 425–525 MHz — — Hz
Resolution3
Output Power Range4 Typical range at 3.3 V
PTX with class E match optimized for best –20 — +20 dBm
PA efficiency.
TX RF Output Steps Using Class E match within 6 dB of max
PRF_OUT — 0.25 — dB
power
TX RF Output Level
PRF_TEMP –40 to +85 C — 2.3 — dB
Variation vs. Temperature
TX RF Output Level
PRF_FREQ — 0.6 — dB
Variation vs. Frequency
Transmit Modulation Gaussian Filtering Bandwith Time
B*T — 0.5 —
Filtering Product
Notes:
1. All minimum and maximum values are guaranteed across the recommended operating conditions of supply voltage and
from –40 to +85 °C unless otherwise stated. All typical values apply at VDD = 3.3 V and 25 °C unless otherwise stated.
2. The maximum data rate is dependent on the XTAL frequency and is calculated as per the formula:
Maximum Symbol Rate = Fxtal/60, where Fxtal is the XTAL frequency (typically 30 MHz).
3. Default API setting for modulation deviation resolution is double the typical value specified.
4. Output power is dependent on matching components and board layout.

Rev 1.0 7
Si4 438- C

Table 5. Auxiliary Block Specifications1

Parameter Symbol Test Condition Min Typ Max Unit
Temperature Sensor TSS — 4.5 — ADC
Sensitivity Codes/
°C
Low Battery Detector LBDRES — 50 — mV
Resolution
Microcontroller Clock FMC Configurable to Fxtal or Fxtal 32.768K — Fxtal Hz
Output Frequency Range2 divided by 2, 3, 7.5, 10, 15, or
30 where Fxtal is the reference
XTAL frequency. In addition,
32.768 kHz is also supported.
Temperature Sensor TEMPCT Programmable setting — 3 — ms
Conversion
XTAL Range3 XTALRange 25 — 32 MHz
30 MHz XTAL Start-Up Time t30M Using XTAL and board layout in — 300 — μs
reference design. Start-up time
will vary with XTAL type and
board layout.
30 MHz XTAL Cap 30MRES — 70 — fF
Resolution
32 kHz XTAL Start-Up Time t32k — 2 — sec
32 kHz Accuracy using 32KRCRES — 2500 — ppm
Internal RC Oscillator
POR Reset Time tPOR — — 6 ms
Notes:
1. All minimum and maximum values are guaranteed across the recommended operating conditions of supply voltage and
from –40 to +85 °C unless otherwise stated. All typical values apply at VDD = 3.3 V and 25 °C unless otherwise stated.
2. Microcontroller clock frequency tested in production at 1 MHz, 30 MHz, 32 MHz, and 32.768 kHz. Other frequencies
tested in bench characterization.
3. XTAL Range tested in production using an external clock source (similar to using a TCXO).

8 Rev 1.0
 Si4 4 38-C

Table 6. Digital IO Specifications (GPIO_x, SCLK, SDO, SDI, nSEL, nIRQ, SDN)1
Parameter Symbol Test Condition Min Typ Max Unit
Rise Time2,3 TRISE 0.1 x VDD to 0.9 x VDD, — 2.3 — ns
CL = 10 pF,
DRV<1:0> = LL
Fall Time3,4 TFALL 0.9 x VDD to 0.1 x VDD, — 2 — ns
CL = 10 pF,
DRV<1:0> = LL
Input Capacitance CIN — 2 — pF
Logic High Level Input Voltage VIH VDD x 0.7 — — V
Logic Low Level Input Voltage VIL — — VDD x 0.3 V
Input Current IIN 0<VIN< VDD –1 — 1 μA
Input Current If Pullup is Activated IINP VIL = 0 V 1 — 4 μA
Drive Strength for Output Low IOmaxLL DRV[1:0] = LL3 — 6.66 — mA
Level IOmaxLH DRV[1:0] = LH3 — 5.03 — mA
IOmaxHL DRV[1:0] = HL3 — 3.16 — mA
IOmaxHH DRV[1:0] = HH3 — 1.13 — mA
Drive Strength for Output High IOmaxLL DRV[1:0] = LL3 — 5.75 — mA
Level IOmaxLH DRV[1:0] = LH3 — 4.37 — mA
IOmaxHL DRV[1:0] = HL3 — 2.73 — mA
IOmaxHH DRV[1:0] = HH3 — 0.96 — mA
Drive Strength for Output High IOmaxLL DRV[1:0] = LL3 — 2.53 — mA
Level for GPIO0 IOmaxLH DRV[1:0] = LH3 — 2.21 — mA
IOmaxHL DRV[1:0] = HL3 — 1.7 — mA
IOmaxHH DRV[1:0] = HH3 — 0.80 — mA
Logic High Level Output Voltage VOH DRV[1:0] = HL VDD x 0.8 — — V
Logic Low Level Output Voltage VOL DRV[1:0] = HL — — VDD x 0.2 V
Notes:
1. All minimum and maximum values are guaranteed across the recommended operating conditions of supply voltage
and from –40 to +85 °C unless otherwise stated. All typical values apply at VDD = 3.3 V and 25 °C unless otherwise
stated.
2. 6.7 ns is typical for GPIO0 rise time.
3. Assuming VDD = 3.3 V, drive strength is specified at Voh (min) = 2.64 V and Vol(max) = 0.66 V at room temperature.
4. 2.4 ns is typical for GPIO0 fall time.

Rev 1.0 9
Si4 438- C
Table 7. Thermal Operating Characteristics
Parameter Value Unit
Operating Ambient Temperature Range TA –40 to +85 C
Thermal Impedance JA 25 C/W
Junction Temperature TJMAX +105 C
Storage Temperature Range TSTG –55 to +150 C

Table 8. Absolute Maximum Ratings*

Parameter Value Unit
VDD to GND –0.3, +3.8 V
Instantaneous VRF-peak to GND on TX Output Pin –0.3, +8.0 V
Sustained VRF-peak to GND on TX Output Pin –0.3, +6.5 V
Voltage on Digital Control Inputs –0.3, VDD + 0.3 V
Voltage on Analog Inputs –0.3, VDD + 0.3 V
Voltage on XIN Input when using a TCXO –0.7, VDD + 0.3 V
RX Input Power +10 dBm
*Note: Stresses beyond those listed under “Absolute Maximum Ratings” may cause permanent damage to the device. These
are stress ratings only and functional operation of the device at or beyond these ratings in the operational sections of
the specifications is not implied. Exposure to absolute maximum rating conditions for extended periods may affect
device reliability. Power Amplifier may be damaged if switched on without proper load or termination connected. TX
matching network design will influence TX VRF-peak on TX output pin. Caution: ESD sensitive device.

10 Rev 1.0
 Si4 4 38-C
2. Functional Description
The Si4438 devices are high-performance, low-current, wireless ISM transceivers that cover the sub-GHz bands.
The wide operating voltage range of 1.8–3.8 V and low current consumption make the Si4438 an ideal solution for
battery powered applications. The Si4438 operates as a time division duplexing (TDD) transceiver where the
device alternately transmits and receives data packets. The device uses a single-conversion mixer to downconvert
the 2-level FSK/GFSK or OOK modulated receive signal to a low IF frequency. Following a programmable gain
amplifier (PGA) the signal is converted to the digital domain by a high performance  ADC allowing filtering,
demodulation, slicing, and packet handling to be performed in the built-in DSP increasing the receiver’s
performance and flexibility versus analog based architectures. The demodulated signal is output to the system
MCU through a programmable GPIO or via the standard SPI bus by reading the 64-byte RX FIFO.
A single high precision local oscillator (LO) is used for both transmit and receive modes since the transmitter and
receiver do not operate at the same time. The LO is generated by an integrated VCO and  Fractional-N PLL
synthesizer. The synthesizer is designed to support configurable data rates from 100 bps to 500 kbps. The transmit
FSK data is modulated directly into the  data stream and can be shaped by a Gaussian low-pass filter to reduce
unwanted spectral content.
The Si4438 contains a power amplifier (PA) that supports output power up to +20 dBm with very high efficiency,
consuming only 75 mA. The integrated +20 dBm power amplifier can also be used to compensate for the reduced
performance of a lower cost, lower performance antenna or antenna with size constraints due to a small
form-factor. Competing solutions require large and expensive external PAs to achieve comparable performance.
The PA is single-ended to allow for easy antenna matching and low BOM cost. The PA incorporates automatic
ramp-up and ramp-down control to reduce unwanted spectral spreading. The Si4438 family supports TX/RX switch
control, and antenna diversity switch control to extend the link range and improve performance. Built-in antenna
diversity can be used to further extend range and enhance performance. Antenna diversity is completely integrated
into the Si4438 and can improve the system link budget by 8–10 dB, resulting in substantial range increases under
adverse environmental conditions. A highly configurable packet handler allows for autonomous encoding/decoding
of nearly any packet structure. Additional system features, such as an automatic wake-up timer, low battery
detector, 64 byte TX/RX FIFOs, and preamble detection, reduce overall current consumption and allows for the
use of lower-cost system MCUs. An integrated temperature sensor, power-on-reset (POR), and GPIOs further
reduce overall system cost and size. The Si4438 is designed to work with an MCU, crystal, and a few passive
components to create a very low-cost system.

