Monday, February 23, 2015

Let's aim above 100GHz

February 20, 2015 // Peter Clarke
Let's aim above 100GHz
Bill Schweber observes that against some earlier expectations connectors are coping at frequencies above 1GHz and even 10GHz. So engineers should have the ambition to go higher.

It wasnt that long ago that most RF circuits operating even slightly above 1 GHz were mostly the-domain of experimental, scientific, or one-off projects, with just a few specialized exceptions such as costly radar systems. The required components were scarce, if available at all, and hard to use; the designs were complex and limited in performance. Equally frustrating, the tools needed to model and design, and the instrumentation needed for test and evaluation, were also costly, awkward, or not up to the job (pick on or more). Mass-production consumer products operating in that part of the spectrum were rare and expensive.

Times have changed, a lot. The development of cell phones, Wi-Fi, and the similarly situated license-free ISM (instrumentation, scientific, medical) 2.4 GHz band has driven availability of all the pieces needed to live and prosper in this >1 GHz region. Now systems are moving higher, with activity in the 5 GHz Wi-Fi band as well, due to crowding and interference issues at 2.4 GHz. Ironically, this at least partially due to the success of the cellular, Wi-Fi ad ISM initiatives at 2.4 GHz.

No single technical development is the enabling factor for this move to higher frequencies; instead, it's the back-and-forth interplay between demand and resources. Demand drives the need for active analog and passive components as well as associated modeling and manufacturing tools, while advances in the components and tools enable the demand to be realized, and then the virtuous cycle begins again.

Of course, 5 or even 10 GHz is not the end of the RF spectrum. There's significant component and end-product work being done at 60 GHz supporting data transfer rates up to 7 Gbit/s, with standards promoted by Wireless Gigabit Alliance via IEEE 802.11ad. Moving up the spectrum, we're seeing automotive radar operating at 77 MHz which will drive design, production, and cost initiatives.

Still, it's a tough battle with many obstacles. While Maxwell's four equations, Figure 1, govern the electromagnetic reality regardless of wavelength, the interplay between the EM fields and materials is a complex function of frequency, and often non-intuitive. For example, his equations do not directly indicate that 2.4 GHz would be the resonant frequency of water molecules, and so the critical frequency for microwave-oven magnetrons due to energy absorption. They do not show that a 60-GHz signal can barely penetrate walls but can propagate via reflections from walls, ceilings and floors.

(Side note: while we call them "Maxwell's equations", they should really be called "the Maxwell-Heaviside equations" as it was really Oliver Heaviside who reduced Maxwell's numerous and extremely complicated expressions to the classic quartet we use today, see "Oliver Heaviside: A first-rate oddity" from Physics Today.)

Even designing and fabricating a coaxial-cable connector is a real challenge as we go past 10 GHz and reach for 100 GHz. A recent article in Microwaves & RF, "Reaching Beyond 100 GHz With Coaxial Connectors," went through the details. It noted that just a few decades ago, the feeling was that coaxial connectors above around 40 GHz would be unlikely because of mechanical limits, and designs would have to rely on more-conventional waveguides. Yet today, vendors such as Anritsu, Keysight Technologies, and others offer coax connectors and cable assembles which go to 110 GHz.

These 100-GHz connectors and cables are hard to fathom. The cable's outer diameter is on the order of 1 to 2 mm, and the connector's inner mating surfaces are smaller, of course; the required slot on the female connector is cut using a saw blade with a 0.05 mm kerf. Even more impressive, these connectors are not one-off, custom-machined and assembled components; instead, they are designed to be manufactured in a production environment with at least modest volumes. Modeling all the EM-field modes in the connector, connector, and their interface is almost magical as practice meets theory at these wavelengths.

Can mass-market use of 100 GHz be that far away? The future is hard to predict, but apparently, the answer is that it is starting to become a real possibility. As we reach further and beyond 100 GHz, we'll cross into terahertz and even optical realms.

As we make the inevitable progress, I'm keeping a poster-size spectrum chart spanning 3 kHz to 300 GHz on my wall for reference, similar to United States Frequency Allocation Chart, Figure 2; the version I have also gives some valuable additional perspective with a small bar chart at its bottom reaching through infrared, visible, ultraviolet, X-ray, and gamma rays, all the way to cosmic rays at 10^25 Hz. 
Bill Schweber, is an electronics engineer and author who has written for EE Times, was analog editor at EDN and prior to that worked in marketing communications for Analog Design and was also editor of its technical journal.

This article first appeared on EE Times' Planet Analog website.

No comments: