You test a 60 GHz radar board without a chamber by splitting the measurement in two: whatever can be reached before the antenna is measured conducted through a 1.85 mm or WR-15 interface, and whatever depends on radiation is measured in a small absorber box at the distance the physics actually asks for. At 60 GHz the free space wavelength is 5 mm, so a 20 mm patch array is in its far field at about 16 cm, and a tabletop enclosure lined with absorber can hold a valid far field measurement that a chamber would only make quieter.
A chamber buys isolation and a low noise floor. It does not buy accuracy. Deciding which of those a given test needs is what keeps a 60 GHz programme off the chamber booking list.
Why the physics moves onto the bench at 60 GHz
Far field distance is 2D squared over lambda. For a 20 mm module aperture at 5 mm wavelength that is 16 cm. For a 100 mm array it is 4 m, which is chamber territory. The aperture, not the frequency, is what decides where the measurement has to happen, and a single chip radar module is a small aperture.
Absorber behaves the same way. A 50 mm pyramid is ten wavelengths thick at 60 GHz, so a box 300 mm on a side gives roughly sixty wavelengths of path with usable absorption on every wall. The same box at 2.4 GHz would be a fraction of a wavelength deep and would do nothing. High frequency is what makes a small shielded volume possible.
Free space loss is 68 dB at one metre at 60 GHz against 60 dB at 24 GHz, so the higher band costs about 8 dB at equal distance. Moving the receive antenna to 16 cm returns about 16 dB, which covers the penalty.
Atmospheric absorption is not the obstacle on a bench either. Oxygen absorbs around 15 dB per kilometre near 60 GHz, which is 0.015 dB The band is short range outdoors and usable indoors.
What a chamber still earns is the pattern measurement below about minus 30 dB, spurious and harmonic search, EIRP for regulatory filing, and repeatability without multipath. Multipath is the real bench problem, because at 5 mm wavelength a reflection path that changes by 2.5 mm moves the measurement through half a cycle, and a hand moving one centimetre changes the reading four times.
What a bench setup has to get right
The interface. A 2.92 mm connector is rated to 40 GHz and a 1.85 mm connector to 67 GHz, so a 60 GHz board brought out on 2.92 mm is already running 20 GHz past its rating with loss and return loss nobody characterised. Above 50 GHz the honest options are 1.85 mm or WR-15 waveguide, which covers 50 to 75 GHz. The cable and the transition have to be measured on a VNA and de-embedded, because a metre of 1.85 mm cable loses a few dB at this frequency and that loss is not flat across the band.
The target. Range resolution is the speed of light over twice the sweep. A 4 GHz sweep gives 3.75 cm, so a corner reflector placed at half a metre has to sit within a few centimetres of where you think it is, or it lands in the next range bin and the amplitude reading is meaningless. For moving targets a rotating plate or a delay line gives repeatable Doppler without a moving part in the beam.
The absorber geometry. Absorber placed inside the near field perturbs the antenna. Keep the first absorber face outside the far field distance of the aperture, and cover the bench surface and the enclosure lid, which are the two reflectors that dominate a tabletop setup.
Three mistakes that make a bench result meaningless
Calling a one metre open air measurement OTA. At one metre the module is in far field, but the room is not anechoic. Bench, lid, monitor and operator all return energy, and the ripple that follows gets read as antenna gain variation. Either shorten the standoff into a shielded box or accept that the number is comparative, not absolute.
Measuring the bare board and shipping the housed unit. A radome, a coating or a plastic housing over the radiator loads the antenna and pulls the resonance down. A 50 micrometre layer of material with a dielectric constant near 3 is enough to move a 60 GHz resonance by hundreds of megahertz, and the shift is invisible on a conducted test. The radiated check has to be run on the assembly as it ships, at least on the first article of every build.
Using the package antenna gain in a link budget after the board is loaded. The gain published for an antenna in package assumes a reference ground and a defined environment. A ground plane with a slot in it, a connector cut into the edge, or a cable routed across the radiator changes the pattern, and the link budget then predicts range the hardware does not deliver. The measured pattern of the assembled unit is the only number worth putting in the budget.
How SUPERB supports a 60 GHz build
SUPERB PCB covers RF and microwave work from 500 MHz to 77 GHz on Rogers RO4003C, RO4350B and RO4835, Taconic RF-35, TLY-5 and CER-10, and PTFE laminates with plasma treatment before plating, including mixed RF and FR-4 stack-ups. Line width and spacing run to 0.05 mm with 0.035 mm fanout on HDI, layer to layer registration is 0.025 to 0.06 mm, impedance is held to plus or minus 5 percent with plus or minus 2 percent on critical nets, and every panel carries a TDR coupon so impedance is measured rather than claimed. High layer count work runs to 68 layers with back-drilled vias held to a residual stub of 0.25 mm or less.
Assembly runs on seven SMT lines in a 5,000 square metre ESD controlled area at 40,000 CPH, with components from 01005 up to 57 mm, 3D SPI before reflow, 3D AOI after, X-ray and CT to IPC-7095 with void acceptance under 5 percent on thermal pads, and twelve zone nitrogen reflow below 500 ppm oxygen. Test covers bed of nails ICT, flying probe and functional test on a fixture built for the product, so a module is proven before it reaches the absorber box. The quality system is certified to ISO 9001, with ISO 13485 for medical work, and components carry a 730 day warranty.
The work is build-to-print: the Gerber, the BOM and the assembly drawing come from you, and the quotation covers fabrication, assembly, test and documentation without an engineering fee. Radar and mmWave programmes are built through the 60 GHz mmWave radar PCBA line and the wider RF module range, on the RF and microwave PCB capability.
If a 60 GHz board is failing a radiated test and the chamber is weeks away, send the stack-up, the connector type and the test distance. The reply states which part of the result the bench can settle on its own, and which part genuinely needs a chamber.