Rev 1.0 11
Si4 438- C
3. Controller Interface
3.1. Serial Peripheral Interface (SPI)
The Si4438 communicates with the host MCU over a standard 4-wire serial peripheral interface (SPI): SCLK, SDI,
SDO, and nSEL. The SPI interface is designed to operate at a maximum of 10 MHz. The SPI timing parameters
are demonstrated in Table 9. The host MCU writes data over the SDI pin and can read data from the device on the
SDO output pin. Figure 1 demonstrates an SPI write command. The nSEL pin should go low to initiate the SPI
command. The first byte of SDI data will be one of the firmware commands followed by n bytes of parameter data
which will be variable depending on the specific command. The rising edges of SCLK should be aligned with the
center of the SDI data.

Table 9. Serial Interface Timing Parameters

Symbol Parameter Min Max Diagram
(ns) (ns)
tCH Clock high time 40

tCL Clock low time 40

tDS Data setup time 20

tDH Data hold time 20

tDD Output data delay time 43

tDE Output disable time 45

tSS Select setup time 20

tSH Select hold time 50

tSW Select high period 80

*Note: CL = 10 pF; VDD = 1.8 V; SDO Drive strength setting = 10.

Figure 1. SPI Write Command

The Si4438 contains an internal MCU which controls all the internal functions of the radio. For SPI read commands
a typical MCU flow of checking clear-to-send (CTS) is used to make sure the internal MCU has executed the
command and prepared the data to be output over the SDO pin. Figure 2 demonstrates the general flow of an SPI
read command. Once the CTS value reads FFh then the read data is ready to be clocked out to the host MCU. The
typical time for a valid FFh CTS reading is 20 μs. Figure 3 demonstrates the remaining read cycle after CTS is set
to FFh. The internal MCU will clock out the SDO data on the negative edge so the host MCU should process the
SDO data on the rising edge of SCLK.

12 Rev 1.0
 Si4 4 38-C

Figure 2. SPI Read Command—Check CTS Value

Figure 3. SPI Read Command—Clock Out Read Data

Rev 1.0 13
Si4 438- C
3.2. Fast Response Registers
The fast response registers are registers that can be read immediately without the requirement to monitor and
check CTS. There are four fast response registers that can be programmed for a specific function. The fast
response registers can be read through API commands, 0x50 for Fast Response A, 0x51 for Fast Response B,
0x53 for Fast Response C, and 0x57 for Fast Response D. The fast response registers can be configured by the
“FRR_CTL_X_MODE” properties.
The fast response registers may be read in a burst fashion. After the initial 16 clock cycles, each additional eight
clock cycles will clock out the contents of the next fast response register in a circular fashion. The value of the
FRRs will not be updated unless NSEL is toggled.
3.3. Operating Modes and Timing
The primary states of the Si4438 are shown in Figure 4. The shutdown state completely shuts down the radio to
minimize current consumption. Standby/Sleep, SPI Active, Ready, TX Tune, and RX tune are available to optimize
the current consumption and response time to RX/TX for a given application. API commands START_RX,
START_TX, and CHANGE_STATE control the operating state with the exception of shutdown which is controlled
by SDN, pin 1. Table 10 shows each of the operating modes with the time required to reach either RX or TX mode
as well as the current consumption of each mode. The times in Table 9 are measured from the rising edge of nSEL
until the chip is in the desired state. Note that these times are indicative of state transition timing but are not
guaranteed and should only be used as a reference data point. An automatic sequencer will put the chip into RX or
TX from any state. It is not necessary to manually step through the states. To simplify the diagram it is not shown
but any of the lower power states can be returned to automatically after RX or TX.

Figure 4. State Machine Diagram

14 Rev 1.0
 Si4 4 38-C

Table 10. Operating State Response Time and Current Consumption

Response Time to Current in State
State/Mode
TX RX /Mode
Shutdown State 15 ms 15 ms 30 nA
Standby State 504 μs 516 μs 40 nA
Sleep State 504 μs 516 μs 740 nA
SPI Active State 288 μs 296 μs 1.35 mA
Ready State 108 μs 120 μs 1.8 mA
TX Tune State 60 μs — 7.8 mA
RX Tune State — 84 μs 7.6 mA
TX State — 132 μs 75 mA @ +20 dBm
RX State 120 μs 75 μs 13.7 mA

Figure 5 shows the POR timing and voltage requirements. The power consumption (battery life) depends on the
duty cycle of the application or how often the part is in either Rx or Tx state. In most applications the utilization of
the standby state will be most advantageous for battery life but for very low duty cycle applications shutdown will
have an advantage. For the fastest timing the next state can be selected in the START_RX or START_TX API
commands to minimize SPI transactions and internal MCU processing.
3.3.1. Power on Reset (POR)
A Power On Reset (POR) sequence is used to boot the device up from a fully off or shutdown state. To execute this
process, VDD must ramp within 1ms and must remain applied to the device for at least 10ms. If VDD is removed,
then it must stay below 0.15V for at least 10ms before being applied again. Please see Figure 5 and Table 11 for
details.

Figure 5. POR Timing Diagram

Rev 1.0 15
Si4 438- C

Table 11. POR Timing

Variable Description Min Typ Max Units
tPORH High time for VDD to fully settle POR circuit 10 ms
tPORL Low time for VDD to enable POR 10 ms
VRRH Voltage for successful POR 90%*Vdd V
VRRL Starting Voltage for successful POR 0 150 mV
tSR Slew rate of VDD for successful POR 1 ms

3.3.2. Shutdown State

The shutdown state is the lowest current consumption state of the device with nominally less than 30 nA of current
consumption. The shutdown state may be entered by driving the SDN pin (Pin 1) high. The SDN pin should be held
low in all states except the shutdown state. In the shutdown state, the contents of the registers are lost and there is
no SPI access. When coming out of the shutdown state a power on reset (POR) will be initiated along with the
internal calibrations. After the POR the POWER_UP command is required to initialize the radio. The SDN pin
needs to be held high for at least 10us before driving low again so that internal capacitors can discharge. Not
holding the SDN high for this period of time may cause the POR to be missed and the device to boot up incorrectly.
If POR timing and voltage requirements cannot be met, it is highly recommended that SDN be controlled using the
host processor rather than tying it to GND on the board.
3.3.3. Standby State
Standby state has the lowest current consumption with the exception of shutdown but has much faster response
time to RX or TX mode. In most cases standby should be used as the low power state. In this state the register
values are maintained with all other blocks disabled. The SPI is accessible during this mode but any SPI event,
including FIFO R/W, will enable an internal boot oscillator and automatically move the part to SPI active state. After
an SPI event the host will need to re-command the device back to standby through the “Change State” API
command to achieve the 40 nA current consumption. If an interrupt has occurred (i.e., the nIRQ pin = 0) the
interrupt registers must be read to achieve the minimum current consumption of this mode.
3.3.4. Sleep State
Sleep state is the same as standby state but the wake-up-timer and a 32 kHz clock source are enabled. The
source of the 32 kHz clock can either be an internal 32 kHz RC oscillator which is periodically calibrated or a
32 kHz oscillator using an external XTAL.The SPI is accessible during this mode but an SPI event will enable an
internal boot oscillator and automatically move the part to SPI active mode. After an SPI event the host will need to
re-command the device back to sleep. If an interrupt has occurred (i.e., the nIRQ pin = 0) the interrupt registers
must be read to achieve the minimum current consumption of this mode.
3.3.5. SPI Active State
In SPI active state the SPI and a boot up oscillator are enabled. After SPI transactions during either standby or
sleep the device will not automatically return to these states. A “Change State” API command will be required to
return to either the standby or sleep modes.
3.3.6. Ready State
Ready state is designed to give a fast transition time to TX or RX state with reasonable current consumption. In this
mode the Crystal oscillator remains enabled reducing the time required to switch to TX or RX mode by eliminating
the crystal start-up time.
3.3.7. TX State
The TX state may be entered from any of the state with the “Start TX” or “Change State” API commands. A built-in
sequencer takes care of all the actions required to transition between states from enabling the crystal oscillator to
ramping up the PA. The following sequence of events will occur automatically when going from standby to TX state.
1. Enable internal LDOs.
2. Start up crystal oscillator and wait until ready (controlled by an internal timer).

16 Rev 1.0
 Si4 4 38-C
3. Enable PLL.
4. Calibrate VCO/PLL.
5. Wait until PLL settles to required transmit frequency (controlled by an internal timer).
6. Activate power amplifier and wait until power ramping is completed (controlled by an internal timer).
7. Transmit packet.
Steps in this sequence may be eliminated depending on which state the chip is configured to prior to commanding
to TX. By default, the VCO and PLL are calibrated every time the PLL is enabled. When the START_TX API
command is utilized the next state may be defined to ensure optimal timing and turnaround.
Figure 6 shows an example of the commands and timing for the START_TX command. CTS will go high as soon
as the sequencer puts the part into TX state. As the sequencer is stepping through the events listed above, CTS
will be low and no new commands or property changes are allowed. If the Fast Response (FRR) or nIRQ is used to
monitor the current state there will be slight delay caused by the internal hardware from when the event actually
occurs to when the transition occurs on the FRR or nIRQ. The time from entering TX state to when the FRR will
update is 5 μs and the time to when the nIRQ will transition is 13 μs. If a GPIO is programmed for TX state or used
as control for a transmit/receive switch (TR switch) there is no delay.

