Frame rate in ultrasound imaging can be increased by simultaneous transmission of multiple beams using coded waveforms. However, the achievable degree of orthogonality among coded waveforms is limited in ultrasound, and the image quality degrades unacceptably due to interbeam interference. In this paper, an alternative combined time-space coding approach is undertaken. In the new method all transducer elements are excited with short pulses and the high time-bandwidth (TB) product waveforms are generated acoustically. Each element transmits a short pulse spherical wave with a constant transmit delay from element to element, long enough to assure no pulse overlapping for all depths in the image. Frequency shift keying is used for per element coding. The received signals from a point scatterer are staggered pulse trains which are beamformed for all beam directions and further processed with a bank of matched filters (one for each beam direction). Filtering compresses the pulse train to a single pulse at the scatterer position with a number of spike axial sidelobes. Cancellation of the ambiguity spikes is done by applying additional phase modulation from one emission to the next and summing every two successive images. Simulation results presented for QLFM and Costas spatial encoding schemes show that the proposed method can yield images with range sidelobes down to -45 dB using only two emissions.