Techniques are provided for phase tracking in wireless communications using frames with some portions that use relatively long symbol durations. One technique includes generating a frame having a plurality of orthogonal training field sequences. Each orthogonal training field sequence is assigned to a different spatial stream. The frame is output for transmission as a single user (SU) transmission, a downlink multi-user (MU) multiple input multiple output (MIMO) transmission, or an orthogonal frequency division multiple access (OFDMA) transmission. After obtaining the frame, channel estimation and/or phase tracking may be performed based on the orthogonal training field sequences.