Broadband suppression of quantum noise below the Standard Quantum Limit (SQL) becomes a top-priority problem for the future generation of large-scale terrestrial detectors of gravitational waves, as the interferometers of the Advanced LIGO project, predesigned to be quantum-noise-limited in the almost entire detection band, are phased in. To this end, among various proposed methods of quantum noise suppression or signal amplification, the most elaborate approach implies a so-called *xylophone* configuration of two Michelson interferometers, each optimised for its own frequency band, with a combined broadband sensitivity well below the SQL. Albeit ingenious, it is a rather costly solution. We demonstrate that changing the optical scheme to a Sagnac interferometer with weak detuned signal recycling and frequency dependent input squeezing can do almost as good a job, as the xylophone for significantly lower spend. We also show that the Sagnac interferometer is more robust to optical loss in filter cavity, used for frequency dependent squeezed vacuum injection, than an analogous Michelson interferometer, thereby reducing building cost even more.