If we use an external oscillator such as a TCXO, MEMS oscillator or usual oscillator with Si4x3x devices, then which parameters need attention?
AN417 refers to the use of crystals but there is no description when using oscillators.In this case focus should be placed on the voltage amplitude value on the XOUT pin.
TCXOs usually works in the range of +/- 1...5 PPM range.
'The crystal oscillator should be enabled, and CINT (xtal capbank) should be programmed to its minimum value. XIN (not shown) should be left open.'
What is the Si4432 PA output impedance?
The PA circuitry in Si4432 is of a type known as a switching power amplifier which means the small signal parameters like impedance, S-parameter can't be defined. The most important design PA parameters are the on state resistance and the output capacitance. The most critical parameter are the peak voltage on the output transistor’s drain which is primarily determined by the applied external matching. The EZRadioPRO family of chips use class E type matching for the low (+13dBm) and middle (+17 dBm) power ranges, and a matching approach which produces a square wave voltage waveform for the highest (+20dBm) power applications. Detailed description, and matching design hints, and examples are given in AN436 and AN435 application notes.
Can EZRadioPRO chips be matched to operate over a wide bandwidth or range of frequencies?
Silicon Labs has received several inquiries regarding whether the EZRadioPRO chips can be matched for wideband operation. That is to say, can the chip use one single fixed-frequency impedance match, but then operate at frequencies other than the center frequency of the match?
The RX/LNA circuitry can easily operate over a considerable bandwidth with a fixed (single frequency) impedance match. Silicon Labs has observed less than 1 dB degradation in RX sensitivity while varying over a +/- 50 MHz tuned bandwidth, relative to the center frequency of the RX impedance match.
The TX circuitry is somewhat more narrow-band in nature, and was not designed or intended for extremely wideband operation, without adjustment or tuning of the TX matching network. However, it is possible to extend the bandwidth of operation with a fixed-frequency impedance matching network, with a reduction in targeted output power level.
Please contact Technical Support for further details regarding wideband operation.
Do I need to use an RF switch between the TX and RX paths on EZRadioPRO Si4x3x devices?
The answer depends upon the output power level that the User is targeting.
For lower power applications (+13 dBm output power, using Si4031/4431 chips), Silicon Labs has developed a matching network and topology that directly ties the TX and RX paths together, without the need for an RF switch to provide isolation between the two paths. We refer to this board configuration as the Direct Tie configuration, and reference designs based upon this matching topology may be downloaded from our website. In this matching approach, component values are carefully selected so that the OFF path does not load down the ON path. That is to say, the circuit must be designed so that the RX input port does not excessively load the TX path (and thus reduce the TX output power) during RX mode, and the TX output circuitry does not degrade the noise figure of the RX path during RX mode. The Direct Tie matching procedure is discussed in detail in AN436 Si4430-4431 PA Matching App Note.
It is not possible to direct-tie these two paths together without some small amount of performance degradation (relative to the performance obtainable with separate TX and RX paths). However, this degradation is minimal (less than 1 dB reduction in TX output power, and less than 2 dB reduction in RX sensitivity) and is often considered an acceptable trade-off to make, given the reduction in cost due to the elimination of the RF switch.
Silicon Labs also has developed several Direct Tie reference designs using the Si4032/4432 chips, targeted at medium output power levels (+17 dBm). These are also available for download from our website.
For high power designs (+20 dBm) using the Si4032/4432 chips, Silicon Labs recommends use of an RF switch between the TX and RX paths. (For comparison, Si4x6x devices do not necessarily need RF switch even for +20 dBm TX power. For these details please refer to AN648: https://www.silabs.com/documents/public/application-notes/AN648.pdf)
Please note that RF switches may not behave in a perfectly linear fashion, and may generate (or re-generate) some amount of harmonics as the TX signal passes through it. Thus in designs that use an RF switch, it may be necessary to place a portion of the harmonic low-pass filtering AFTER the switch (between the switch and the antenna).
For more information, please refer to: AN436: https://www.silabs.com/documents/public/application-notes/AN436.pdf
How can I reduce the harmonics at the TX output for EZRadioPRO devices?
Control of harmonics is not a trivial issue. Depending upon the regulatory standard (e.g., FCC, ETSI, ARIB etc.) the requirements for attenuation of harmonics may be quite severe.
The power amplifier circuitry inside the EZRadioPRO family of chips is of a type known as a 'switching amplifier', and inherently may have a higher harmonic content than a more conventional class of amplifier (such as Class-A or Class-B). Care must be taken to attenuate the harmonic signals present at the TX output pin, prior to reaching the antenna.
Depending upon the applicable regulatory standard, a low-pass filter of between 3rd order and 5th order may be necessary to obtain the required amount of harmonic attenuation.
However, it should be noted that simply designing a high-order low-pass filter may not be sufficient. It is absolutely necessary to also use good PCB layout practices in order to obtain the desired filter response. These design techniques include:
1) Minimum length traces and pads, in order to minimize parasitic inductance and capacitance.