Figure 6. Start_TX Commands and Timing

3.3.8. RX State
The RX state may be entered from any of the other states by using the “Start RX” or “Change State” API command.
A built-in sequencer takes care of all the actions required to transition between states. The following sequence of
events will occur automatically to get the chip into RX mode when going from standby to RX state:
1. Enable the digital LDO and the analog LDOs.
2. Start up crystal oscillator and wait until ready (controlled by an internal timer).
3. Enable PLL.
4. Calibrate VCO
5. Wait until PLL settles to required receive frequency (controlled by an internal timer).
6. Enable receiver circuits: LNA, mixers, and ADC.
7. Enable receive mode in the digital modem.
Depending on the configuration of the radio, all or some of the following functions will be performed automatically
by the digital modem: AGC, AFC (optional), update status registers, bit synchronization, packet handling (optional)
including sync word, header check, and CRC. Similar to the TX state, the next state after RX may be defined in the
“Start RX” API command. The START_RX commands and timing will be equivalent to the timing shown in Figure 6.

Rev 1.0 17
Si4 438- C
3.4. Application Programming Interface
The host MCU communicates with an application programming interface (API) embedded inside the device. The
API is divided into two sections, commands and properties. The commands are used to control the chip and
retrieve its status. The properties are general configurations which will change infrequently. For API description
details, refer to the EZRadioPRO API Documentation.zip file available on www.silabs.com.
3.5. Interrupts
The Si4438 is capable of generating an interrupt signal when certain events occur. The chip notifies the
microcontroller that an interrupt event has occurred by setting the nIRQ output pin LOW = 0. This interrupt signal
will be generated when any one (or more) of the interrupt events (corresponding to the Interrupt Status bits) occur.
The nIRQ pin will remain low until the microcontroller clears all the interrupts. The nIRQ output signal will then be
reset until the next change in status is detected.
The interrupts sources are grouped into three groups: packet handler, chip status, and modem. The individual
interrupts in these groups can be enabled/disabled in the interrupt property registers. An interrupt must be enabled
for it to trigger an event on the nIRQ pin. The interrupt group must be enabled as well as the individual interrupts in
API properties described in the API documentation. Once an interrupt event occurs and the nIRQ pin is low there
are two ways to read and clear the interrupts. All of the interrupts may be read and cleared in the
“GET_INT_STATUS” API command. By default all interrupts will be cleared once read. If only specific interrupts
want to be read in the fastest possible method the individual interrupt groups (Packet Handler, Chip Status,
Modem) may be read and cleared by the “GET_MODEM_STATUS”, “GET_PH_STATUS” (packet handler), and
“GET_CHIP_STATUS” API commands. The instantaneous status of a specific function maybe read if the specific
interrupt is enabled or disabled. The status results are provided after the interrupts and can be read with the same
commands as the interrupts. The status bits will give the current state of the function whether the interrupt is
enabled or not. The fast response registers can also give information about the interrupt groups but reading the
fast response registers will not clear the interrupt and reset the nIRQ pin.
3.6. GPIO
Four general purpose IO pins are available to utilize in the application. The GPIO are configured by the
GPIO_PIN_CFG command in address 13h. For a complete list of the GPIO options please see the API guide.
GPIO pins 0 and 1 should be used for active signals such as data or clock. GPIO pins 2 and 3 have more
susceptibility to generating spurious in the synthesizer than pins 0 and 1. The drive strength of the GPIOs can be
adjusted with the GEN_CONFIG parameter in the GPIO_PIN_CFG command. By default the drive strength is set
to minimum. The default configuration for the GPIOs and the state during SDN is shown below in Table 12.The
state of the IO during shutdown is also shown in Table 12. As indicated previously in Table 6, GPIO 0 has lower
drive strength than the other GPIOs.

Table 12. GPIOs

Pin SDN State POR Default
GPIO0 0 POR
GPIO1 0 CTS
GPIO2 0 POR
GPIO3 0 POR
nIRQ resistive VDD pull-up nIRQ
SDO resistive VDD pull-up SDO
SDI High Z SDI
SCLK High Z SCLK
NSEL High Z NSEL

18 Rev 1.0
 Si4 4 38-C
4. Modulation and Hardware Configuration Options
The Si4438 supports three different modulation options and can be used in various configurations to tailor the
device to any specific application or legacy system for drop in replacement. The modulation and configuration
options are set in property, MODEM_MOD_TYPE. Refer to the EZRadioPRO API Documentation.zip file available
on www.silabs.com for details.
4.1. Modulation Types
The Si4438 supports five different modulation options: Gaussian frequency shift keying (GFSK), frequency-shift
keying (FSK), on-off keying (OOK). Minimum shift keying (MSK) can also be created by using GFSK settings.
GFSK is the recommended modulation type as it provides the best performance and cleanest modulation
spectrum. The modulation type is set by the “MOD_TYPE[2:0]” registers in the “MODEM_MOD_TYPE” API
property. A continuous-wave (CW) carrier may also be selected for RF evaluation purposes. The modulation
source may also be selected to be a pseudo-random source for evaluation purposes.
4.2. Hardware Configuration Options
There are different receive demodulator options to optimize the performance and mutually-exclusive options for
how the RX/TX data is transferred from the host MCU to the RF device.
4.2.1. Receive Demodulator Options
There are multiple demodulators integrated into the device to optimize the performance for different applications,
modulation formats, and packet structures. The calculator built into WDS will choose the optimal demodulator
based on the input criteria.
4.2.1.1. Synchronous Demodulator
The synchronous demodulator's internal frequency error estimator acquires the frequency error based on a
101010 preamble structure. The bit clock recovery circuit locks to the incoming data stream within four transactions
of a “10” or “01” bit stream. The synchronous demodulator gives optimal performance for 2-level FSK or GFSK
modulation that has a modulation index less than 2.
4.2.1.2. Asynchronous Demodulator
The asynchronous demodulator should be used OOK modulation and for FSK/GFSK under one or more of the
following conditions:
Modulation index > 2
Non-standard preamble (not 1010101... pattern)
When the modulation index exceeds 2, the asynchronous demodulator has better sensitivity compared to the
synchronous demodulator. An internal deglitch circuit provides a glitch-free data output and a data clock signal to
simplify the interface to the host. There is no requirement to perform deglitching in the host MCU. The
asynchronous demodulator will typically be utilized for legacy systems and will have many performance benefits
over devices used in legacy designs. Unlike the Si4432/31 solution for non-standard packet structures, there is no
requirement to perform deglitching on the data in the host MCU. Glitch-free data is output from Si4438 devices,
and a sample clock for the asynchronous data can also be supplied to the host MCU; so, oversampling or bit clock
recovery is not required by the host MCU. There are multiple detector options in the asynchronous demodulator
block, which will be selected based upon the options entered into the WDS calculator. The asynchronous
demodulator's internal frequency error estimator is able to acquire the frequency error based on any preamble
structure.
4.2.2. RX/TX Data Interface With MCU
There are two different options for transferring the data from the RF device to the host MCU. FIFO mode uses the
SPI interface to transfer the data, while direct mode transfers the data in real time over GPIO.

Rev 1.0 19
Si4 438- C
4.2.2.1. FIFO Mode
In FIFO mode, the transmit and receive data is stored in integrated FIFO register memory. The TX FIFO is
accessed by writing Command 66h followed directly by the data/clk that the host wants to write into the TX FIFO.
The RX FIFO is accessed by writing command 77h followed by the number of clock cycles of data the host would
like to read out of the RX FIFO. The RX data will be clocked out onto the SDO pin.
In TX FIFO mode, the data bytes stored in FIFO memory are “packaged” together with other fields and bytes of
information to construct the final transmit packet structure. These other potential fields include the Preamble, Sync
word, and CRC checksum. In TX mode, the packet structure may be highly customized by enabling or disabling
individual fields; for example, it is possible to disable both the Preamble and Sync Word fields and to load the entire
packet structure into FIFO memory. For further information on the configuration of the FIFOs for a specific
application or packet size, see "6. Data Handling and Packet Handler" on page 31. In RX mode, the Packet
Handler must be enabled to allow storage of received data bytes into RX FIFO memory. The Packet Handler is
required to detect the Sync Word, and proper detection of the Sync Word is required to determine the start of the
Payload. All bytes after the Sync Word are stored in RX FIFO memory except for the CRC checksum and
(optionally) the variable packet length byte(s). When the FIFO is being used in RX mode, all of the received data
may still be observed directly (in real time) by properly programming a GPIO pin as the RXDATA output pin; this
can be quite useful during application development. When in FIFO mode, the chip will automatically exit the TX or
RX State when either the PACKET_SENT or PACKET_RX interrupt occurs. The chip will return to the state
programmed in the argument of the “START TX” or “START RX” API command, TXCOMPLETE_STATE[3:0] or
RXVALID_STATE[3:0]. For example, the chip may be placed into READY mode after a TX packet by sending the
“START TX” command and by writing 30h to the TXCOMPLETE_STATE[3:0] argument. The chip will transmit all of
the contents of the FIFO, and the PACKET_SENT interrupt will occur. When this event occurs, the chip will return
to the READY state as defined by TXCOMPLETE_STATE[3:0] = 30h.
4.2.2.2. FIFO Direct Mode (Infinite Receive)
In some applications, there is a need to receive extremely long packets (greater than 40 kB) while relying on
preamble and sync word detection from the on-chip packet handler. In these cases, the packet length is unknown,
and the device will load the bits after the sync word into the RX FIFO forever. Other features, such as Data
Whitening, CRC, Manchester, etc., are supported in this mode, but CRC calculation is not because the end of
packet is unknown to the device. The RX data and clock are also available on GPIO pins. The host MCU will need
to reset the packet handler by issuing a START_RX to begin searching for a new packet.
4.2.2.3. Direct Mode
For legacy systems that perform packet handling within the host MCU or other baseband chip, it may not be
desirable to use the FIFO. For this scenario, a Direct mode is provided, which bypasses the FIFOs entirely. In TX
Direct mode, the TX modulation data is applied to an input pin of the chip and processed in “real time” (i.e., not
stored in a register for transmission at a later time). Any of the GPIOs may be configured for use as the TX Data
input function. Furthermore, an additional pin may be required for a TX Clock output function if GFSK modulation is
desired (only the TX Data input pin is required for FSK). To achieve direct mode, the GPIO must be configured in
the “GPIO_PIN_CFG” API command as well as the “MODEM_MOD_TYPE” API property. For GFSK,
“TX_DIRECT_MODE_TYPE” must be set to Synchronous. For 2FSK or OOK, the type can be set to asynchronous
or synchronous. The MOD_SOURCE[1:0] should be set to 01h for are all direct mode configurations. In RX Direct
mode, the RX Data and RX Clock can be programmed for direct (real-time) output to GPIO pins. The
microcontroller may then process the RX data without using the FIFO or packet handler functions of the RFIC.

