With much interest I read Yeap Yean Wei's article, "Design A Simple, Low-Cost UWB Source," in the December 2006 issue of Microwaves & RF (p. 68). It reminded me of my old work with frequency multipliers using step-recovery diodes.

I am finding one problem in both types of devices: your UWB and my frequency multipliers. The problem is that the load must be the 50-ohm termination. In my frequency multipliers, this load had to be well matched at a particular frequency of interest. Your UWB source has to be loaded into a 50-ohm load, which has to be well matched over a very wide spectrum: DC to 1/100 ps ~ i.e., >10 GHz.

I am no expert in UWB but I have not seen any antenna "well matched" to 50 ohms over such wide band, and I am afraid such an antenna does not exist.

I have seen a study by Picosecond Pulse Labs in which several antenna structures were evaluated for UWB use.

I think it would be a useful next step in the work that you reported to connect a pair of antennas to your UWB source, and to adjust the complete response to an optimum.

By the way, several manufacturers have reported developing UWB chips promising wonderful data-transmission features, like >480 Mb/s over 10 meters, etc. It seems to me that so far nothing has happened in their introduction to waiting markets. One problem could be that UWB devices seem to function well in the ideal laboratory conditions, but connecting them to real antennas, mismatched by themselves, and mismatched even more by surrounding objects, might well limit the promised features.

Thank you for an interesting article.

Still, more must be done . . .

Jiri Polivka
Spacek Labs, Inc.

Yeap Yean Wei responds: Research on the UWB components and antennas has proliferated since the FCC released its First Report and Order 1 to permit the unlicensed use of UWB devices in the 3.1-to-10.6-GHz frequency band. I do agree that many of the UWB antennas reported are wideband but not "well matched" to 50 ohms across the whole band. However, the challenge does not lie merely on the matching from the impulse generator to the antenna. In order to meet the UWB emission mask, the pulse-shaping circuit must be inserted between the impulse generator and the antenna. Introducing an additional impulse-shaping circuit will further impair the impedance matching across the whole band. If a conventional planar bandpass filter is chosen as the shaping element, the group delay will be affected too. Since low-power UWB technology is targeting the consumer electronic devices for PAN, the size of each building block in a UWB system is particularly crucial. In this context, some trade off must be made. The impulse generator discussed in the article was the initial work carried out two-to-three years ago. Subsequent works such as the pulse shaping using active components and UWB signal propagation with this impulse generator has not published publicly. If time permits, I will try to consolidate the findings into an article. My current focus has been diverted to RF module design since joining Fujitsu Media Devices in Singapore.

According to the WiMedia Alliance's specifications (already approved as ECMA's standard), the targeted PHY data rate of 480 Mb/s is specified at 3 m. Therefore, the end-user experience at application layer will definitely be less than 480 Mb/s due to an overhead introduced at every layer. Mismatch within the components is just one of the problems. Other issues such as the resistance of signal propagation due to device casing (most likely to be fully covered), the antenna placement and orientation, coupling within the circuitries and propagation environment are more predominant.