2) Minimum length traces also help to control harmonics that may radiate directly from the traces and/or components themselves.
3) Liberal use of vias to obtain good connection to GND.
4) Physical placement of filter components in such a fashion to prevent unwanted coupling of the signal from input to output.
5) Use of high-quality discrete components to obtain minimum insertion loss.
6) Use of a (at least) 4-layer PCB (if possible), with at least one internal solid GND plane layer.
7) Use of a highly-linear RF switch (if applicable to the design) that does not additionally generate harmonics within the switch itself.
8) If an RF switch is used, consider placing a portion of the required low-pass filtering AFTER the RF switch, to help attenuate harmonics created within the switch.
9) Use of a shield can, if necessary.
These techniques are discussed further within our application notes on P.A. Matching and Board Layout Techniques.
For more information, please refer to:
What are the FCC requirements for harmonic levels?
The FCC requirements on harmonic levels (spurious emissions) depends upon the section or 'Part' of the FCC regulatory standard under which the device is operating.
Silicon Labs has observed that the most common unlicensed applications for the EZRadioPRO family of chips are governed by one of the following two FCC Parts:
FCC Part 15.247
FCC Part 15.231
Part 15.247 concerns itself with higher-power frequency hopping spread spectrum (FHSS) systems and wideband digital modulation systems, within the 902-928 MHz frequency range.
Part 15.231 concerns itself with low-power devices intended for periodic operation (e.g., garage door openers, remote keyless entry systems, security and control links, etc.). The most commonly-used frequencies for devices operating under this Part are generally within the 315 MHz to 434 MHz frequency range. The general requirements for harmonic levels under Part 15.231 are that they be no higher than -20 dBc, UNLESS the harmonic will within a 'Restricted Band' (see discussion below).
The requirements of Part 15.247 call for harmonic levels no higher than -20 dBc, UNLESS the harmonic falls within a 'Restricted Band' as defined in FCC Part 15.205. If a harmonic signal falls within a Restricted Band, then a more stringent harmonic level is specified. The maximum allowed levels of harmonics are defined in FCC Part 15.209, and are specified in terms of radiated field strength (measured at a distance of 3 meters). The maximum allowed level depends upon whether the harmonic falls above 960 MHz or below 960 MHz. If the harmonic falls below 960 MHz, the allowed field strength is 200 microvolts/meter. If the harmonic falls above 960 MHz, the allowed field strength is 500 microvolts/meter.
Users often desire to measure harmonic levels as a conducted measurement, rather than as a radiated field strength measurement. Proper measurement of radiated field strength requires a calibrated antenna chamber, and often may not be immediately available to a user. In order to convert 'conducted TX output power' into 'radiated field strength at a specified distance' (or vice versa), it is necessary to know the characteristics (i.e., spatial gain) of the antenna for the device-under-test. This will obviously vary from one user to another, and thus Silicon Labs cannot provide one unique answer that will be appropriate for all customers. However, if an ideal isotropic point-source radiator is assumed for the antenna, the following mathematical conversions may be made:
200 microvolts/meter (at 3 meters), equivalent to -49.2 dBm conducted input power to an isotropic point-source radiator
500 microvolts/meter (at 3 meters), equivalent to -41.2 dBm conducted input power to an isotropic point-source radiator
Again, we must emphasize that compliance with FCC spurious emission requirements is performed using radiated measurements, and not conducted measurements.
The provisions of FCC Part 15.35 may (in certain instances) allow use of time-averaging of a periodic or packet-like burst signal in order to reduce the apparent level of harmonics. This is often useful in achieving compliance with FCC requirements.
For more information regarding FCC requirements, please download the latest FCC documents from the government FCC website.
It will be necessary to browse or search through 'CFR 47 Telecommunications' for 'Part 15' documents.
Silicon Labs suggests that the user become familiar with the following documents:
CFR 47 Part 15.231
CFR 47 Part 15.247
CFR 47 Part 15.205
CFR 47 Part 15.209
CFR 47 Part 15.35
I noticed there are a few different types of testcards - What are their purposes?
There are four test card design types currently supporting EZRadioPRO, the designs are available under the Technical Resource search at www.silabs.com.
A) Antenna Diversity:
This design implements antenna diversity.
Dual 50 Ω SMA connectors for antenna diversity designs
Ideal for indoor fading rich environments
These are often referred to as the 'A' type testcards
B) TX/RX split:
This design has two SMA connectors for use with coax cable rather than antennas in a controlled lab environment (such as spectrum analysis on the Tx path or BER/PER analysis on the Rx path).
They are not intended for use with antennas, the separated channels allow for independent testing of the Rx and Tx.
These are often referred to as the 'B' type testcards
C) TX/RX Antenna Switch:
Implements an inexpensive RF switch between the PA and LNA of the radio.
Single 50 Ω antenna designs
Ideal for range testing
These are often referred to as the 'C' type testcards
D) TX/RX Direct Tie:
This design directly ties the LNA and PA together providing the lowest cost solution.
Single 50 Ω antenna designs
Ideal for range testing
These are often referred to as the 'D' type testcards.