20 Rev 1.0
 Si4 4 38-C
4.3. Preamble Length
4.3.1. Digital Signal Arrival Detector
Traditional preamble detection requires 20 bits to detect preamble. This device introduces a new approach to
signal detection that can detect a preamble pattern in as little as one byte. If AFC is enabled, a preamble length of
two bytes is sufficient to reliably detect signal arrival and settle a one shot AFC. The impact of this is significant for
low-power solutions as it reduces the amount of time the receiver has to stay active to detect the preamble. This
feature is used with Preamble Sense Mode (see "8.6. Preamble Sense Mode" on page 35) and the latest WMBus
N modes as well as with features, such as frequency hopping, which may use signal arrival as a condition to hop.
The traditional preamble detector is also available to maintain backward compatibility. Note that the DSA is using
the RSSI jump detector. When used for collision detection, the RSSI jump detector may need to be reconfigured
after preamble detection. Refer to the API documentation for details on how to configure the device to use the
signal arrival detector.
4.3.2. Traditional Preamble Detection
Optimal performance of the chip is obtained by qualifying reception of a valid Preamble pattern prior to continuing
with reception of the remainder of the packet (e.g., Sync Word and Payload). Reception of the Preamble is
considered valid when a minimum number of consecutive bits of 101010... pattern have been received; the
required threshold for preamble detection is specified by the RX_THRESH[6:0] field in the
PREAMBLE_CONFIG_STD_1 property. The appropriate value of the detection threshold depends upon the
system application and typically trades off speed of acquisition against the probability of false detection. If the
detection threshold is set too low, the chip may readily detect the short pattern within noise; the chip then proceeds
to attempt to detect the remainder of the non-existent packet, with the result that the arrival of an actual valid
packet may be missed. If the detection threshold is set too high, the required number of transmitted Preamble bits
must be increased accordingly, leading to longer packet lengths and shorter battery life. A preamble detection
threshold value of 20 bits is suitable for most applications. The total length of the transmitted Preamble field must
be at least equal to the receive preamble detection threshold, plus an additional number of bits to allow for
acquisition of bit timing and settling of the AFC algorithm. The recommended preamble detection thresholds and
preamble lengths for a variety of operational modes are listed in Table 13. Configuration of the preamble detection
threshold in the RX_THRESH[6:0] field is only required for reception of a standard Preamble pattern (i.e., 101010...
pattern). Reception of a repetitive but non-standard Preamble pattern is also supported in the chip but is configured
through the PREAMBLE_CONFIG_NSTD and PREAMBLE_PATTERN properties.

Rev 1.0 21
Si4 438- C
Table 13. Recommended Preamble Length
Antenna Recommended Recommended Preamble
Mode AFC Preamble Type
Diversity Preamble Length Detection Threshold
(G)FSK Disabled Disabled Standard 4 Bytes 20 bits
(G)FSK Enabled Disabled Standard 5 Bytes 20 bits
(G)FSK Disabled Disabled Non-standard 2 Bytes 0 bits
(G)FSK Enabled Non-standard Not Supported
(G)FSK Disabled Enabled Standard 7 Bytes 24 bits
(G)FSK Enabled Enabled Standard 8 Bytes 24 bits
OOK Disabled Disabled Standard 4 Bytes 20 bits
OOK Disabled Disabled Non-standard 2 Bytes 0 bits
OOK Enabled Not Supported
Notes:
1. The recommended preamble length and preamble detection thresholds listed above are to achieve 0% PER. They may
be shortened when occasional packet errors are tolerable.
2. All recommended preamble lengths and detection thresholds include AGC and BCR settling times.
3. “Standard” preamble type should be set for an alternating data sequence at the max data rate (…10101010…)
4. “Non-standard” preamble type can be set for any preamble type including …10101010...
5. When preamble detection threshold = 0, sync word needs to be 3 Bytes to avoid false syncs. When only a 2 Byte sync
word is available the sync word detection can be extended by including the last preamble Byte into the RX sync word
setting.

22 Rev 1.0
 Si4 4 38-C
5. Internal Functional Blocks
The following sections provide an overview to the key internal blocks and features.
5.1. RX Chain
The internal low-noise amplifier (LNA) is designed to be a wide-band LNA that can be matched with three external
discrete components to cover any common range of frequencies in the sub-GHz band. The LNA has extremely low
noise to suppress the noise of the following stages and achieve optimal sensitivity; so, no external gain or front-end
modules are necessary. The LNA has gain control, which is controlled by the internal automatic gain control (AGC)
algorithm. The LNA is followed by an I-Q mixer, filter, programmable gain amplifier (PGA), and ADC. The I-Q
mixers downconvert the signal to an intermediate frequency. The PGA then boosts the gain to be within dynamic
range of the ADC. The ADC rejects out-of-band blockers and converts the signal to the digital domain where
filtering, demodulation, and processing is performed. Peak detectors are integrated at the output of the LNA and
PGA for use in the AGC algorithm.
5.2. RX Modem
Using high-performance ADCs allows channel filtering, image rejection, and demodulation to be performed in the
digital domain, which allows for flexibility in optimizing the device for particular applications. The digital modem
performs the following functions:
Channel selection filter
TX modulation
RX demodulation
Automatic Gain Control (AGC)
Preamble detection
Invalid preamble detection
Radio signal strength indicator (RSSI)
Automatic frequency compensation (AFC)
Cyclic redundancy check (CRC)
The digital channel filter and demodulator are optimized for ultra-low-power consumption and are highly
configurable. Supported modulation types are GFSK, FSK, GMSK, and OOK. The channel filter can be configured
to support bandwidths ranging from 850 down to 1.1 kHz. A large variety of data rates are supported ranging from
100 bps up to 500 kbps. The configurable preamble detector is used with the synchronous demodulator to improve
the reliability of the sync-word detection. Preamble detection can be skipped using only sync detection, which is a
valuable feature of the asynchronous demodulator when very short preambles are used in protocols, such as
MBus. The received signal strength indicator (RSSI) provides a measure of the signal strength received on the
tuned channel. The resolution of the RSSI is 0.5 dB. This high-resolution RSSI enables accurate channel power
measurements for clear channel assessment (CCA), carrier sense (CS), and listen before talk (LBT) functionality.
A comprehensive programmable packet handler including key features of Silicon Labs’ EZMAC is integrated to
create a variety of communication topologies ranging from peer-to-peer networks to mesh networks. The extensive
programmability of the packet header allows for advanced packet filtering, which, in turn enables a mix of
broadcast, group, and point-to-point communication. A wireless communication channel can be corrupted by noise
and interference, so it is important to know if the received data is free of errors. A cyclic redundancy check (CRC)
is used to detect the presence of erroneous bits in each packet. A CRC is computed and appended at the end of
each transmitted packet and verified by the receiver to confirm that no errors have occurred. The packet handler
and CRC can significantly reduce the load on the system microcontroller allowing for a simpler and cheaper
microcontroller. The digital modem includes the TX modulator, which converts the TX data bits into the
corresponding stream of digital modulation values to be summed with the fractional input to the sigma-delta
modulator. This modulation approach results in highly accurate resolution of the frequency deviation. A Gaussian
filter is implemented to support GFSK, considerably reducing the energy in adjacent channels.
5.2.1. Automatic Gain Control (AGC)
The AGC algorithm is implemented digitally using an advanced control loop optimized for fast response time. The
AGC occurs within a single bit or in less than 2 μs. Peak detectors at the output of the LNA and PGA allow for
optimal adjustment of the LNA gain and PGA gain to optimize IM3, selectivity, and sensitivity performance.

