Laboratory test report
Under Newport Test Labs’ standard testing conditions, Spec RSA-717RX did not meet its claimed power output into either 8Ω or 4Ω test loads. (The lab didn’t test the third load specified by Spec, of 6Ω). Into standard 8Ω loads, the Spec delivered only 40-watts per channel, both channels driven, at 1kHz, so overall output was around 10-watts (0.98dB) lower than Spec’s specification of 50-watts per channel.
Driven into 4Ω loads, for which Spec claims an output of 100-watts per channel, our test sample managed only 77-watts both channels driven at 1kHz, slightly more (78-watts) when driving a 10kHz load and only 64-watts when a 20Hz test signal was employed. All these figures fall well short of Spec’s specification of 100-watts per channel for this load.
When tested with 2Ω loads (for which the RSA-717EX is not rated) the amplifier self-protected at low frequencies, yet delivered 88-watts when a 10kHz test signal was used. This would appear to be a quirk of the particular circuit protection system employed by Spec.
The frequency response when driving a standard non-inductive resistive laboratory load was very flat at low and midrange frequencies, but started rising at 4kHz to be +1dB at 20kHz, after which it fell to be 1dB down at 58kHz and 3dB down at 70kHz. As is common with some Class-D designs, the response into a load that simulates that of a loudspeaker was not as flat, ending up +1.5dB at 20kHz and showing minor deviations from ‘flat’ at 2kHz (+0.3dB) and 80Hz (+0.1dB). This suggested to me that the frequency response of the Spec RSA-717EX will vary depending on the impedance of the particular speakers you connect to it—again, a not-unusual scenario with older Class-D designs. (Most of the newer implementations of Class-D designs are load-independent.)
Channel separation was adequate, at around 59dB at low and midrange frequencies, and 52dB at 20kHz, as was channel balance, which Newport Test Labs measured as being 0.1dB at 1kHz. Interchannel phase was good at low and mid-frequencies, but the figure of –25.4° at 20kHz was a degree of magnitude higher than I am used to seeing.
Total harmonic distortion (THD) was low, with Newport Test Labs measuring overall results of 0.04% at an output of one watt, and 0.02% at 40-watts, but at an output of one watt the spread of distortion components was quite unusual, as can be seen in Graph 1 (into an 8Ω load) and Graph 2 (into a 4Ω) load. You can see that the ‘good-sounding’ second-order, fourth-order and sixth-order harmonics are relatively low down, while the odd-order harmonics (third, fifth and seventh-order) are higher. However since the highest is at –90dB (0.0031%) at 8Ω, and at –85dB (0.0056%) at 4Ω, the distortion would not be audible. Also interesting is the slew of seemingly unrelated higher-order distortion components between 10kHz and 20kHz. However, once again because all are more than 90dB down (0.0031%) at 8Ω and nearly 100dB down (0.001%) at 4Ω, there would be no audible effects.
The distortion spectrum was more typical at 40-watts into 8Ω (Graph 3) and 60-watts into 4Ω (Graph 4), but also fairly atypically high, with the first two distortion components at around –80dB (0.01%), the next two at around –90dB (0.0031%). With an 8Ω load quite a few odd-order high-frequency distortion components are visible on Graph 3, but again all are more than 90dB down. When driving 4Ω loads, the odd-order components still dominated, but differences were less and the levels lower. It does point, however, to the amplifier’s performance being more dependent on the load it’s driving than is usual.
Signal-to-noise ratios were quite good, with Newport Test Labs reporting figures of 66dB unweighted and 81dB A-weighted referred to an output of one watt, and 83dB unweighted and 97dB A-weighted referred to an output of 40-watts. This would seem to indicate a fair amount of out-of-band electrical noise, which is hardly surprising given that the RSA-717EX has both a switch mode power supply and a Class-D output stage.
Square wave performance at 100Hz showed the tilt expected from the low-frequency roll-off noted during the frequency response
testing but none of the bending that would otherwise suggest phase issues. The ‘spike’ is indicative of a rise in the high-frequency response: it’s also visible in the other square wave responses at 1kHz and 10kHz. Also visible on the other traces is high-frequency switching noise that’s typical of a Class-D design. This noise is inaudible, being well above the high-frequency limit of human hearing. When loaded down with a highly capacitative load, the Spec RSA-717EX proved to be stable. Power consumption was relatively high for a Class-D design, at around 12-watts during normal operation, increasing to just under 100-watts when the amplifier was operating at full power, but low compared to a conventional amplifier. Standby power draw met Australian standards, at 0.5-watts.
Readers should note that the results mentioned in the report, tabulated in performance charts and/or displayed using graphs and/or photographs should be construed as applying onlyto the specific sample tested.