Microwave Engineering Europe - March 2008 - (Page 12)

12 WIRELESS INFRASTRUCTURE — RECEIVER DESIGN up to 30 dB of baseband voltage gain in the architecture shown in Figure 1, without the need for AC-coupling or DC offset cancellation. Extraneous DC offsets can also be generated from self-mixing of the LO signal due to LO-RF coupling, or from self-mixing of the RF signal due to RF-LO coupling. This DC offset tends to be dynamic in nature due to the widely varying RF input signal that can either be the desired or from an interference source. This offset component is largely suppressed by the LT5575’s inherent 60 dB isolation between the RF and LO ports. However poor external circuit layout can degrade this isolation performance. Therefore particular attention must be paid to the PC board routing of the connections to these two ports in order to preserve the device isolation (more on this in the IC Layout Consideration section later in this article). Second order distortion contributes to errors For a conventional superheterodyne receiver, third order distortion is dominant, producing intermodulation terms of the form cos {(2ωi – ωj)t }, which may be in-band, depending on the frequencies of the interfering signals. For direct conversion receivers, however, second order distortion can also play an important role in limiting performance. That’s because 2nd order distortion contributes a baseband term of the form cos (ωi – ωj)t. This term is out-of-band for a superheterodyne receiver, but is likely to be in-band for a direct conversion design. In an actual application, 2nd order distortion can become a problem in the presence of a single strong nearby interferer. Because channel selection is performed at baseband, and usually in the DSP, even those interferers which are not “on-channel” may still pass freely through the RF & baseband ﬁlters. These interferers are then more likely to produce unwanted products that fall directly on top of the desired baseband signal, at which point they cannot be ﬁltered out. Second order distortion is most troublesome in this regard, because it can yield such products as a result of even a single interfering signal. Speciﬁcally, 2nd order distortion due to a tone interferer will give rise to a DC offset at the mixer output. If the interferer is modulated, then a modulated signal due to 2nd order nonlinearity will appear at the baseband output. The second order intercept point (IP2) of a direct conversion receiver system is therefore a critical performance parameter. It is a measure of second order non-linearity and helps quantify the receiver’s susceptibility to single and two tone interfering signals. The LT5575 minimizes the impact of 2nd order distortion Figure 1: Example of a direct conversion receiver for a W-CDMA application. due to its exceptionally high IIP2, 60 dBm at 1900 MHz and 54 dBm at 900 MHz. Port-to-port coupling In a superheterodyne receiver, careful selection of the LO and IF frequencies can usually minimize the resultant mixing products in the IF passband that arise from port-to-port coupling. No such protection is afforded with the direct conversion receiver architecture, where the RF and LO frequencies are the same, and where their mixing products appear directly at the baseband I/Q outputs, as described previously. The LT5575 speciﬁes its LO–RF leakage of -60 dBm or better (to 2100 MHz), and RF–LO suppression of 57 dBc or better, to largely eliminate this concern. I – Q mismatch Ideally, the I and Q channels of a radio signal carry orthogonal, i.e. non-interfering, channels of information. However, mismatch in the gain or phase of the I and Q channels results in interference between the channels, which makes it more difﬁcult to recover the information they contain. Modern digital communication systems specify a maximum Error Vector Magnitude (EVM), typically on the order of a few percent, which can be related to the gain and phase mismatch error of the I/Q channels. Phase mismatch error, arising from unequal time delays in the I and Q signal paths, from inaccuracy in the LO quadrature generator, and from port-toport coupling is particularly problematic. The higher the frequency, the more severe is the phase error problem. This is why I – Q mismatch is more of an issue for direct conversion receivers. The LT5575 speciﬁes a typical phase error of 0.5º, and a typical gain error of 0.04 dB, resulting in an EVM penalty of about 1 percent for QPSK-type modulation formats. High frequency termination of the baseband I/Q outputs One key advantage of the direct conversion architecture is that it eliminates the classical image rejection problem. This is because there is no longer an image frequency that will yield a mixer output signal at the desired baseband frequency. There is, however, still an unwanted product that will be generated by even a perfectly linear mixer. This signal is the sum of the RF and LO frequencies, and it will appear at the mixer output well above the baseband frequency. Consider for example a 1900 MHz RF application; the LO frequency will also be at 1900 MHz. The desired output at baseband will be accompanied also by a signal at 3800 MHz, which is the sum of the RF + LO frequencies. It may seem obvious that the baseband ﬁlters following the mixer will completely suppress this sum product. Well, maybe not. Any integrated circuit uses internal wire bonds to connect the IC chip to the terminals of its package. These wire bonds act as small inductors which at high frequency tend to isolate the chip from any external ﬁltering elements. If no on-chip ﬁltering is used, then the sum frequency signal as well as high frequency distortion products will manifest themselves on-chip in unpredictable ways. Most importantly, any improperly terminated high frequency signals will consume the signal headroom of the chip, causing degradation of the chip’s intrinsic linearity. Figure 2 shows the equivalent output circuit of the LT5575, which contains on-chip 5 pF capacitors on each of the IOUT+, IOUT-, QOUT+ and QOUT- outputs. These on-chip capacitors, augmented by offchip termination capacitors as needed, serve to mitigate the Image problem and optimize linearity for any given application. More detailed discussion of many of these factors and others can be found in [1]. Microwave Engineering Europe ● March 2008 ● www.mwee.com 011_012_013_014_MWEE.indd 12 20/02/08 12:03:51 http://www.mwee.com

Table of Contents Feed for the Digital Edition of Microwave Engineering Europe - March 2008

Microwave Engineering Europe - March 2008 News Contents Comment Wireless Infrastructure: A Direct Conversion I/Q Demodulatordrives Favorable Basestation Cost-performance Metrics Wireless Infrastructure: Mobile World Set to Reshape the Internet RF Amplifiers: Latest Advances in RF Amplifiers Include a CMOS PA Operating at 77 GHz and Significant Advances in PAs for WiMAX and Broadband Applications Many Applications Still Require Unique Performance Benefits of BeO ACE Automated Circuit Extraction Returns to Real Design by Exploring Design Alternatives and Changes in Seconds Exceeding the Standard for Wireless Sensor Networks Products Calendar

Microwave Engineering Europe - March 2008

For optimal viewing of this digital publication, please enable JavaScript and then refresh the page. If you would like to try to load the digital publication without using Flash Player detection, please click here.