Rev 1.0 23
Si4 438- C
5.2.2. Auto Frequency Correction (AFC)
Frequency mistuning caused by crystal inaccuracies can be compensated for by enabling the digital automatic
frequency control (AFC) in receive mode. There are two types of integrated frequency compensation: modem
frequency compensation, and AFC by adjusting the PLL frequency. With AFC disabled, the modem compensation
can correct for frequency offsets up to ±0.25 times the IF bandwidth. When the AFC is enabled, the received signal
will be centered in the pass-band of the IF filter, providing optimal sensitivity and selectivity over a wider range of
frequency offsets up to ±0.35 times the IF bandwidth. When AFC is enabled, the preamble length needs to be long
enough to settle the AFC. As shown in Table 13 on page 22, an additional byte of preamble is typically required to
settle the AFC.
5.2.3. Received Signal Strength Indicator
The received signal strength indicator (RSSI) is an estimate of the signal strength in the channel to which the
receiver is tuned. The RSSI measurement is done after the channel filter, so it is only a measurement of the
desired or undesired in-band signal power. There are two different methods for reading the RSSI value and several
different options for configuring the RSSI value that is returned. The fastest method for reading the RSSI is to
configure one of the four fast response registers (FRR) to return a latched RSSI value. The latched RSSI value is
measured once per packet and is latched at a configurable amount of time after RX mode is entered. The fast
response registers can be read in 16 SPI clock cycles with no requirement to wait for CTS. The RSSI value may
also be read out of the GET_MODEM_STATUS command. In this command, both the current RSSI and the latched
RSSI are available. The current RSSI value represents the signal strength at the instant in time the
GET_MODEM_STATUS command is processed and may be read multiple times per packet. Reading the RSSI in
the GET_MODEM_STATUS command takes longer than reading the RSSI out of the fast response register. After
the initial command, it will take 33 μs for CTS to be set and then the four or five bytes of SPI clock cycles to read
out the respective current or latched RSSI values.
The RSSI configuration options are set in the MODEM_RSSI_CONTROL API property. The latched RSSI value
may be latched and stored based on the following events: preamble detection, sync detection, or a configurable
number of bit times measured after the start of RX mode (minimum of 4 bit times). The requirement for four bit
times is determined by the processing delay and settling through the modem and digital channel filter. In
MODEM_RSSI_CONTROL, the RSSI may be defined to update every bit period or to be averaged and updated
every four bit periods. If RSSI averaging over four bits is enabled, the latched RSSI value will be delayed to a
minimum of 7 bits after the start of RX mode to allow for the averaging. The latched RSSI values are cleared when
entering RX mode so they may be read after the packet is received or after dropping back to standby mode. If the
RSSI value has been cleared by the start of RX but not latched yet, a value of 0 will be returned if it is attempted to
be read.
The RSSI value read by the API may be translated into dBm by the following linear equation:

RF_Input_Level_dBm =  RSSI_value
------------------------------- – MODEM_RSSI_COMP – 70
 2 
The MODEM_RSSI_COMP property provides for fine adjustment of the relationship between the actual RF input
level (in dBm) and the returned RSSI value. That is, adjustment of this property allows the user to shift the RSSI vs
RF Input Power curve up and down. This may be desirable to compensate for differences in front-end insertion loss
between multiple designs (e.g., due to the presence of a SAW preselection filter, or an RF switch). A value of
MODEM_RSSI_COMP = 0x40 = 64d is appropriate for most applications.
Clear channel assessment (CCA) or RSSI threshold detection is also available. An RSSI threshold may be set in
the MODEM_RSSI_THRESH API property. If the Current RSSI value is above this threshold, an interrupt or GPIO
may notify the host. Both the latched version and asynchronous version of this threshold are available on any of
the GPIOs. Automatic fast hopping based on RSSI is available. See "5.3.1.2. Automatic RX Hopping and Hop
Table" on page 26. Clear channel assessment (CCA) or RSSI threshold detection is also available. An RSSI
threshold may be set in the MODEM_RSSI_THRESH API property. If the RSSI value is above this threshold, an
interrupt or GPIO may notify the host.
Both the latched version and asynchronous version of this threshold are available on any of the GPIOs. Automatic
fast hopping based on RSSI is available. See “5.3.1.2. Automatic RX Hopping and Hop Table”.

24 Rev 1.0
 Si4 4 38-C
5.2.4. RSSI Jump Indicator (Collision Detection)
The chip is capable of detecting a jump in RSSI in either direction (i.e., either a signal increase or a signal
decrease). Both polarities of jump detection may be enabled simultaneously, resulting in detection of a Jump-Up or
Jump-Down event. This may be used to detect whether a secondary interfering signal (desired or undesired) has
“collided” with reception of the current packet. An interrupt flag or GPIO pin may be configured to notify the host
MCU of the Jump event. The change in RSSI level required to trigger the Jump event is programmable through the
MODEM_RSSI_JUMP_THRESH API property.
The chip may be configured to reset the RX state machine upon detection of an RSSI Jump, and thus to
automatically begin reacquisition of the packet. The chip may also be configured to generate an interrupt. This
functionality is intended to detect an abrupt change in RSSI level and to not respond to a slow, gradual change in
RSSI level. This is accomplished by comparing the difference in RSSI level over a programmable time period. In
this fashion, the chip effectively evaluates the slope of the change in RSSI level.
The arrival of a desired packet (i.e., the transition from receiving noise to receiving a valid signal) will likely be
detected as an RSSI Jump event. For this reason, it is recommended to enable this feature in mid-packet (i.e., after
signal qualification, such as PREAMBLE_VALID.) Refer to the API documentation for configuration options.
5.3. Synthesizer
An integrated Sigma Delta () Fractional-N PLL synthesizer capable of operating over 425–525 MHz. Using a 
synthesizer has many advantages; it provides flexibility in choosing data rate, deviation, channel frequency, and
channel spacing. The transmit modulation is applied directly to the loop in the digital domain through the fractional
divider, which results in very precise accuracy and control over the transmit deviation. The frequency resolution in
the 425–525 MHz band is 14.3 Hz with more resolution in the other bands. The nominal reference frequency to the
PLL is 30 MHz, but any XTAL frequency from 25 to 32 MHz may be used. The modem configuration calculator in
WDS will automatically account for the XTAL frequency being used. The PLL utilizes a differential LC VCO with
integrated on-chip inductors. The output of the VCO is followed by a configurable divider, which will divide the
signal down to the desired output frequency band.
5.3.1. Synthesizer Frequency Control
The frequency is set by changing the integer and fractional settings to the synthesizer. The WDS calculator will
automatically provide these settings, but the synthesizer equation is shown below for convenience. The APIs for
setting the frequency are FREQ_CONTROL_INTE, FREQ_CONTROL_FRAC2, FREQ_CONTROL_FRAC1, and
FREQ_CONTROL_FRAC0.

fc_frac 2  freq_xo
RF_channel =  fc_inte + ------------------  -----------------------------  Hz 
 19  8
2

Note: The fc_frac/219 value in the above formula has to be a number between 1 and 2.
5.3.1.1. EZ Frequency Programming
In applications that utilize multiple frequencies or channels, it may not be desirable to write four API registers each
time a frequency change is required. EZ frequency programming is provided so that only a single register write
(channel number) is required to change frequency. A base frequency is first set by first programming the integer
and fractional components of the synthesizer. This base frequency will correspond to channel 0. Next, a channel
step size is programmed into the FREQ_CONTROL_CHANNEL_STEP_SIZE_1 and
FREQ_CONTROL_CHANNEL_STEP_SIZE_0 API registers. The resulting frequency will be:

RF Frequency = Base Frerquency + Channel  Stepsi ze

The second argument of the START_RX or START_TX is CHANNEL, which sets the channel number for EZ
frequency programming. For example, if the channel step size is set to 1 MHz, the base frequency is set to
490 MHz with the INTE and FRAC API registers, and a CHANNEL number of 5 is programmed during the
START_TX command, the resulting frequency will be 495 MHz. If no CHANNEL argument is written as part of the
START_RX/TX command, it will default to the previous value. The initial value of CHANNEL is 0; so, if no
CHANNEL value is written, it will result in the programmed base frequency.

