Download this article in .PDF format
This file type includes high resolution graphics and schematics when applicable.

Figure 2 shows a group of 2- and 3-b representations of a digital sequence. Larger numbers of bit grouping (equivalent to higher modulation levels) have more amplitude levels. Thus, for fixed average power, higher-modulation-level signals will have relatively larger peak levels or higher peak-to-average-ratio (PAR) values. Higher PAR requires better linearity to maintain a required bit error rate (BER), signal integrity, and other performance parameters.

QAM Is Rising: 1024QAM And Beyond, Fig. 2

As the number of amplitude levels is increased, the signals with larger modulation levels are more susceptible to noise. Thus, a system with higher SNR or better noise floor is desired and an oscillator with superior phase noise is required to achieve low system noise levels and minimum signal degradation.

Given the symbol rate, fs, of a QAM signal, the approximate RF bandwidth (BW) can be estimated by applying Eq. 2:

BW = fl(1 + α)   (2)

where:

α = the filter rolloff factor.

Generally, a raised-cosine filter is used for pulse-shaping in digital systems as it provides lower intersymbol interference (ISI) and the rolloff factor of the raised-cosine filter determines the excess bandwidth the filter occupies beyond the Nyquist bandwidth (fs/2).

Measuring the performance of link quality in the presence of impairments such as phase noise, system thermal noise, and clock jitter can be described in terms of error-vector-magnitude (EVM) measurements. In the case of direct-conversion modulators and demodulators, additional impairments exist such as local-oscillator (LO) leakage in transmitters, DC offset in receivers, and in-phase/quadrature (I/Q) magnitude and phase imbalance. EVM measurements evaluate the difference between actual and ideal symbol locations of a digitally modulated waveform.

EVM is related to the system’s SNR. An ideal system with zero noise, no nonlinearity distortion, no frequency error, and no I/Q imbalance will have an excellent (theoretically infinite) SNR and zero EVM. Degradation in EVM is due to longer error distance between the referenced and measured symbol locations, which is due to system noise and distortion.

QAM Is Rising: 1024QAM And Beyond, Fig. 3a

QAM Is Rising: 1024QAM And Beyond, Fig. 3b

QAM Is Rising: 1024QAM And Beyond, Fig. 3c

Texas Instruments has developed a wide variety of highly linear and high-performance signal-chain devices suitable for large-bandwidth, high-modulation-rate requirements. For example, model TRF3720, is a fully integrated I/Q modulator and phase-locked-loop (PLL) voltage-controlled oscillator (VCO) which exhibits a 1024QAM EVM of 0.589%. Figure 3(a) shows an evaluation setup for the device, where a model TSW3100 pattern generator is used as the baseband signal source.

The digital-to-analog converter (DAC) in this test setup is a model DAC34H84, which is cascaded with the TRF3720 device under test. The evaluation system was locked to a low-noise 61.44-MHz reference oscillator, with a model TSW3003 evaluation board used as a buffer to divide the 61.44-MHz reference clock to 30.72 MHz to lock the instruments. Figure 3(b) shows the evaluation setup with the modulator, DAC, pattern generator, and signal splitter. Figure 3(c) shows the 56-MHz 1024QAM signal EVM of 0.589% at the modulated output centered at 2.21 GHz.

Download this article in .PDF format
This file type includes high resolution graphics and schematics when applicable.