May Expert Advice

May 2009


Gayle Collins is a RF High Power Amplifier Applications and Design Engineer at Freescale Semiconductor in Tempe AZ. She has over 16 years experience in control, power and RF engineering. Having spent most of her career in R&D she decided to move into applications and design to add depth to her engineering experience. She previously was a senior design engineer in Advanced Development at RFMD in the handsets side of the business. Prior to that Gayle was a R&D RFIC Design and Product Engineer at Agilent Technologies in Santa Rosa, CA. She holds a Master of Science in Electrical Engineering from Arizona State University. Her interests include nonlinear analysis of RF PAs, high power PA design and filter synthesis for broadband matching.

To comment or ask Gayle Collins a question, use the comment link at the bottom of the entry. The first 5 people to comment will receive a copy of the Electrical Engineering Handbook (please include your e-mail and mailing address).

This years IMS holds some stellar offerings for the RF high power amplifier engineer. The cellular market demands RF power amplifiers that are highly efficient, linear, have good output power and are not cost prohibitive. The all gold and not to be missed session this year is TH3A High Power Amplifiers (Tuesday starting at 1:20 pm).

This is true high power PA design with all output powers well over 50 dBm (100 W). The third paper, ‘A 33 W GaN HEMT Doherty Amplifier…’ is a very complete paper in terms of analysis, design and implementation. From a commercial standpoint this is probably the most useful GaN PA paper in the conference. A 2.6 GHz PA with an average Pout of 45.2 dBm and a Psat of 52.5 dBm is presented, fully characterized in all practical aspects including DPD with some very good results. The methodology promises to be quite practical and easily implemented, independent of the technology.

Another stunning paper in this session is the ‘250 Watt Doherty PA (TH3A-4) based on Two-stage Power IC for 1.8 GHz GSM Applications’. This is high power nirvana. The results show 70 dBc correction with a two carrier GSM signal at 47 dBm out and 60 dBc correction with a four carrier GSM signal at 45.5 dBm out. TH3A-2 is a very good offering as a novel example of a multichip module from the Freescale RF power design team presented by one of their up and coming engineers. The innovation here is the development of a new over-molded plastic package that reduces cost while handling an output power of 200 W, the highest power level reported for this package type that is accomplished through the use of integrated passive devices. The first paper, TH3A-1 benchmarks a 340 W C-band PA with over 50% PAE.

Session TH1A (Thursday starting at 8:00 am), Power Amplifier Linearization, includes two must-see papers, as well. TH1A-1, ‘An Improved Doherty Amplifier…’ describes a technique for improving linearity at the same time as the efficiency and peak power in a Doherty amplifier. High order memory-less nonlinearities and low order memory effects are compensated with pre-distortion. The gate voltage of the peaking amplifier is digitally modulated as a function of the signal envelope between Class C and Class AB bias. This is done in such a manner as to reduce the variance of the AM/AM distortion of the full amplifier. The boost in gate voltage at higher envelope levels increases the peak power capability of the amplifier and improves efficiency. The ability of this technique to improve all three performance metrics of the Doherty amplifier is very promising.

TH1A-2, ‘ Fully Orthogonal Multi-Carrier Linearization…’ presents a methodology for the mitigation of strong memory effects and a scheme is presented for the removal of unwanted in-band frequency components generated by the inter-band pre-distortion correction. The linearization technique is demonstrated with an LDMOS PA at 3.5GHz using a two-carrier 8MHz OFDM signal with 16MHz channel separation. ACLR of up to 45 dBc is reported for both in-band and inter-band.