Rev 1.0 25
Si4 438- C
5.3.1.2. Automatic RX Hopping and Hop Table
The transceiver supports an automatic hopping feature that can be fully configured through the API. This is
intended for RX hopping where the device has to hop from channel to channel and look for packets. Once the
device is put into the RX state, it automatically starts hopping through the hop table if the feature is enabled.
The hop table can hold up to 64 entries and is maintained in firmware. Each entry is a channel number; so, the hop
table can hold up to 64 channels. The number of entries in the table is set by RX HOP TABLE_SIZE API.
The specified channels correspond to the EZ frequency programming method for programming the frequency. The
receiver starts at the base channel and hops in sequence from the top of the hop table to the bottom. The table will
wrap around to the base channel once it reaches the end of the table. An entry of 0xFF in the table indicates that
the entry should be skipped. The device will hop to the next non 0xFF entry.
There are three conditions that can be used to determine whether to continue hopping or to stay on a particular
channel. These conditions are:
RSSI threshold
Preamble timeout (invalid preamble pattern)
Sync word timeout (invalid or no sync word detected after preamble)
These conditions can be used individually, or they can be enabled all together by configuring the
RX_HOP_CONTROL API. However, the firmware will make a decision on whether or not to hop based on the first
condition that is met.
The RSSI that is monitored is the current RSSI value. This is compared to the threshold, and, if it is above the
threshold value, it will stay on the channel. If the RSSI is below the threshold, it will continue hopping. There is no
averaging of RSSI done during the automatic hopping from channel to channel. Since the preamble timeout and
the sync word timeout are features that require packet handling, the RSSI threshold is the only condition that can
be used if the user is in “direct” or “RAW” mode where packet handling features are not used.
Note that the RSSI threshold is not an absolute RSSI value; instead, it is a relative value and should be verified on
the bench to find an optimal threshold for the application.
The turnaround time from RX to RX on a different channel using this method is 115 μs. The time spent in receive
mode will be determined by the configuration of the hop conditions. Manual RX hopping will have the fastest
turn-around time but will require more overhead and management by the host MCU.
The following are example steps for using Auto Hop:
1. Set the base frequency (inte + frac) and channel step size.
2. Define the number of entries in the hop table (RX_HOP_TABLE_SIZE).
3. Write the channels to the hop table (RX_HOP_TABLE_ENTRY_n)
4. Configure the hop condition and enable auto hopping- RSSI, preamble, or sync (RX_HOP_CONTROL).
5. Set preamble and sync parameters if enabled.
6. Program the RSSI threshold property in the modem using “MODEM_RSSI_THRESH”.
7. Set the preamble threshold using “PREAMBLE_CONFIG_STD_1”.
8. Program the preamble timeout property using “PREAMBLE_CONFIG_STD_2”.
9. Set the sync detection parameters if enabled.
10. If needed, use “GPIO_PIN_CFG” to configure a GPIO to toggle on hop and hop table wrap.
11. Use the “START_RX” API with channel number set to the first valid entry in the hop table (i.e., the first non
0xFF entry).
12. Device should now be in auto hop mode.
5.3.1.3. Manual RX Hopping
The RX_HOP command provides the fastest method for hopping from RX to RX but it requires more overhead and
management by the host MCU. Using the RX_HOP command, the turn-around time is 75 μs. The timing is faster
with this method than Start_RX or RX hopping because one of the calculations required for the synthesizer
calibrations is offloaded to the host and must be calculated/stored by the host, VCO_CNT0. For information about
using fast manual hopping, contact customer support.

26 Rev 1.0
 Si4 4 38-C
5.4. Transmitter (TX)
The Si4438 contains an integrated +20 dBm transmitter or power amplifier that is capable of transmitting from –20
to +20 dBm. The output power steps are less than 0.25 dB within 6 dB of max power but become larger and more
non-linear close to minimum output power. The Si4438 PA is designed to provide the highest efficiency and lowest
current consumption possible. PA options are single-ended to allow for easy antenna matching and low BOM cost.
Automatic ramp-up and ramp-down is automatically performed to reduce unwanted spectral spreading.
Chip’s TXRAMP pin is disabled by default to save current in cases where on-chip PA will be able to drive the
antenna. In cases where on-chip PA will drive the external PA, and the external PA needs a ramping signal,
TXRAMP is the signal to use. To enable TXRAMP, set the API Property PA_MODE[7] = 1. TXRAMP will start to
ramp up, and ramp down at the SAME time as the internal on-chip PA ramps up/down.
The ramping speed is programmed by TC[3:0] in the PA_RAMP_EX API property, which has the following
characteristics:

Table 14. Ramp Times as a Function of TC[3:0] Value

TC Ramp Time (μs)
0 1.25
1 1.33
2 1.43
3 1.54
4 1.67
5 1.82
6 2.00
7 2.22
8 2.50
9 2.86
10 3.33
11 4.00
12 5.00
13 6.67
14 10.00
15 20.00

The ramping profile is close to a linear ramping profile with smoothed out corner when approaching Vhi and Vlo.
The TXRAMP pin can source up to 1 mA without voltage drooping. The TXRAMP pin’s sinking capability is
equivalent to a 10 k pull-down resistor.
Vhi = 3 V when Vdd > 3.3 V. When Vdd < 3.3 V, the Vhi will be closely following the Vdd, and ramping time will be
smaller also.
Vlo = 0 V when NO current needed to be sunk into TXRAMP pin. If 10uA need to be sunk into the chip, Vlo will be
10 μA x 10k = 100 mV.

Rev 1.0 27
Si4 438- C
Number Command Summary
0x2200 PA_MODE Sets PA type.
0x2201 PA_PWR_LVL Adjust TX power in fine steps.
Adjust TX power in coarse steps and opti-
0x2202 PA_BIAS_CLKDUTY
mizes for different match configurations.
0x2203 PA_TC Changes the ramp up/down time of the PA.

5.4.1. Si4438: +20 dBm PA

The +20 dBm configuration utilizes a class-E matching configuration. Typical performance for output power steps,
voltage, and temperature are shown in Figures 7–9. The output power is changed in 128 steps through
PA_PWR_LVL API. For detailed matching values, BOM, and performance at other frequencies, refer to the PA
Matching application note.

TXPowervs.PA_PWR_LVL
25

20

15

10

0
TXPower(dBm)

Ͳ5

Ͳ10

Ͳ15

Ͳ20

Ͳ25

Ͳ30

Ͳ35

Ͳ40
0 20 40 60 80 100 120
PA_PWR_LVL

Figure 7. +20 dBm TX Power vs. PA_PWR_LVL

TX Power vs. VDD

22
20
TX Power (dBm)

18
16
14
12
10
1.8 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6
Supply Voltage (VDD)

Figure 8. +20 dBm TX Power vs. VDD

28 Rev 1.0
 Si4 4 38-C

TX Power vs Temp
20.5

20

TX Power (dBm)
19.5

19

18.5

18
-40 -30 -20 -10 0 10 20 30 40 50 60 70 80
Temperature (C)

Figure 9. +20 dBm TX Power vs. Temp

5.5. Crystal Oscillator
The Si4438 includes an integrated crystal oscillator with a fast start-up time of less than 250 μs. The design is
differential with the required crystal load capacitance integrated on-chip to minimize the number of external
components. By default, all that is required off-chip is the crystal. The default crystal is 30 MHz, but the circuit is
designed to handle any XTAL from 25 to 32 MHz. If a crystal different than 30 MHz is used, the POWER_UP API
boot command must be modified. The WDS calculator crystal frequency field must also be changed to reflect the
frequency being used. The crystal load capacitance can be digitally programmed to accommodate crystals with
various load capacitance requirements and to adjust the frequency of the crystal oscillator. The tuning of the crystal
load capacitance is programmed through the GLOBAL_XO_TUNE API property. The total internal capacitance is
11 pF and is adjustable in 127 steps (70 fF/step). The crystal frequency adjustment can be used to compensate for
crystal production tolerances. The frequency offset characteristics of the capacitor bank are demonstrated in
Figure 10.

Figure 10. Capacitor Bank Frequency Offset Characteristics

Utilizing the on-chip temperature sensor and suitable control software, the temperature dependency of the crystal

Rev 1.0 29
Si4 438- C
can be canceled.
A TCXO or external signal source can easily be used in place of a conventional XTAL and should be connected to
the XIN pin. The incoming clock signal is recommended to be peak-to-peak swing in the range of 600 mV to 1.4 V
and ac-coupled to the XIN pin. If the peak-to-peak swing of the TCXO exceeds 1.4 V, then dc coupling to the XIN
pin should be used. The maximum allowed swing on XIN is 1.8 V peak-to-peak.
The XO capacitor bank should be set to 0 whenever an external drive is used on the XIN pin. In addition, the
POWER_UP command should be invoked with the TCXO option whenever external drive is used.

30 Rev 1.0
 Si4 4 38-C
6. Data Handling and Packet Handler
6.1. RX and TX FIFOs
Two 64-byte FIFOs are integrated into the chip, one for RX and one for TX, as shown in Figure 11. Writing to
command Register 66h loads data into the TX FIFO, and reading from command Register 77h reads data from the
RX FIFO. The TX FIFO has a threshold for when the FIFO is almost empty, which is set by the “TX_FIFO_EMPTY”
property. An interrupt event occurs when the data in the TX FIFO reaches the almost empty threshold. If more data
is not loaded into the FIFO, the chip automatically exits the TX state after the PACKET_SENT interrupt occurs. The
RX FIFO has one programmable threshold, which is programmed by setting the “RX_FIFO_FULL” property. When
the incoming RX data crosses the Almost Full Threshold, an interrupt will be generated to the microcontroller via
the nIRQ pin. The microcontroller will then need to read the data from the RX FIFO. The RX Almost Full Threshold
indication implies that the host can read at least the threshold number of bytes from the RX FIFO at that time. Both
the TX and RX FIFOs may be cleared or reset with the “FIFO_RESET” command.

