Memory effect was checked on the single-building block HBT by varying the two-tone frequency spacing from 1 MHz to 120 MHz. Figure 3 demonstrates the test results for average power levels from 10 dBm to 30 dBm.

The test was done from 1 MHz to 120 MHz. The curves are from 10 dBm to 30 dBm of the average power (PEP is 3 dB higher). The IMD3 is fairly flat across this frequency band.

Figure 4 compares the 1BB power HBT (with P_sat ~ 33 dBm) tested under two-tone and WCDMA signals at 2.14 GHz with and without the dynamic bias circuit in the near- class B bias condition. The improvement in linearity by the dynamic bias circuit is most noticeable over the output power range within 15 dB of the P1dB. At 20 dBm average P_out, the ACLR of WCDMA signal improves over 10 dB. The IM3 improves by 15 dB at 23 dBm output power. At high power levels where the peak power exceeds P_sat, the clipping of current or voltage waveforms becomes the dominant non-linear effect and cannot be improved by the dynamic bias circuit.

Temperature response of the dynamic bias circuitry: The RF performance of the 28 V HBT with dynamic bias circuit scales well with the sizes up to four building blocks (or ~38 dBm saturated power). The dynamic bias circuit also demonstrates excellent temperature stability in RF operation as evaluated over the -40 C to +85 C range. A 2BB size HBT with the dynamic bias circuit was evaluated over temperature with its result shown in Figure 5. At ACLR = -50 dBc under the WCDMA signal, the output power is maintained within 1 dB; the efficiency at ACLR = -50 dBc is 16% over the same temperature range.

The power HBT with dynamic bias circuit also performs well on other modulation schemes.

Several rounds of accelerated lifetime tests were conducted. At 28 V bias and 0.05 mA/m² quiescent bias current, 3000 hour tests were repeated at a junction temperature of 310 C. This level of lifetime is expected as the 28 V HBT operates at lower current density than its 5 V counterpart since the current density is one key factor to the lifetime. The collector sidewall in the 28 V HBT has not shown any leakage in the accelerated lifetime test. These tests demonstrate that 28 V HBT not only maintains the same level of lifetime as the 5 V InGaP/GaAs HBT^[12], but also is rugged against mismatch, RF power overdrive, and V_CC fluctuations.

The result is shown in Figure 6.

A Gummel plot was taken before and after the 3000 hours accelerated life test. Figure 6 presents the outcome. Little change can be noticed down to the nA range.

The power device, under normal usage, may be subjected to mismatch or overdrive conditions and it must be rugged enough to survive such conditions. The ruggedness is related to the level of ballasting. Under the presently chosen ballasting level, the three HBTs of 1BB, 2BB and 4BB were individually tested for output mismatch. The input power is held constant at the level providing output power of P_1dB under normal operation; then the output load is changed from 50 Ω to 10:1 under all phases. The test started at V_CC = 24 V and went all the way up to 30 V without any observed device degradation.

With the peak-to-average ratio (PAR) of the modulation signal in the 6 dB to 10 dB range, it is not uncommon to have the peak power level overdrive the amplifier way beyond the 1 dB gain compression point. Therefore, another test was conducted with the RF overdrive under the normal matching condition with the collector biased at 28 V. For all three sizes, 8 dB gain compression at 2.14 GHz was achieved without damage or degradation with the sine wave CW signal.

A small number of samples also underwent an RF burn-in. The HBT hybrid amplifier was driven to P1dB level at room temperature for more than 500 hours, and no change can be noticed. All these experiments demonstrate that a InGaP/GaAs HBT can be a rugged, reliable, and linear device for industrial applications such as a base station amplifiers.

InGaP/GaAs HBTs were designed for the 28 V power application. The semiconductor device structure and microfabrication procedure were introduced. The 28 V process borrows many steps from its low-voltage counterpart. Power transistor design to the 10 W level was discussed. Special attention to the balance of the phase and magnitude of the RF signal across all the HBT fingers is essential to maintain the performance. High linearity in class AB and near-class B operation was achieved with the dynamic bias circuit approach.

The combination of the high linearity RF performance in the back-off power level, the ruggedness in RF power overdrive and the output mismatch condition, and the long lifetime demonstrated that InGaP/GaAs HBT technology is mature to serve the 28 V linear power operation in infrastructure applications.