In HF/VHF/UHF Power Amplifiers, session WE4A (Wednesday starting at 3:40 pm) the first and the 4th papers look quite interesting. The first paper is the result of a collaboration between Frederick Raab’s Green Mountain Radio Research and CICESE Research Center. High efficiency and good dynamic range are provided through an asymmetric combing method in a Chireix out-phasing system that uses a Class E PA. As a result linear amplification of amplitude-modulated signals with high average efficiency is achieved. The technique is verified experimentally with a 28W transmitter at 1.82MHz that attains 85% or greater efficiency over 10dB of dynamic range. The 4th paper by Allen Katz, et. al. reports a stunning 70% linear efficiency and at a saturated power of 100W, a PAE of 90% is reported. This is from a GaN device developed for avionics and space applications but results like these are definitely worth a look from those of us in the cellular infrastructure end of the business.

The UCSD group et. al. have put in an engaging contribution in session WE2A (Wednesday starting at 10:20 am), Millimeter-Wave and CMOS Based Power Amplifiers (WE2A-1). A stacked FET structure is exploited in order to overcome the low breakdown voltage of a MOSFET. Four transistors are stacked and connected in series such that the output voltages combine in phase. This architecture increases the breakdown voltage of the amplifier to a 9V supply voltage. The single stage amplifier is demonstrated to attain a PAE of 47% at 1.9GHz and a saturated output power of 33dBm at a 6.5V supply. The second paper in the session, ’60 GHz CMOS Power Amplifier with 20dB Gain and 12dBm Psat’ reports the highest gain with high output power and high efficiency for a CMOS PA at 60 GHz. The PA is demonstrated to obtain a linear gain of 20 dB with a P1dB level of 8.2dBm.

Thursday’s interactive forum on High-Power Amplifiers has a number of papers of particular interest. THPC-5, ‘Dynamic-Load Modulation …’ with varactor based matching networks describes a method similar to Doherty design but with dynamic loading rather than static. This is combined with a dynamically controlled input signal to achieve the desired performance. It will be intriguing to see the performance metrics that were obtained. Next, ‘Inward Nonlinear Characterization of Doherty Power Amplifiers’ is very timely given the current level of interest in the performance achievable from Doherty configured amplifiers. A time domain model is used for the nonlinear models of the carrier and peaking amplifiers of a Doherty. Time domain is the natural home of nonlinear behavior and memory effects are captured for free. The authors were able to identify input-power dependent phase mismatch between the carrier and peaking sides that was causing distortion. This technique promises insight into the inner function of the Doherty amplifier. THPC-1, ‘Design of a Current Mode Class-D RF Amplifier…’ uses a measurement based design method based of characterizing the LDMOS FETs using active harmonic loadpull. Recognizing that CMCD amplifier is inverted Class F in push-pull configuration, the loadpull impedances were determined. 61.5% PAE at 25 watts out is reported at 900MHz. THPC-3, ‘A General Method for Passband Quantization Noise Suppression…’ offers a technique for selective passband noise suppression in pulsed transceiver architectures that will alleviate the limitations of dynamic range by the noise present thereby facilitating correctibility and improving performance.

In working with high power PAs, the engineer is most rightly concerned with the non-linear behavior of the amplifier. To this end the session on Nonlinear Modeling, WE3G is of interest. The Thursday session on ‘Instrumentation for Nonlinear Characterization’ is also useful for engineers in the high-power PA field.

Session TH4A (Thursday starting at 3:40 pm), High Power Amplifier Techniques, contains several papers on novel high-efficiency PA architectures using adaptive bias control and drain modulation techniques, or Envelope-Tracking, from the university groups at UCSD and POSTECH, Korea, who are well established in these fields. Are these the techniques for future high-efficiency RF PAs? Why not show up and find out how these new circuit and control methods can improve the PA efficiency. Also in this session is a paper on a high-power, high-frequency gallium nitride PA, from Toshiba, setting benchmark power and frequency performance. Talking of GaN, there is a session devoted to GaN PA applications, WE3A – Application of Gallium Nitride Technology from L to V Band, showing PA design examples at the higher frequencies where GaN has an advantage over LDMOS technology.

All in all, there are a couple of very full days of interesting presentations for RF PA designers at IMS2009.

Post a comment to this article