No abstract

No abstract

This paper presents an analysis of Error VectorMagnitude(EVM) performance of power amplifier using the amplitude/phase distortion coefficients derived from probability density function(PDF) of OFDM input signal. Amplitude coefficient values are calculated from the amplitude/phase distortion coefficients. The amplitude/phase distortion coefficient can be calculated from summation of the measured AM/AM(amplitude distortion) and AM/PM(phase distortion) data considered the probability of an input OFDM signal level. A complete analysis of the EVM model for output OFDM symbol with variable envelop through power amplifier can be derived from amplitude/phase distortion coefficients. To investigate, we use 1024(2 10 ) IFFT -OFDM signal based on 802.16e mobile system, 9.7dB peak to average power ratio (PAPR) at 0.01% CCDF and 10MHz overall occupation bandwidth, and the fabricated HBT MMIC power amplifier. The error ratio between the predicted EVM and measured EVM is as small as 1%.

Optical frequency division multiplexing (OFDM) has frequenctly been tried as one of means to increase a transmission capacity beyond 20GHz. Equi-spaced channel allocation has some advantages for network applications. Frequency slots for equispaced channel locking are generally produced by using either a fiber Fabry-Perot (FFP) resonator or a waveguide ring resonator[l] [2]. However, the change of the effective length in a resonator causes transmission peak frequency variation and, thereby, makes it difficult to identify each channel from system to system. Therefore, stability and synchronization of a resonator based on a certain frequency reference are needed. Recently, frequency synchronization using an angled etalon filter was reported [3].In this paper, we demonstrate a stabilized and synchronized channel meultiplexing scheme. The experimental setup is shown in Fig. 1. We used a tunable FFP filter with the free spectral range (FSR) of -1OOGHz as a reference resonator for equi-spaced channel allocation. For generation of absolute frequency reference, a I3C2H2 cell which has been known to have an absorption line (frequency fluctuation is less than 5MHz)[4] near the zero chromatic dispersion in the dispersion shifted fiber was selected. Two absolute frequency references to synchronize the FFP filter were obtained by individually locking two DFB-LDs to two absorption lines far away from each other. All DFB-LDs were dithered with sine waves (about a few KHz) for locking. Multiplexed output and absolute frequency references are monitored with an optical spectrum analyzer. Also, the mismatch between the second absortion line and the resonance at the corresponding transmission peak of the filter is measured by monitoring a beat frequency between both frequencies with a microwave spectrum analyzer. Each linewidth of the five lasers is around 2OMHz and the threshold current is about 20mA. The rate of frequency change at dc, which is measured by a scanning Fabry-Perot interferometer, is -330MHz/mA and -11GHz/K for the DFB-LDs, respectively. All lasers are thermally stabilized within a few mK before they are individually locked to the resonances of the FFP filter. Sine waves of slight different frequencies increasing from 1 KHZ with 100 Hz interval are applied to the lasers for the purpose of frequency locking. In order to set the FSR of the tunable FFP filter (finesse = l50), two absorption lines, 1,545.06nm and l,549.06nm7 from the 13C2H2 cell are selected, and two DFB-LDs are locked to those two absorption lines. The FSR is finally tuned to the frequency difference between two absolute frequency references divided by N (= 5, integer). In case that we try to put the first channel very close to the first absorption line, maximum deviation is occurred between the sixth channel and the second 376