Waveguide Filters: New Design Techniques for Exploiting a Mature Technology
By Andy Trusler, Microwave Division Manager, Luso Electronics
Although waveguide technology has been around for many years, there are still applications where it offers huge advantages over RF coaxial cables or digital-based solutions. This is particularly true in the area of satellite and military communications, where the power handling, low loss, reliability, stability, and electromagnetic shielding capabilities of waveguide assemblies make them the technology of choice for RF distribution networks. At the same time, the latest advances in design and manufacturing techniques are continuing to improve the performance of these components, meaning that their mechanical and electrical properties can help optimize system performance for the most demanding applications.
Much of the theory of waveguides and their performance was refined around the middle of the 20th century, and were most notably laid out in 1951 in the Waveguide Handbook  — one of the classic MIT Radiation Laboratory Series of books, which is still in print today. However, while the basic principles remain the same, the practice of designing waveguide components has now changed beyond recognition. Because commercial electromagnetic simulation software now provides such accurate results, prototyping is kept to a minimum and time-to-market has been dramatically reduced. Also, the latest CNC machines routinely work to a tolerance of ±0.02mm, which produces a reliable and repeatable physical interpretation of the modeled components.
The benefits of this approach are illustrated here by two examples: firstly, the design of a custom filter for a C-band receive system, which was required to block interference from altimeters in aircraft landing at a nearby airport; and secondly, a WR284 waveguide to 7/16 coaxial end launch adaptor for use in radar systems at airports.
Filter Design and Simulation
In the first case, the customer requirement was for just two filters. Conventionally such a small custom production run would have necessitated a high non-recurring engineering (NRE) charge and a long lead time, during which a theoretical design would be converted to a machined or fabricated prototype and then optimized empirically. However, the use of accurate electromagnetic simulation tools, followed by rapid transfer of the resulting design into manufacturing, has reduced both the cost and the time required to realize a practical component or subsystem.
The WR229 radar altimeter filter was specified with a passband in the 3.70-4.20GHz C-Band satellite communications band, and was required to eliminate interference both from S-Band radars in the band 2.40-3.10 GHz and from aircraft altimeters in the band 4.25-4.40GHz, specifically for use when the receive station is located in close proximity to an airport.
The waveguide filter was designed with a standard WR229 aperture at the flanges, tapered down internally to just above half the standard waveguide height, which reduces the characteristic impedance of the waveguide. The filter was realized as a single sided E-plane iris-coupled filter with rounded corners, having six resonant bandstop cavities each coupled to the main waveguide run by a 3-4mm thick iris across the waveguide broad wall. The iris aperture is rectangular, and the thickness of the irises is small relative to the wavelength at this frequency, hence the attenuation of each iris is negligible. Since the required power handling capability was relatively low, it was not considered necessary to use a double-sided iris filter design.
CST Microwave Studio was used to design the filter, which was created and then optimized on an overnight run. It was decided to optimize the design as a band rejection filter, in order to achieve the tight rejection required in the higher frequency band.
Figure 1 shows the S-parameters for the simulated design. Figure 2 shows the optimized mechanical design of the filter, which was immediately transferred into production on Luso Electronics’ in-house CNC machines. For ease of manufacturing, the unit was manufactured as a split-block aluminum fabrication, where the split was along the center line of the H-plane, since this is where the current flow is at a minimum. No special machining processes were required, but the piece parts were passivated before assembly to prevent oxidation, which could otherwise affect the electrical integrity of the joint between the two halves of the filter assembly. A set of tuning screws was also incorporated, to allow fine tuning of both insertion loss and return loss. The tuning process was very simple, requiring a total tuning time of around 5 minutes for each unit.
The filter was measured using an Anritsu VectorStar network analyzer with coaxial to waveguide adaptors at each port. The adaptors were de-embedded during the open-short-open calibration process. Figure 3 shows the practical results for the filter, which demonstrate a very strong correlation with the simulated results in Figure 1. Maximum insertion loss is 0.4dB across the 3.70-4.20GHz passband, with a return loss of 20dB minimum. Minimum stopband rejection at both high and low bands is 40dB. The overall length of the filter was 264mm.
The production filter is shown in the photo in Figure 4.
Coaxial End-Launch Adaptor
The second design was a WR284 waveguide to 7/16 coaxial end launch adaptor, and again this was fully designed and optimized on CST Microwave Studio. The coaxial adaptor was designed with a standard WR284 aperture, and a multi-ridge platform was used to minimize return loss. It was discovered that there was variation in the match between the 7/16 connectors and the adaptor body, so a tuning stub was incorporated so that this deviation could be compensated for.
The adaptor was simulated using realistic machining tolerances to show the expected deviation in performance for small dimensional variations. Figures 5 and 7 show the simulated and measured S-parameters of the adaptor respectively, for which the mechanical model is shown in Figure 6. The optimization was sufficiently accurate—within a few percentage points—that no prototype was required, and consequently it was not necessary to charge NRE.
This design also demonstrated the ability to make adjustments to the design in order to ease the manufacturing process, both to reduce the cost and to improve delivery timescales.
Overall, initial design to finished manufacture was accomplished within 4 weeks.
The platform was machined as an integral part of the internal aperture; a wire erosion machine was used to machine both the aperture and the platform in one operation. This ensured a very tight tolerance could be achieved to aid repeatable RF results. It also reduced the manual assembly required on the unit.
This article has described the advantages of using electromagnetic simulation to design and put into production two custom components: a high-performance waveguide filter and a waveguide-to-coaxial adaptor. Had conventional design techniques been used, lead time would have been significantly increased, with an associated increase in cost. In each of these cases it was not necessary to charge NRE, since the entire design and optimization process took just one day or less.
Additionally, this process instills complete confidence that the end product will meet the required specification. Modern design and simulation tools have enabled the waveguide design engineer to design real world solutions to customer requirements, and to achieve this within realistic timescales and at minimum cost. Transferring the designs into Luso Electronics’ newly commissioned manufacturing facility in the UK also allows the design to be produced rapidly, accurately and repeatably.
 Waveguide Handbook by Nathan Marcuvitz, MIT Radiation Laboratory Series.
this article to a friend!