Figure 11. TX and RX FIFOs

Rev 1.0 31
Si4 438- C
6.2. Packet Handler
When using the FIFOs, automatic packet handling may be enabled for TX mode, RX mode, or both. The usual
fields for network communication, such as preamble, synchronization word, headers, packet length, and CRC, can
be configured to be automatically added to the data payload. The fields needed for packet generation normally
change infrequently and can therefore be stored in registers. Automatically adding these fields to the data payload
in TX mode and automatically checking them in RX mode greatly reduces the amount of communication between
the microcontroller and Si4438. It also greatly reduces the required computational power of the microcontroller. The
general packet structure is shown in Figure 12. Any or all of the fields can be enabled and checked by the internal
packet handler.

Figure 12. Packet Handler Structure

The fields are highly programmable and can be used to check any kind of pattern in a packet structure. The
general functions of the packet handler include the following:
Detection/validation of Preamble quality in RX mode (PREAMBLE_VALID signal)
Detection of Sync word in RX mode (SYNC_OK signal)
Detection of valid packets in RX mode (PKT_VALID signal)
Detection of CRC errors in RX mode (CRC_ERR signal)
Data de-whitening and/or Manchester decoding (if enabled) in RX mode
Match/Header checking in RX mode
Storage of Data Field bytes into FIFO memory in RX mode
Construction of Preamble field in TX mode
Construction of Sync field in TX mode
Construction of Data Field from FIFO memory in TX mode
Construction of CRC field (if enabled) in TX mode
Data whitening and/or Manchester encoding (if enabled) in TX mode
For details on how to configure the packet handler, see “AN626: Packet Handler Operation for Si4438 RFICs”.

32 Rev 1.0
 Si4 4 38-C
7. RX Modem Configuration
The Si4438 can easily be configured for different data rate, deviation, frequency, etc. by using the WDS settings
calculator, which generates an initialization file for use by the host MCU.

8. Auxiliary Blocks
8.1. Wake-up Timer and 32 kHz Clock Source
The chip contains an integrated wake-up timer that can be used to periodically wake the chip from sleep mode. The
wake-up timer runs from either the internal 32 kHz RC Oscillator, or from an external 32 kHz XTAL.
The wake-up timer can be configured to run when in sleep mode. If WUT_EN = 1 in the GLOBAL_WUT_CONFIG
property, prior to entering sleep mode, the wake-up timer will count for a time specified defined by the
GLOBAL_WUT_R and GLOBAL_WUT_M properties. At the expiration of this period, an interrupt will be generated
on the nIRQ pin if this interrupt is enabled in the INT_CTL_CHIP_ENABLE property. The microcontroller will then
need to verify the interrupt by reading the chip interrupt status either via GET_INT_STATUS or a fast response
register. The formula for calculating the Wake-Up Period is as follows:

WUT_R
42
WUT = WUT_M  -----------------------------  ms 
32.768
The RC oscillator frequency will change with temperature; so, a periodic recalibration is required. The RC oscillator
is automatically calibrated during the POWER_UP command and exits from the Shutdown state. To enable the
recalibration feature, CAL_EN must be set in the GLOBAL_WUT_CONFIG property, and the desired calibration
period should be selected via WUT_CAL_PERIOD[2:0] in the same API property. During the calibration, the
32 kHz RC oscillator frequency is compared to the 30 MHz XTAL and then adjusted accordingly. The calibration
needs to start the 30 MHz XTAL, which increases the average current consumption; so, a longer CAL_PERIOD
results in a lower average current consumption. The 32 kHz XTAL accuracy is comprised of both the XTAL
parameters and the internal circuit. The XTAL accuracy can be defined as the XTAL initial error + XTAL aging +
XTAL temperature drift + detuning from the internal oscillator circuit. The error caused by the internal circuit is
typically less than 10 ppm. Refer to API documentation for details on WUT related commands and properties.
8.2. Low Duty Cycle Mode (Auto RX Wake-Up)
The low duty cycle (LDC) mode is implemented to automatically wake-up the receiver to check if a valid signal is
available or to enable the transmitter to send a packet. It allows low average current polling operation by the
Si4438 for which the wake-up timer (WUT) is used. RX and TX LDC operation must be set via the
GLOBAL_WUT_CONFIG property when setting up the WUT. The LDC wake-up period is determined by the
following formula:

WUT_R
42
LDC = WUT_LDC  -----------------------------  ms 
32.768

where the WUT_LDC parameter can be set by the GLOBAL_WUT_LDC property. The WUT period must be set in
conjunction with the LDC mode duration; for the relevant API properties, see the wake-up timer (WUT) section.

Rev 1.0 33
Si4 438- C

Figure 13. RX and TX LDC Sequences

The basic operation of RX LDC mode is shown in Figure 14. The receiver periodically wakes itself up to work on
RX_STATE during LDC mode duration. If a valid preamble is not detected, a receive error is detected, or an entire
packet is not received, the receiver returns to the WUT state (i.e., ready or sleep) at the end of LDC mode duration
and remains in that mode until the beginning of the next wake-up period. If a valid preamble or sync word is
detected, the receiver delays the LDC mode duration to receive the entire packet. If a packet is not received during
two LDC mode durations, the receiver returns to the WUT state at the last LDC mode duration until the beginning
of the next wake-up period.

Figure 14. Low Duty Cycle Mode for RX

In TX LDC mode, the transmitter periodically wakes itself up to transmit a packet that is in the data buffer. If a
packet has been transmitted, nIRQ goes low if the option is set in the INT_CTL_ENABLE property. After
transmitting, the transmitter immediately returns to the WUT state and stays there until the next wake-up time
expires.
8.3. Temperature, Battery Voltage, and Auxiliary ADC
The Si4438 family contains an integrated auxiliary ADC for measuring internal battery voltage, an internal
temperature sensor, or an external component over a GPIO. The ADC utilizes a SAR architecture and achieves
11-bit resolution. The Effective Number of Bits (ENOB) is 9 bits. When measuring external components, the input
voltage range is 1 V, and the conversion rate is between 300 Hz to 2.44 kHz. The ADC value is read by first
sending the GET_ADC_READING command and enabling the inputs that are desired to be read: GPIO, battery, or
temp. The temperature sensor accuracy at 25 °C is typically ±2 °C. For API details, refer to the EZRadioPRO API
Documentation.zip file available on www.silabs.com.

34 Rev 1.0
 Si4 4 38-C
8.4. Low Battery Detector
The low battery detector (LBD) is enabled and utilized as part of the wake-up-timer (WUT). The LBD function is not
available unless the WUT is enabled, but the host MCU can manually check the battery voltage anytime with the
auxiliary ADC. The LBD function is enabled in the GLOBAL_WUT_CONFIG API property. The battery voltage will
be compared against the threshold each time the WUT expires. The threshold for the LBD function is set in
GLOBAL_LOW_BATT_THRESH. The threshold steps are in increments of 50 mV, ranging from a minimum of
1.5 V up to 3.05 V. The accuracy of the LBD is ±3%. The LBD notification can be configured as an interrupt on the
nIRQ pin or enabled as a direct function on one of the GPIOs.
8.5. Antenna Diversity
To mitigate the problem of frequency-selective fading due to multipath propagation, some transceiver systems use
a scheme known as antenna diversity. In this scheme, two antennas are used. Each time the transceiver enters RX
mode the receive signal strength from each antenna is evaluated. This evaluation process takes place during the
preamble portion of the packet. The antenna with the strongest received signal is then used for the remainder of
that RX packet. The same antenna will also be used for the next corresponding TX packet. This chip fully supports
antenna diversity with an integrated antenna diversity control algorithm. The required signals needed to control an
external SPDT RF switch (such as a PIN diode or GaAs switch) are available on the GPIOx pins. The operation of
these GPIO signals is programmable to allow for different antenna diversity architectures and configurations. The
antdiv[2:0] bits are found in the MODEM_ANT_DIV_CONTROL API property descriptions and enable the antenna
diversity mode. The GPIO pins are capable of sourcing up to 5 mA of current; so, it may be used directly to
forward-bias a PIN diode if desired. The antenna diversity algorithm will automatically toggle back and forth
between the antennas until the packet starts to arrive. The recommended preamble length for optimal antenna
selection is 8 bytes.
8.6. Preamble Sense Mode
This mode of operation is suitable for extremely low power applications where power consumption is important.
The preamble sense mode (PSM) takes advantage of the Digital Signal Arrival detector (DSA) which can detect a
preamble within 8 bit times with no sensitivity degradation. This fast detection of an incoming signal can be
combined with duty cycling of the receiver during the time the device is searching or sniffing for packets over the
air. The average receive current is lowered significantly when using this mode. In applications where the timing of
the incoming signal is unknown, the amount of power savings is primarily dependent on the data rate and preamble
length as the RX inactive time is determined by these factors. In applications where the sleep time is fixed and the
timing of the incoming signal is known, the average current also depends on the sleep time. The PSM mode is
similar to the low duty cycle mode but has the benefit of faster signal detection and autonomous duty cycling of the
receiver to achieve even lower average receive currents. This mode can be used with the low power mode (LP)
which has an active RX current of 10 mA or with the high-performance (HP) mode which has an active RX current
of 13 mA.

Figure 15. Preamble Sense Mode

Rev 1.0 35
Si4 438- C

Table 15. Data Rates*

Data Rate
1.2 kbps 9.6 kbps 50 kbps 100 kbps
PM length = 4 bytes 6.48 6.84 8.44 10.43 mA
PM length = 8 bytes 3.83 3.96 4.57 5.33 mA
*Note: Typical values. Active RX current is 13 mA.

