Skip to Main Content
In digital radio applications, the error vector magnitude (EVM) is a figure of merit that characterizes modulation accuracy. The EVM highlights modulation impairments caused by either the in-phase/quadrature (I/Q) baseband transmit chain, filters, modulator, or amplifier linearity. EVM testing in production implies the use of high-cost test instrumentation in automated test equipment (ATE), which requires a complex I/Q baseband modulator supporting digital modulation and a built-in high-performance receiver. In this paper, an alternate EVM measurement technique is proposed that not only eliminates digital signal processing but also relies on standard continuous-wave (CW) tone measurements to evaluate the EVM performance of the device under test (DUT). The presented method comes at no extra test time cost since the EVM is computed from a series of CW tests, which are traditionally tested to guarantee the RF modulator yield in mass production. Assuming each EVM contributor is uncorrelated, the method consists of using simple sine-wave/cosine-wave input I/Q stimuli applied to the DUT to quantify each impairment and then compute each EVM contribution to finally derive the resulting composite EVM. The presented ATE EVM computation results are in excellent agreement with the laboratory EVM measurements using ATE conformance test equipment. The technique is applied to a release'99 wideband code-division multiple-access (W-CDMA) band I [(Universal Mobile Telecommunications System (UMTS) 2100] direct-conversion RF integrated circuit (IC) modulator.