The authors acknowledge the fruitful discussions with Walter Strifler and the technical assistance of Huy Pham and Sindolfo Gacayan. The effort by C. Dunnrowicz in the early phase of the project is appreciated.

1. D. Hill, T.S. Kim, 28 V Low Thermal-Impedance HBT with 20 W CW Output Power, IEEE Trans. Microwave Theory and Tech., vol. MTT-45, No. 12, pp. 2224-2228, December 1997.

2. P. Kurpas et al, High-voltage GaAs Power-HBTs for Base-station Amplifiers, 2001 IEEE MTT-S Int. Microwave Symposium Digest, pp. 633-636, June 2001.

3. N. L. Wang, C. Dunnrowicz, X. Chen, H. F. Chau, X. Sun, Y. Chen, B. Lin, I. L. Lo, C. H. Huang and M.H. T. Yang, High-efficiency 28 V class AB InGaP/GaAs HBT MMIC Amplifier with Integrated Bias Circuit, 2003 IEEE MTT-S International Microwave Symposium Digest, vol 3, pp, June 2003, pp. 2397-2402.

4. N. L. Wang, C. Dunnrowicz, W. Ma, X. Chen, H. F. Chau, X. Sun, Y. Chen, B. Lin, I. L. Lo, C. H. Huang and M.H. T. Yang, Linearity Improvement of Multi-Watts 24 V-28 V InGaP/GaAs HBT by Low-frequency Low-source Impedance Matching, 2004 IEEE MTT-S International Microwave Symposium Digest, vol. 3, pp. 1721-1724, June 2004.

5. N.L. Wang, W. Ma, C. Dunnrowicz, X. Chen, H.F. Chau, X. Sun, Y. Chen, B. Liin, I.L. Lo, C.H. Huang, M.H.T. Yang, C.P. Lee, 28 V High-efficiency High-linearity InGaP/GaAs Power HBT, 2003 PA Workshop 6-2, UCSD.

6. J.H.K. Vuolevi, T. Rahkonen, J.P.A. Manninen, Measurement Technique for Characterizing Memory Effects in RF Power Amplifier, IEEE Trans. MTT, vol. 49, No. 8, pp. 1383-1389, August 2001.

7. N.L. Wang, W. Ma, S Xu, e. Camargo, XP Sun, P. Hu, Z. Tang, H.F. Chau, A. Chen, CP Lee, 28 V High-linearity and Rugged InGaP/GaAs Power HBT, 2006 IEEE MTT-S International Microwave Symposium Digest, WE4-B, June 2006.

8. M. Rickelt, H-M Rein, E. Rose, Influence of Impact-ionization-induced Instabilities on the Maximum Usable Output Voltage of Si-Bipolar Transistors, IEEE Trans. ED, vol. 48, No. 4, pp. 774-779, April 2001.

9. D.E. Dalson et al, CW measurement of HBT Thermal Resistance,IEEE Trans. Elec. Dev. 39-10, pp. 2235-2239, 1992.

10. K. Goverdhanam, W. Dai, M. Frei, D. Farrell, J. Bude, H. Safar, M. Mastrapasqua, T. Bambridge, Distributed Effects in High- power RF LDMOS Transistors, 2005 IEEE MTT-S International Microwave Symposum Digest, pp. 455-458, June 2005.

11. J. C. Pedro and N. B. Carvalho, Intermodulation Distortion in Microwave and Wireless Circuits, Artech House 2003.

12. H.F. Chau et al, Wafer Level Reliability Tests of InGaP HBTs Using High Current Stress, 2002 IEEE GaAs Mantech Digest.

Nan-Lei Larry Wang received his BSEE from National Taiwan University, and MSEE and PhD from UC Berkeley. He has more than 20 years industry experience on RF, microwave, millimeter-wave IC and cellular phone RF transceiver design, which includes work at Raytheon Research Division, Rockwell International Science Center and Denso's cellular phone design center. He co-founded EiC Corp., which pioneered the development of high-reliability InGaP/GaAs HBT for wireless infrastructure base station and handset power amplifiers. In 2004, the business was merged into WJ Communications, where he is the vice president of Advanced Power Devices.

---