Guest columnist Jamie Lunn, vector network analyser product manager at Rohde & Schwarz UK, looks at some of the measurement challenges designers face at RF frequencies above 30GHz
Extremely high frequency, which can be defined as the frequencies between 30 and 300GHz, remained relatively undeveloped for many years, except for somewhat esoteric applications such as radio astronomy and remote atmospheric sensing.
That is because this frequency range is highly susceptible to atmospheric attenuation – when resonance of the oxygen molecule attenuates the signal, or when rain absorbs the signal, reducing signal strength.
But with spectrum scarcity in the sub-10GHz band, applications for millimetre-wave systems are found in consumer and non-consumer communications, security, imaging and radar. In particular, with increasing demand for multi-gigabit-per-second communication created by the likes of high-definition (HD) and 3D video, it was soon recognised that the 7GHz of unlicensed bandwidth available around 60GHz was a big opportunity.
So, for several years, the extremely high frequency region has become an area of intense development, with several applications actually being able to capitalise on its signal attenuation characteristics.
Wireless HD
Among the many short-range communications standards that have appeared in the 60GHz frequency range is WirelessHD, a media standard for video in the home that effectively untethers the TV from other home video and audio devices.
Conceived as an in-room, point-to-point, non-line-of-sight standard that uses the band between 57GHz and 64GHz, it is capable of transmitting HD video images and the first products have a limited range of up to10 metres, pleasing copyright owners.
The increased signal attenuation at high frequencies is due to the very small wavelengths generated.
So WirelessHD antennas use various beamforming techniques to concentrate power in the direction of the receiver and bounce signals off nearby objects, using both direct and reflected signals to achieve a stronger, more stable signal.
As you move into higher frequencies and the wavelength size decreases, so the physical structure of devices also comes down in size. This can increase cost dramatically because of the machine costs related to making smaller components. The tolerances involved become much greater too.
Instrumentation can also prove more expensive at higher frequencies, because the frequencies and modulation bandwidth are outside the range of most standard test equipment. One reason for the cost increase is economies of scale.
SiGe and CMOS transistors that can run fast enough for circuits in the millimetre-wave region are a relatively recent arrival, so the growth in millimetre-wave applications has created a correspondingly recent demand for millimetre-wave instrumentation for device specification verification.
Costs can be trimmed by using a down converter module to convert the 60GHz signal to lower frequencies that can be analysed on standard equipment. But there is a risk when using a downconverter or harmonic mixer that the mixing produces multiple image frequencies unsuitable for spurious measurements.
Waveguide flange
Users of harmonic mixers would also need to take account of the waveguide flange, requiring a further conversion to 1.85mm coax to interface with the DUT.
Any additional hardware will also add measurement uncertainty into the signal observed. So it is advisable to use millimetre-wave test equipment that allows measurements to be performed directly on the signal.
When calculating uncertainty in RF measurements stemming from voltage-standing-wave-ratio mismatch, it is important to recognise that uncertainties increase as the frequency rises. So although many spectrum analysers have good power measurement capabilities, a dedicated RF power meter provides better accuracy and impedance matching – critical at higher frequencies.
Precise calibration
It has been said that a vector network analyser (VNA) is only as useful as its measurement accuracy. So to get the best out of a high-frequency VNA, precise calibration is vital. Sources of error come from less than perfect input impedance at the RF receiver and/or output impedance by the RF source, as well as wear and tear to cables and connectors.
Things are further complicated when measurements cannot be made easily at the coaxial connectors, for instance on wafer devices. Calibration won’t fix every error, but you can minimise them by choosing a VNA with good raw system performance and unique calibration techniques.
www.rohde-schwarz.co.uk