Vertically, the Revel's tweeter is a very high 47" from the floor, which at first worried me: Was I doing the right thing by measuring the response on the tweeter axis? (In the three-dimensional puzzle you have to solve when doing a set of speaker measurements, it helps to choose an easily repeatable axis for the microphone. In this respect, the tweeter is the best choice.) But as fig.7 reveals, the Salon's balance hardly changes over a wide window, thanks to its use of high-order crossover filters. As long as the listener's ears are between 36" and 50" from the floor, he or she will hear what the speaker is capable of doing.

Fig.7 Revel Salon, vertical response family at 50", from back to front: responses 15 degrees-5 degrees above HF axis; reference response; responses 5 degrees-15 degrees below HF axis.

However, the high-order crossover means that the Salon cannot be time-coherent, as shown by its step response on the tweeter axis (fig.8). The tweeter output arrives first at the microphone, followed by the outputs of the midrange unit, then the upper woofer, then the three lower woofers. It is arguable whether time coherence is necessary or not. Certainly, LG's very positive reaction to the Revel's resolution of detail and its imaging accuracy was not negatively affected by the speaker's time-domain behavior.

Fig.8 Revel Salon, step response on tweeter axis at 50" (5ms time window, 30kHz bandwidth).

The Salon's waterfall plot (fig.9) was not as clean as I had expected. I suspect that some of the HF hash present in the floor of this graph is due to early reflections of the tweeter's output from the tops of the speaker's side cheeks. There is a hint of delayed energy just below the on-axis notch in the presence region, but this is probably too subtle to introduce any coloration.

Fig.9 Revel Salon, cumulative spectral-decay plot at 50" (0.15ms risetime).

Another finely engineered loudspeaker design from Kevin Voecks and his team.—John Atkinson