Overview of CustomRF's Capabilities

Monopole Antenna Near Obstruction

This example, a 3D EMF simulation in CST Microwave Studio, shows how the impedance, the near-field and the far-field gain pattern of a monopole WiFi antenna are affected by a nearby obstruction.
Antenna Unobstructed
Antenna Near Obstruction
Unobstructed Monopole Antenna Obstructed Monopole Antenna
Unobstructed Return Loss

Obstructed Return Loss

E-Field Magnitude in YZ-Plane

Unobstructed Antenna E-Field

E-Field Magnitude in YZ-Plane

Obstructed Antenna E-Field

Far-Field Antenna Pattern

Unobstructed Antenna Pattern

Far-Field Antenna Pattern

Obstructed Antenna Pattern

Waveguide Launch

WR42 waveguide is recommended to be used in the 17.6 - 26.7 GHz frequency range, but, with a cut-off frequency of 14.058 GHz, it could be used at lower frequencies, albeit with somewhat degraded performance such as higher dispersion. In this case WR42 with a 17 - 22 GHz operating band is desired. The launch is from a .141 semi-rigid coaxial cable. CustomRF optimized a launch with a single probe as shown below.

Coaxial to WR42 Waveguide Transition, Single Probe
WR42 Launch, NoStubs

Return Loss, Coaxial to WR42 Waveguide Transition, Single Probe
Return Loss, No Stubs

This return loss is too poor for the application. CustomRF reoptimized the structure with two #0-80 tuning screws incorporated and thus significantly improved the return loss. The improved structure and return loss are shown below.

Coaxial to WR42 Waveguide Transition, With Tuning Stubs
WR42 Launch with 2 Stubs

Return Loss, Coaxial to WR42 Waveguide Transition, With Tuning Stubs
Return Loss with 2 Stubs

Critical Path Isolation on PCB

This example, a 3D EMF simulation in CST Microwave Studio, shows the isolation between 2 critical points on a 2-sided PCB. Port 1 is connected to the power amplifier, Port 3 to the low noise receiver of a transceiver, so that S31 is the transmit to receive (TX/RX) isolation. Also shown is how the isolation can be significantly improved by surrounding each via between top and bottom side with a "via fence" that connect the top and bottom-side ground planes. These via fences inhibit signals that otherwise propagate between the ground planes much like inside waveguide.

PCB Without Via Fence
PCB Without Via Fence
PCB with Via Fence
PCB With Via Fence
TX/RX Isolation
Isolation Without Via Fence

TX/RX Isolation
Isolation With Via Fence

RFID Tag Near Water

Radio Frequency Identification (RFID) is being used for many purposes, including to keep track of patients in hospitals. In the United States passive RFID tags in the UHF band operate in the band around 915 MHz. At this frequency, a small tag can easily be detuned in the presence of certain matter, including water. The human body consists mostly of water, so performance of passive tags is a concern. In this example we show simulation results of a typical flat RFID tag in both free space and when it is placed on top of water.

Antenna Structure In Free Space
RFID Tag in Free Space

Antenna Structure On Water
RFID Tag on Water

Return Loss, Free Space
Return Loss, Free Space

Return Loss, On Water
Return Loss, On Water
Antenna Pattern, Free Space
Pattern, Free Space

Antenna Pattern, On Water
Pattern On Water

It can be seen that the proximity of water has a severe impact on return loss and on antenna gain.

Reconstruction Bandpass Filter with sin(x)/x Compensation

The output spectrum of a digital-to-analog converter (DAC) inherently contains not only the fundamental band but alias bands as well. For this reason a reconstruction filter is used to capture the desired band. This is often the fundamental band, but in some cases an alias band is preferred (digital upconversion). Then the reconstruction filter must be a bandpass filter. This is shown below as implemented to generate a 60 MHz IF for a radio transmitter.

Block Diagram

The output spectrum is not inherently flat as desired but has a roll-off, the voltage magnitude of which follows the shape of a sin(x)/x, or sinc(x), function, wherein the nulls are at multiples of the 80 MHz sample frequency. This is shown below.

Unfiltered DAC Output Spectrum
Note: Desired alias is spectrally inverted. This is pre-compensated in the baseband.

This roll-off could be pre-equalized in the digital baseband section before the DAC, but that would require additional DAC resolution, thus detracting from its dynamic range. To avoid this it was decided to use analog compensation.

The reconstruction filter had been slated to be a 3-pole Chebychev filter, with a topology as shown below.

Filter Schematic

This topology was maintained, but the element values changed so that, in the passband, the response was no longer quasi-flat, but the inverse of a sinc function, while still providing 24 dB of rejection at the edge of the closest alias band. The filter response (ideal element) is shown below.

Filter Response

The spectrum at the upconverter input then appears as shown below.

Filtered DAC Output Spectrum

While the resulting IF spectrum is not perfectly flat, this filter nevertheless saves 9 dB in DAC dynamic range.

Effect of Gain Compression on Spectrum and EVM

An OFDM signal according to IEEE 802.11a, 54 Mbps, has a peak to average power ratio of approximately 22 dB. For this reason it is very susceptible to gain compression in any active stage, such as an amplifier. The level of the signal passing through such needs to be well below the 1dB compression point or the modulated carrier will experience spectral regrowth and so may not meet the specified spectral mask, and the receiver will get bit errors. This example shows simulation results, done in Mathcad, of the effect of an amplifier operating 3 dB and 6 dB below its 1 dB output compression point. The block diagram is shown below.

OFDM Compression Simulation Block Diagram

Amplifier with OFDM

An ideal OFDM generator produces 256 OFDM symbols, each consisting of 48 64-QAM subcarriers + 4 pilots as specified in 802.11a. This signal is used to modulate a 60 MHz IF, which is then applied to a unity-gain amplifier with gain compression per the Rapp model. The input spectrum, with no gain compression, and the output spectrum, when operating 3 dB and 6 dB below the 1 dB compression point, are shown below.

Amplifier Spectrum
It can be seen that spectral regrowth increases noticeably as the level gets closer to the 1 dB compression point. At the same time there is an increase in the constellation error, or error vector magnitude, as shown below.

Constellation Without Compression
Uncompressed Constellation

Constellation With Compression
Compressed Constellation
The simulated constellation error is -21.6 dB with a 3 dB and -25.6 dB with a 6 dB backoff from the 1 dB compression point, so with this amplifier 6 dB backoff is required to meet the -25 dB constellation error required per IEEE 802.11. In practice, with most amplifiers the backoff must be 8 dB or greater because there are other contributors to error (thermal noise, phase noise from the synthesizer, clock synchronization errors, etc.).

CustomRF, Inc. |  28 Lowell Drive  |  Stow  |  MA  |   01775  |  978.637.2070  |  info@customrf.com