There are really good reasons for considering commercial, off-the-shelf (COTS) products rather than custom designs in many technology and application areas. Indeed considering COTS is mandatory for many organisations and even enshrined in law for US government procurement.¹
The main reasons for using COTS components are rapid availability, low cost and low risk. Clearly in many markets it is far better to buy a product that is in stock, already sold by the thousand and is well proven than to design something special for the purpose.
But this isn’t true for millimeter wave waveguides!
Why not? Because the demand for waveguide components is relatively limited, chances are that the component will have to be made to order. Whilst Flann has been making millimeter wave components to 500 GHz for many years, some other manufacturers are just starting to explore the design and manufacturing challenges, and may not even have the necessary test equipment available, so picking a component out of a catalogue isn’t necessarily low risk.
The main reason that COTS makes no sense for millimeter wave waveguides has to do with flanges. Up to around 18 GHz the standard waveguide flanges, used to connect waveguide components together, are in reasonable proportion to the component itself.
Above 50 GHz though, flanges based on the ¾ inch (19 mm) UG-387 (or IEEE 1785.2a) predominate.
In these, the flange diameter might be more than 20 times bigger than the waveguide width! By the time the design engineer has allowed space for fixing screws, and clearance for the tool to do them up, the waveguide device ends up far bigger and heavier than necessary.
The following photos show a waveguide low-pass filter structure – the first image is for a filter working up to 40 GHz, whilst the second is the same design, but scaled to 110 GHz. The millimeter wave filter is much bigger than it needs to be, just because of the flange design.
The second reason to avoid COTS for millimeter wave waveguide is performance. Every pair of flanges causes loss and unwanted reflections from misalignment, dimensional tolerances or flange cocking (see picture). If the manufacturer combines the components into a single block then this performance limitation disappears. In fact in many cases knowing more about the requirement allows the manufacturer to use design tricks to improve the performance beyond that achievable using discrete waveguide components.
The final reason to avoid COTS for millimeter wave waveguide is cost – all that extra machining for an over-sized body and accurate flange alignment detail costs money, and combining components into single block sub-systems saves it!
With this in mind, buying COTS for millimeter wave waveguide should be limited to prototyping (where flexibility is required in system configuration and layout) and Test & Measurement (where system performance might be less important than performance and repeatability of individual components). Even in these cases, though, it would be worth choosing a supplier like Flann with the engineering expertise to build components into a single unit as the project progresses in the future.
So in summary, please talk to us before you design in a millimeter wave system based on separate waveguide components – we might be able to save you space, improve performance and save you money!
 “Agencies shall […] Require prime contractors and subcontractors at all tiers to incorporate, to the maximum extent practicable, commercial items or non-developmental items as components of items supplied to the agency.” (Federal Acquisition Regulation (FAR), Subpart 12.101 2007)