Complex RF Mixers, Zero-IF Architecture, and Advanced Algorithms: The Black Magic in Next-Generation SDR Transceivers

Transcription

1 Complex RF Mixers, Zero-F Architecture, and Advanced Algorithms: The Black Magic in Next-Generation SDR Transceivers By Frank Kearney and Dave Frizelle Share on ntroduction There is an interesting interaction between complex mixers, zero-f architecture, and advanced algorithm development. The objective o this article is to establish the basic undamentals o each: the principles o operation and the value they deliver in terms o system design, and then to discuss the interdependability o the three. RF engineering is oten regarded as the black art o electronics. t can be a strange mix o mathematics, mechanics, and, in some instances, just trial and error. t unsettles many a good engineer and many others settle or understanding the outcome rather than the detail. Much o the existing literature jumps straight into the theoretical and mathematical explanation without establishing the underlying concepts. Demystiying the Complex RF Mixer Figure 1 provides an overview o the complex mixer in an upconverter (transmitter) coniguration. Two parallel paths with independent mixers are ed rom a common local oscillator whose phase is oset to one o the mixers. The independent outputs are then summed in a summing ampliier to produce the desired RF output. Figure 2. path analysis. Alternatively, let s assume that a signal tone at requency x is applied solely to the input. The mixer in turn produces an output with tones at the requency ±x. With nothing applied to the input, its mixer output is muted and the output rom the mixer goes straight to the RF output. Path Mixer Summing Ampliier Path Mixer Figure 1. Basic architecture o a complex transmitter. The coniguration has a very useul application. Let s assume, as shown in Figure 2, that we eed a tone signal only on the input, and the input is undriven. Given that the tone at the input has a requency o x MHz, the mixer in the path produces an output at the requency ±x. As there is no signal applied to the input, the mixer in its path produces an empty spectrum, and the output rom the mixer passes straight to the RF output. Figure 3. path analysis. At irst glance it may seem that the outputs rom Figure 2 and Figure 3 are identical. However, there is one critical dierence, namely phase. Let s assume, as shown in Figure 4, that we apply the same tone to both and inputs, and that there is a phase shit between the input channels. Analog Dialogue 51-02, February 2017 analogdialogue.com 1

2 Figure 4. Simultaneous and signal path analysis. n theory it should be possible to have all the energy on only one side o the. However, as the result rom the lab experiments in Figure 5 show, in practice ull cancellation may not occur, leaving some energy on the other side o, known as the image. Also note that energy at the requency is present, known as leakage or L. Other energy is also evident in the results these are harmonics o the wanted signal and are not discussed in this article. For perect image cancellation, the outputs o the and mixers must be o precisely the same amplitude, and be exactly 180 out o phase with respect to each other on the image side o the. the phase and amplitude requirements are not met, then the summing/cancellation process, as shown in Figure 4, becomes less than perect and energy at the image requency will remain. we look closely at the output o the mixers, we observe that signals at the requency plus the input requency are in phase, whereas signals produced at the requency minus the input requency are out o phase. This results in the tones on the upper side o adding while the tones on the lower side cancel. Without any iltering we have removed one o the tones (or sidebands) and created an output that sits entirely on one side o the requency. mplications The use o a conventional, single-mixer architecture produces ± products. Beore transmission, one o the sidebands will need to be removed, usually through the addition o a band-pass ilter. The ilter roll o must be such that it removes the unwanted image signal without aecting the wanted signal. The example shown in Figure 4 has the signal leading the signal by. the coniguration was to change such that the signal led the signal by, then we could expect a similar summing and cancellation, but in this instance all the signal would appear on the lower side o the. Figure 5 shows the results o lab measurements o a complex transmitter. The let hand side shows the test case when leads by, resulting in the output tone placement on the upper side o the. The right hand side o Figure 5 shows the relationship swapped so that now leads by and the resultant output tone sits on the lower side o the. mage Figure 6. Single-mixer image ilter requirements. Required Filter Proile Wanted vs. Phase Delay vs. Phase Delay 2500 Value 2500 Value 2000 Value 2000 Value Wanted Leakage Third Harmonic mage Third Harmonic Second Harmonic Figure 5. Tone placement dependent o the and phase relationship. 2 Analog Dialogue 51-02, February 2017