36 Rev 1.0
 Si4 4 38-C
9. Pin Descriptions: Si4438-C

Pin Pin Name I/0 Description

Shutdown Input Pin.
0–VDD V digital input. SDN should be = 0 in all modes except Shutdown mode.
1 SDN I
When SDN = 1, the chip will be completely shut down, and the contents of the
registers will be lost.
2 RXp I Differential RF Input Pins of the LNA.
3 RXn I See application schematic for example matching network.

Transmit Output Pin.

4 TX O The PA output is an open-drain connection, so the L-C match must supply
VDD (+3.3 VDC nominal) to this pin.
It is recommended to connect this pin to GND per the reference design
5 NC schematic. Not connected internally to any circuitry.
+1.8 to +3.8 V Supply Voltage Input to Internal Regulators.
6 VDD VDD
The recommended VDD supply voltage is +3.3 V.

Programmable Bias Output with Ramp Capability for External FET PA.
7 TXRAMP O
See "5.4. Transmitter (TX)" on page 27.

+1.8 to +3.8 V Supply Voltage Input to Internal Regulators.

8 VDD VDD
The recommended VDD supply voltage is +3.3 V.

9 GPIO0 I/O General Purpose Digital I/O.

May be configured through the registers to perform various functions including:
10 GPIO1 I/O Microcontroller Clock Output, FIFO status, POR, Wake-Up timer, Low Battery
Detect, TRSW, AntDiversity control, etc.

Rev 1.0 37
Si4 438- C
Pin Pin Name I/0 Description

General Microcontroller Interrupt Status Output.

When the Si4438 exhibits any one of the interrupt events, the nIRQ pin will be
11 nIRQ O set low = 0. The Microcontroller can then determine the state of the interrupt
by reading the interrupt status. No external resistor pull-up is required, but it
may be desirable if multiple interrupt lines are connected.

Serial Clock Input.

12 SCLK I 0–VDD V digital input. This pin provides the serial data clock function for the
4-line serial data bus. Data is clocked into the Si4438 on positive edge transi-
tions.

0–VDD V Digital Output.

13 SDO O
Provides a serial readback function of the internal control registers.

Serial Data Input.

14 SDI I 0–VDD V digital input. This pin provides the serial data stream for the 4-line
serial data bus.

Serial Interface Select Input.

15 nSEL I 0–VDD V digital input. This pin provides the Select/Enable function for the
4-line serial data bus.

Crystal Oscillator Output.

16 XOUT O Connect to an external 25 to 32 MHz crystal, or leave floating when driving
with an external source on XIN.

Crystal Oscillator Input.

17 XIN I
Connect to an external 25 to 32 MHz crystal, or connect to an external source.
When using a XTAL, leave floating per the reference design schematic. When
18 GND GND using a TCXO, connect to TCXO GND which should be separate from the
board reference ground plane.
19 GPIO2 I/O General Purpose Digital I/O.
May be configured through the registers to perform various functions, including
20 GPIO3 I/O Microcontroller Clock Output, FIFO status, POR, Wake-Up timer, Low Battery
Detect, TRSW, AntDiversity control, etc.
The exposed metal paddle on the bottom of the Si4438 supplies the RF and
circuit ground(s) for the entire chip. It is very important that a good solder con-
PKG PADDLE_GND GND nection is made between this exposed metal paddle and the ground plane of
the PCB underlying the Si4438.

38 Rev 1.0
 Si4 4 38-C
10. Ordering Information

Part Number* Description Package Type Operating

Temperature
Si4438-C2A-GM ISM EZRadioPRO Transceiver QFN-20 –40 to 85 °C
Pb-free
*Note: Add an “(R)” at the end of the device part number to denote tape and reel option.

Rev 1.0 39
Si4 438- C
11. Package Outline: Si4438
Figure 16 illustrates the package details for the Si4438. Table 16 lists the values for the dimensions shown in the
illustration.

Figure 16. 20-Pin Quad Flat No-Lead (QFN)

Table 16. Package Dimensions

Dimension Min Nom Max
A 0.80 0.85 0.90
A1 0.00 0.02 0.05
A3 0.20 REF
b 0.18 0.25 0.30
D 4.00 BSC
D2 2.45 2.60 2.75
e 0.50 BSC
E 4.00 BSC
E2 2.45 2.60 2.75
L 0.30 0.40 0.50
aaa 0.15
bbb 0.15
ccc 0.10
ddd 0.10
eee 0.08
Notes:
1. All dimensions are shown in millimeters (mm) unless otherwise noted.
2. Dimensioning and tolerancing per ANSI Y14.5M-1994.
3. This drawing conforms to the JEDEC Solid State Outline MO-220, Variation
VGGD-8.
4. Recommended card reflow profile is per the JEDEC/IPC J-STD-020
specification for Small Body Components.

40 Rev 1.0
 Si4 4 38-C
12. PCB Land Pattern: Si4438
Figure 17 illustrates the PCB land pattern details for the Si4438. Table 17 lists the values for the dimensions shown
in the illustration.

Figure 17. PCB Land Pattern

Rev 1.0 41
Si4 438- C

Table 17. PCB Land Pattern Dimensions

Symbol Millimeters
Min Max
C1 3.90 4.00
C2 3.90 4.00
E 0.50 REF
X1 0.20 0.30
X2 2.55 2.65
Y1 0.65 0.75
Y2 2.55 2.65
Notes:
General
1. All dimensions shown are in millimeters (mm) unless otherwise noted.
2. This land pattern design is based on IPC-7351 guidelines.
Solder Mask Design
3. All metal pads are to be non-solder mask defined (NSMD). Clearance
between the solder mask and the metal pad is to be 60 μm minimum, all
the way around the pad.
Stencil Design
4. A stainless steel, laser-cut and electro-polished stencil with trapezoidal
walls should be used to assure good solder paste release.
5. The stencil thickness should be 0.125 mm (5 mils).
6. The ratio of stencil aperture to land pad size should be 1:1 for the
perimeter pads.
7. A 2x2 array of 1.10 x 1.10 mm openings on 1.30 mm pitch should be
used for the center ground pad.
Card Assembly
8. A No-Clean, Type-3 solder paste is recommended.
9. The recommended card reflow profile is per the JEDEC/IPC J-STD-020
specification for small body components.

42 Rev 1.0
 Si4 4 38-C
13. Top Marking
13.1. Si4438 Top Marking

13.2. Top Marking Explanation

Mark Method YAG Laser

Line 1 Marking Part Number 44382A = Si4438 Rev 2A1

Line 2 Marking TTTTTT = Internal Code Internal tracking code.2

YY = Year Assigned by the Assembly House. Corresponds to the last
Line 3 Marking significant digit of the year and workweek of the mold date.
WW = Workweek
Notes:
1. The first letter after the part number is part of the ROM revision. The last letter indicates the firmware revision.
2. The first letter of this line is part of the ROM revision.

Rev 1.0 43
 Simplicity Studio
One-click access to MCU tools,
documentation, software, source
code libraries & more. Available
for Windows, Mac and Linux!

www.silabs.com/simplicity

MCU Portfolio SW/HW Quality Support and Community

www.silabs.com/mcu www.silabs.com/simplicity www.silabs.com/quality community.silabs.com

Disclaimer
Silicon Laboratories intends to provide customers with the latest, accurate, and in-depth documentation of all peripherals and modules available for system and software implementers
using or intending to use the Silicon Laboratories products. Characterization data, available modules and peripherals, memory sizes and memory addresses refer to each specific
device, and "Typical" parameters provided can and do vary in different applications. Application examples described herein are for illustrative purposes only. Silicon Laboratories
reserves the right to make changes without further notice and limitation to product information, specifications, and descriptions herein, and does not give warranties as to the accuracy
or completeness of the included information. Silicon Laboratories shall have no liability for the consequences of use of the information supplied herein. This document does not imply
or express copyright licenses granted hereunder to design or fabricate any integrated circuits. The products must not be used within any Life Support System without the specific
written consent of Silicon Laboratories. A "Life Support System" is any product or system intended to support or sustain life and/or health, which, if it fails, can be reasonably expected
to result in significant personal injury or death. Silicon Laboratories products are generally not intended for military applications. Silicon Laboratories products shall under no
circumstances be used in weapons of mass destruction including (but not limited to) nuclear, biological or chemical weapons, or missiles capable of delivering such weapons.

Trademark Information
Silicon Laboratories Inc., Silicon Laboratories, Silicon Labs, SiLabs and the Silicon Labs logo, CMEMS®, EFM, EFM32, EFR, Energy Micro, Energy Micro logo and combinations
thereof, "the world’s most energy friendly microcontrollers", Ember®, EZLink®, EZMac®, EZRadio®, EZRadioPRO®, DSPLL®, ISOmodem ®, Precision32®, ProSLIC®, SiPHY®,
USBXpress® and others are trademarks or registered trademarks of Silicon Laboratories Inc. ARM, CORTEX, Cortex-M3 and THUMB are trademarks or registered trademarks of
ARM Holdings. Keil is a registered trademark of ARM Limited. All other products or brand names mentioned herein are trademarks of their respective holders.

Silicon Laboratories Inc.

400 West Cesar Chavez
Austin, TX 78701
USA

http://www.silabs.com