It's important to note that if the return loss on either antenna is significantly worse than 14 dB, serious consideration should be given to improving the match either by adjusting element values in the matching network (if present) or by adding a new "L" matching network. While working on the matching networks, it is usually more convenient to use the Smith Chart view with appropriate port extensions as this affords the best visual indication as the match progresses. Recall that the return loss measurement does not contain any phase information.

Troubleshooting Receiver Problems
Until now, we've focused on solving problems in the transmit section of a WLAN design. Now, let's turn our attention to the receive portion.

Clearly, when implementing a WLAN receiver, sensitivity is the big concern. Poor receiver sensitivity is usually caused by high insertion loss in the front end; poor antenna matching (for solution, see antenna matching section above); and the pickup of RF or IF spurious signals. It is easy to differentiate between these two sources: RF spurious signals will change with channel frequency whereas IF spurs are independent of channel frequency. Let's look at each of these in detail starting with insertion loss in the receiver front end.

The first step in measuring the receiver's front-end insertion loss is to measure the sensitivity using a direct coaxial connection to the antenna port of the device under test DUT. The procedure for this measurement is outlined in a separate technical bulletin, TB382², which may be downloaded from at http://www.intersil.com/data/tb/tb3/tb382/TB382.pdf.

The next recommendation would be to check the sensitivity directly into the low noise amplifier (LNA) by carefully breaking the 50-ohm connection leading to LNA and then inserting a pipe at the break. The difference in sensitivity between this measurement and that measured at the connectorized input to the assembly (previous paragraph) is the insertion loss of the front end.

Note that reference designs may contain "slider" components whose exact location and value may be adjusted to optimize the sensitivity. Typically, "slider" components are located at the LNA input and in the interstage circuit between the LNA output and mixer input. With the coax connected directly to the antenna port of the DUT, experiment with the location of these components and note the effect on sensitivity. Also try altering the value up and down one standard component value and repeat the location experiment. Once the optimum position and value is determined on a typical assembly, the PCB design files should be altered to reflect this optimum position so that the part may be robotically installed correctly during manufacture.

Pickup of RF or IF Spurious Signals
Once designers complete loss and antenna matching measurements, they should move on to diagnosing spurious noise problems in the receiver front end. Many contemporary reference designs use differential (i.e. balanced) in-phase (I) and quadrature (Q) receiver signaling. To assess possible pickup of spurious RF or IF signals, solder short wires to the I+ and I- signal lines leading to the baseband processor or combo baseband/MAC IC. Connect these wires through the differential probe fitted with a 10:1 attenuator head to the spectrum analyzer. If single-ended signaling is employed, however, connect the I signal to the probe and ground the second lead of the probe.

During this measurement, the spectrum analyzer should be set to:

• Start Frequency: 0 MHz
• Stop Frequency: 20 MHz
• Amplitude: 0 dBm

Leave all other analyzer settings at their default values. Set the operating channel to CH6 using the configuration utility. Then, connect a signal generator to the antenna port of the DUT. Set the generator amplitude to minimum (or to standby) and the generator frequency to 2442 MHz (i.e. offset 5 MHz above the channel's center frequency.) Note the spectrum analyzer display and compare it to the typical display shown in Figure 4.

The spectrum consists of band limited noise from the front end, occupying the frequency range from DC to 10 MHz, and gradually tapering off to the background noise level of the IF amplifier. As Figure 4 points out, the background noise of the IF amp should ideally be at least 20 dB below that of the front end signifying that the background noise would not substantially influence the system noise figure.

Observe that there are two significant spurious signals visible in the passband, appearing at frequencies of approximately 11 and 11.5 MHz. There are also two additional out-of-band spurious signals appearing at approximately 16.5 and 17 MHz.

In most reference designs, the IF gain is accurately controlled by an automatic gain control (AGC) loop so we must use an indirect measurement to assess the relative level of spurious signals relative to the noise. The level of the spurious signals may be easily determined with the aid of the signal generator.

Advance the level of the signal generator until the generator's signal becomes visible on the spectrum analyzer. Adjust the frequency of the generator until the signal appears adjacent to one of the spurious signals and then change the amplitude of the generator until its amplitude just equals that of the spurious signal. The RF level of the generator is then equal to the equivalent spurious level reflected to the antenna port.

The noise occupies a broad spectrum of frequencies whereas the spurious signals are usually at discrete frequencies. The total noise power, integrated over the bandwidth of 10 MHz, is therefore much greater than that shown on the analyzer at any one discrete frequency.

In general, if the level of the spurious signal is at least 10 dB below the anticipated sensitivity of the radio, the spurious will not degrade the sensitivity. For example, in the case of a receiver with a "normal" front end, the spurious level should be less than -93 dBm (-83 dBm -10 dB). In the example illustrated in Figure 4 above, the spurious signals at 11 and 11.5 MHz are at -97 dBm, so the above criteria is met.

If the spurious signal is at a very low frequency, it may not be easily discernable with the 0 to 20 MHz scan suggested above. It is therefore prudent to take an additional sweep from 0 to 2 MHz to disclose such spurious signals.

Troubleshooting Spurious Signals
There are two possible sources of spurious signals, RF and IF. RF spurious signals are usually caused by harmonics of the various clock frequencies from the MAC or baseband/MAC chip being picked up in the sensitive front-end circuitry. In order to gain insight as to the exact point of the spurious pickup, try the following:

• If the reference design uses antenna diversity, exercise the diversity switch and note if the spurious levels change with the antenna selection.
• Work your way from the antenna port to the LNA input, removing one component at a time and placing a pipe at the output of the component. Note changes in the level of the spurious signals. Any significant change points to spurious injection at that point.
• Carefully inspect the layout for any vias carrying data, digital power distribution or control signals located near the RF signal path. Given the extremely good sensitivity of a typical receiver, every control line can act as an antenna to inject spurious signals and should therefore be scrutinized.
• Carefully inspect the digital and RF/analog ground system, and ensure that the RF and digital grounds are well isolated from each other. They normally only connect together in one location, usually at the bus connector.

Unlike RF spurious components, IF spurious signals are usually injected between the output of the RF IC and input of the IF IC in the vicinity of the surface-acoustic-wave (SAW) filter. Many reference designs uses a balanced differential structure in this area in order to take advantage of common mode rejection. It is therefore extremely important to maintain precise balance of these critical signal lines in order to reject IF spurious signals. The differential signal lines should be run close together and should be exactly equal in length.

It is also important to maintain exact symmetry in the matching circuitry on either side of the SAW filter. As noted in the section above on RF spurious signals, carefully scrutinize any vias in the vicinity of the IF signal path and ensure that none "pop up" close to a signal line. Any power distribution, control line or signal line can act as an injection source.

Wrap Up
That wraps up our two-part series on troubleshooting 802.11b designs. Using the above techniques, designers can track down problems in their radio designs and, in turn, improve overall system performance.

Editor's Note: To view part 1 of this article, click here.

References

1. Polar Instruments (UK), www.polarinstruments.com
2. Abrahams, Richard L. "Measurement of WLAN Receiver Sensitivity", Technical Bulletin, TB382, ( http://www.intersil.com/data/tb/tb3/tb382/TB382.pdf)

About the Author
Richard Abrahams is a senior principal engineer in the Wireless Applications Group of Intersil Corporation. He received the BSEE and MSEE from Rensselaer Polytechnic Institute. Richard can be reached at rabraham@intersil.com.