3 The spacing between the image and the wanted signal directly aects the ilter requirements. Where the spacing is large, a simple low cost ilter with a gentle roll o can be used. the spacing is narrow, then designs must implement a ilter with a sharp response; typically employing multipole or SAW ilters. Hence it would be correct to state that spacing must be maintained between the image and the wanted signal so that the image can be iltered without aecting the wanted signal, and that the spacing is inversely proportional to the complexity and cost o the ilter. Furthermore, the ilter must be tunable in requency i the requency is variable, which urther increases the complexity o the ilter. The spacing between the image and the wanted signal will be determined by the signal that we apply to the mixer. The example in Figure 6 shows a bandwidth signal shited away rom dc. The resultant output rom the mixer places the image 20 MHz rom the wanted signal. n this coniguration, to achieve a wanted signal spectrum at the output, we had to have a 20 MHz baseband signal path to the mixer. o the baseband bandwidth is unused, and the data interace rate to the mixer circuit is higher than necessary. Returning to the complex mixer as shown in Figure 5, we know that its architecture eliminates the image without the need or external iltering. What s more, in a zero-f architecture we can optimize eiciency so that the signal path processing bandwidth is equal to that o the wanted signal. Figure 7 shows a conceptual diagram o how this is achieved. As previously shown, i leads by, there will be an output on the upper side o only. leads by, there will be an output on the lower side o only. Thereore, i two independent baseband signals are generated, where one is designed to produce an upper sideband output only and the other is designed to produce a lower sideband output only, they can be summed in baseband and applied to the complex transmitter. The result will be an output with dierent signals appearing above and below. n a practical application the combined baseband signal would be produced digitally. The summing nodes shown in Figure 7 are solely to illustrate the concept SUM SUM Figure 7. Zero-F complex mixer architecture. The Zero-F Dividend The use o the complex transmitter to generate a single sideband output provides substantial advantages in terms o the RF iltering required to remove the image. However, i the image cancellation perormance is good enough to make the image negligible, we can exploit the architecture more by using it in zero-f mode. Zero-F allows us to take specially created baseband data and produce an RF output with independent signals appearing on either side o the. Figure 8 provides an illustration o how this might be done. We have two sets o and data, where each is independent and encoded with symbol data that can be decoded at the receiver with respect to the phase o the reerence carrier. Reerence Sum 12 Sum 12 Symbol #1 Symbol #2 Symbol #3 Time Figure 8. Taking a closer look at / signaling in a zero-f complex mixer coniguration. nitial observation shows that 1 leads 1 by and that the amplitude o both are matched. Likewise, 2 leads 2 by and their amplitudes are also matched. The independent signals are combined so that = Sum12 and = Sum12. The summed and signals no longer exhibit phase and amplitude correlation their amplitudes are not equal at all times and the phase relationship between them varies. The resultant output rom the mixer places 1/1 data on one side o the carrier and 2/2 data on the other side o the carrier as previously explained and shown in Figure 7. The use o zero-f complements the advantages o the complex transmitter by positioning independent data blocks directly adjacent to each other on either side o. The data processing path bandwidth never exceeds that o the RF data bandwidth. So in theory, the use o a complex mixer used in a zero-f architecture provides a solution that requires no RF iltering while also optimizing baseband power eiciency, delivering lower cost per unit o unusable signal bandwidth. Up to this point, the ocus o this article has been on the complex mixer used as a zero-f transmitter. The same principles work in reverse and the complex mixer architecture can be used as a zero-f receiver. The same advantages that have been described or the transmitter equally apply to the receiver. When using a single-mixer to receive a signal, the image requency must irst be iltered out using an RF ilter. n the zero-f mode o operation there is no image requency to worry about, and signals above will be received independently o signals below. A complex receiver is shown below. The input spectrum is applied to both and mixers. One mixer is driven with, the other with +. The outputs o the receiver are and. n the case o a receiver, it is not as easy to prove empirically what the output will look like or a given input, but i a tone is input above, as shown, the and outputs will be at the (tone ) requency and there will be an expected phase shit between and where leads. Similarly, i the tone were input below, the and outputs would again be at ( tone) requency but this time will lead. n this way the complex receiver can distinguish energy above rom energy below. Analog Dialogue 51-02, February

