Combining high power and high frequency in rf and microwave.
The following paragraphs are excerpts of an interview with Alen Fejzuli, president of Res-Net Microwave, Inc. His plain-speak comments are intentionally "teched down" for the listener and writer, a non-engineer.
On the design approach:
Designing for rf and microwave requires solid understanding of transmission line theory, and application of "high frequency rules," if you will, to the design of our components. The Smith Chart, wavelength, VSWR -- these items are all foreign to the world of digital design.
On the products and basic challenge:
Res-Net Microwave has been known for more than 25 years as a manufacturer of high power components -- resistors, attenuators, and terminations. Typically, when you're designing high power components there are physical limitations. The world of physics is against you. You're trying to make high frequency components that also need to dissipate a lot of power. In order for a resistor or terminator to dissipate a lot of power, it tends to require considerable size. However, as frequency increases, wavelength decreases, and once wavelength becomes smaller than the component, a number of effects take place that change the properties and performance of the component. For example, at low-frequency resistor looks like a resistor. On the other hand, at high frequency a resistor will have parasitic capacitance and inductance, which limits the high frequency performance.
On size and power:
How well a component dissipates power is driven by the type of material that is used, for example, we use Alumina, Aluminum Nitride (AlN), Beryllium Oxide (BeO), and, now, synthetic diamond (CVD diamond). Each of these materials dissipate power at a specific rate, AlN, BeO, and CVD in increasing order (lowest to highest, Figure 1). With the introduction of CVD diamond material into our manufacturing processes, the differences have become dramatic. For example, we have little (CVD) resistors that are 40 by 20 mils that can dissipate 20 watts, even in that small size. A BeO component of the same size might dissipate 5 or 6 Watts, an ALN maybe 3 Watts.
On size and LOTS of power:
If you look at a 1000-watt resistor on BeO material, it is 1 square inch in size, pretty big! We can reduce the size of that component down to roughly __0.2 square inch___ by using a CVD diamond substrate to create a component of the same power rating. Furthermore, due to its size, the larger high power resistor also acts like a big parallel plate capacitor. The combined effects of the resistor and capacitor create a low pass filter which limits the high frequency performance of these high power components.
More generally, capacitance is always present in our world, and that's why there's always a tradeoff between power and frequency.
On size and frequency:
This benefit of reducing component size without sacrificing power dissipation brings us back around to the wavelength factor -- once the wavelength becomes smaller than the component, reflected power and VSWR increase with power (presuming frequency remains the same), parasitic capacitance increases with frequency (presuming power remains the same), and general performance diminishes quickly with increases in both power and frequency.
And that's how Res-Net has made the world of physics work for us, instead of against us. By minimizing the size required to dissipate the power called for, we're not only maximizing power -- we're also maximizing the high frequency range of the component and circuit.
That 40 by 20 mil CVD diamond resistor I mentioned previously operates well up to 35 GHz at 20 watts. We have miniature terminations spec'd at 26.5 GHz, 50 watts.
Wrap-up: Small is good. Very very good.
In the digital world, the push to smaller and smaller components is driven mostly by space and weight limitations and the general push for miniaturization. In the rf and microwave world, there's a 4-way relationship between size, power, frequency/wavelength, and performance. We like to think we're pushing the limits on all fronts.