4 The output o the complex receiver will be the sum o the / inormation representing the spectrum that was received above and the / inormation representing the spectrum that was received below. This concept was described earlier or the complex transmitter where a summed and summed signal is applied to the complex transmitter. n the case o the complex receiver, the baseband processor receiving the summed and summed inormation will easily be able to distinguish upper and lower requencies using a complex FFT. 2 1 SUM n-band mage Baseband Processing SUM SUM = SUM = Rx 1 = 1 + Ø 2 = 2 Ø SUM = ( 1 Ø) + ( 2 + Ø) 1 = SUM 2 2 SUM SUM SUM = ( SUM 2 ) Ø+ ( 2 + Ø) Figure 11. Zero-F implementation restrictions. Figure 9. Zero-F complex mixer receiver coniguration. When the summed and summed signals are received, there are two knowns the summed signal and the summed signal but there are our unknowns, namely 1, 1, 2, and 2. Because there are more unknowns than knowns, it would seem impossible to solve or 1, 1, 2, and 2. However, it is also known that 1 = 1 + and that 2 = 2, and with these two additional knowns it is now possible to solve or 1, 1, 2, and 2 using the received summed and summed signals. n act, we only need to solve or 1 and 2 because the signals are just copies o the signals with a ± phase shit. Limitations n practice, the perormance o the complex mixer has struggled to completely eliminate the image signal. This limitation could be considered as having two pronounced eects on radio architecture design. Even with the perormance limitation, complex F does bring tangible beneits. Let us consider the low F example in Figure 10. Accepting the perormance limitation, we do still see an image. However, that image is heavily attenuated rom that which we would expect to see rom a single-mixer design (see Figure 6). Although the complex mixer continues to require a ilter, the ilter proile can be much more relaxed and its implementation simpler and lower cost. mage Required Filter Proile Wanted Figure 10. Practical implementation o the complex mixer. Note the attenuated image. The ilter complexity is inversely proportional to the distance between the image and the wanted signal. we go to a zero-f coniguration, then that distance becomes zero and the image sits within the wanted signal band. The practical application o zero-f theory has struggled, resulting in in-band image levels that degraded perormance beyond an acceptable level (see Figure 11). The principles o the complex transmitter and receiver only hold true when the phase and amplitude requirements o the and data paths are met. Mismatches in the signal paths will cause inaccurate cancellation o the image signals on both sides o the. Examples o such issues can be seen in Figure 10 and 11. n instances where zero-f is not being used, iltering could be used to remove the image. However, i a zero-f architecture is to be used, then the unwanted image alls directly within the spectrum o the wanted signal and a ailure condition will occur i the image power is large enough. Thereore, the use o zero-f and complex mixing can deliver an optimal system design solution but only when the design can eliminate the phase o amplitude mismatches along the signal paths. Advanced Algorithm Enablement The concept o the complex mixer architecture has existed or many years but the challenges o meeting the phase and amplitude requirements in a dynamic radio environment have restricted its use in a zero-f mode. Analog Devices has overcome the challenge by a combination o smart silicon design and advanced algorithms. The design accepts that there will be signal path impairments; however, these are minimized by smart silicon design. The remaining imperections are calibrated out by seloptimizing quadrature error correction (EC) algorithms. Figure 12 provides a conceptual overview. Silicon Path EC Adjust System EC Advanced Algorithm Control Output Path Mixer Path Mixer Summing Ampliier Figure 12. Advanced EC algorithm and smart silicon design enabling zero-f architecture. 4 Analog Dialogue 51-02, February 2017

5 On AD transceiver devices such as the AD9371, the EC algorithm sits within the on-chip ARM processor. t has constant knowledge o the silicon signal path, the modulated RF output, and the input signal. t uses this knowledge to intelligently adapt the signal path proile in a controlled predictive ashion rather than a kneejerk reactive one. The algorithm perormance is such that it can be best described as digitally assisting the perormance o the analog signal path. The dynamic EC calibration algorithm is just one example, albeit a prominent one, o the advanced algorithms that reside and operate inside AD transceivers. Others such as leakage cancellation coexist and lit the zero-f architecture to an optimal level o perormance. While these irst generation o transceiver algorithms were primarily required or technology enablement, the second generation, such as digital predistortion (DPD), enhance the perormance not just o the transceiver, but o the entire system. All systems have imperections that limit their perormance. Whereas the irst generation o algorithms have primarily ocused on calibrating out on-chip limitations, the next generation uses the intelligence o algorithms to compensate or system perormance and eiciency limitations external to the transceiver. Examples include PA distortion and eiciency (DPD and CFR), duplexer perormance (TxNc), and passive intermodulation issues (PM). Conclusion Complex mixers have existed or many years, but the image rejection perormance that they provided did not allow them to be used in a zero-f coniguration. The combination o smart silicon design and advanced algorithms remove the perormance barriers that had previously impeded the adaption o zero-f architectures in high perormance systems. With the perormance limitations removed, the use o zero-f architecture delivers saving in terms o iltering, power, system complexity, size, heat, and weight (the topic is extensively covered in an earlier article rom Brad Brannon 1 ). n the case o complex mixers and zero-f, we can consider the EC and L algorithms as an enablement unction. However, as the scope o the algorithmic development extends, it provides system designers with increased perormance levels that allow them more lexibility in their radio designs. They may choose enhanced perormance but they may also use the gains achieved rom the algorithm to compensate or lower cost or size components in their radio designs. Reerences 1 Brad Bannon. Where Zero-F Wins: 50% Smaller PCB Footprint at ⅓ the Cost. Analog Dialogue, Sept Dave Frizzelle Dave Frizelle [david.rizelle@analog.com] works as an applications manager in the Transceiver Product Group at Analog Devices Limerick supporting the integrated transceiver amily o products. He has worked at AD since graduation in His previous engineering roles include six years working in Japan and Korea supporting the development and design-in o AD components into advanced consumer products. Frank Kearney [rank.kearney@analog.com] works as algorithm development manager in the Communication Systems Engineering team at Analog Devices Limerick. He has worked at AD since graduation in He recently returned rom China where he held the position o senior applications manager or the systems engineering team in Asia Paciic. He is currently a doctoratal candidate at University College Dublin. Frank Kearney Analog Dialogue 51-02, February