Flying Mole CA-S10 integrated amplifier Measurements

Sidebar 3: Measurements

Before performing any measurements on an amplifier, I precondition it by running it for 60 minutes at one-third its specified power into 8 ohms—the worst case for an amplifier with a class-B output stage. With a class-D amplifier such as the Flying Mole CA-S10, however, the output stage is so efficient that almost all the power taken from the wall is supplied to the load at all levels, rather than being dissipated in the output devices. At the end of the hour, the CA-S10's case was faintly warm. This is indeed an environmentally friendly amplifier.

The Flying Mole delivered a maximum voltage gain of 45.33dB into 8 ohms. The maximum gain from the preamp outputs was 17.2dB, meaning that the power-amp section was responsible for around 28dB, which is typical of power amplifiers in general. The amplifier preserved absolute polarity; ie, was non-inverting. The input impedance didn't change with the setting of the volume control and was a uniform 10k ohms across the audioband—this will be a little on the low side for tubed source components. The output impedance from the Preamp Out jacks was a very high 3.8k ohms. From the speaker terminals, the output impedance was a reasonably low 0.15 ohm at low and midrange frequencies. However, it rose to >1.5 ohms at 20kHz, due to the need for an internal low-pass filter to be used on the output to reduce the amount of RF switching noise present.

Though it will always roll off ultrasonic frequencies, this filter will interact with the load to modify the amplifier's frequency response at the top of the audioband. The solid traces in fig.1 show how the response changes into loads ranging from 16 ohms (top) to 2 ohms (bottom). The Flying Mole's output peaks by almost 2dB at 40kHz into 16 ohms, but is down by almost 5dB at 20kHz into 2 ohms. However, with its low audioband output impedance, the modification of the CA-S10's response into our standard simulated loudspeaker (fig.1, top dotted trace) remains within +0.8/–1.1dB limits.

Fig.1 Flying Mole CA-S10, frequency response at 2.83V into (from top to bottom at 2kHz): simulated loudspeaker load, 16, 8, 4, 2 ohms (1dB/vertical div., right channel dashed).

This filter-load interaction also affects the CA-S10's behavior with squarewaves: fig.2 shows the amplifier's reproduction of a 1kHz squarewave into 8 ohms, which overshoots a little, while fig.3 shows the same squarewave into 4 ohms. The leading edges are now significantly slowed down. As the Avantgarde Uno 3.0 speaker used by Robert Deutsch has an impedance of 10–16 ohms above 10kHz (see, it is possible that this behavior contributed to his feeling that the amplifier's sound suffered from some leading-edge emphasis. The Paradigm Reference Studio/20's impedance also rises above 8 ohms in the high treble, which will also give rise to some transient overshoot.

Fig.2 Flying Mole CA-S10, small-signal 10kHz squarewave into 8 ohms.

Fig.3 Flying Mole CA-S10, small-signal 10kHz squarewave into 4 ohms.

Channel separation was excellent, however, at >90dB in the midrange and below, though it did decrease to a still more-than-adequate 60dB at 40kHz (fig.4). Measuring the CA-S10's signal/noise ratio was tricky because of the large amount of ultrasonic switching noise present on its output: 139.5mV with a frequency of 365Hz with no audio signal being amplified. The wideband, unweighted ratio was therefore poor, at 27.1dB ref. 2.83V into 8 ohms (1W), and even reducing the measurement bandwidth to the audioband (22Hz–22kHz) improved the ratio to just 51.1dB, while the A-weighted ratio was a modest 59dB. This is not an amplifier that will be optimal for use with high-sensitivity speakers.

Fig.4 Flying Mole CA-S10, channel separation (20dB/vertical div.).

The presence of the switching noise made assessing the Flying Mole's distortion problematic because of the possibility of it driving my Audio Precision System One's input stage into slew-rate limiting. The active, sixth-order, low-pass filter I use to measure class-D amplifiers won't handle very large voltage swings, so I examined the CA-S10's maximum power delivery without a filter, which should be borne in mind when looking at fig.5. This graph plots the percentage of distortion and noise present in a 1kHz tone (vertical scale) in the Flying Mole's output as the output power increases (horizontal scale). Below the discontinuity, the traces actually show the contribution of the switching noise. But above that point, true distortion raises its head. The CA-S10's power rating of 100Wpc into 8 ohms (20dBW) is met at just under 3% THD, as is its rating of 160Wpc into 4 ohms (19dBW). Into 2 ohms, the amplifier reached 3% THD at 90Wpc (13.5dBW), though it is fair to note that the CA-S10 is not rated into this load.

Fig.5 Flying Mole CA-S10, distortion (%)vs 1kHz continuous output power into (from bottom to top at 100W): 2, 4, 8 ohms.

To examine how the small-signal THD+noise percentage changed with frequency, I inserted an active 30kHz low-pass filter to minimize the effect of switching noise on the measurement. The results are shown in fig.6, which is plotted only up to 10kHz because of the effect of the low-pass filter. Unusually, the distortion at low and midrange frequencies decreases with decreasing load impedance, the opposite of what happens with a conventional amplifier. But at higher frequencies the distortion rises into all loads, the amplifier apparently becoming less linear.

Fig.6 Flying Mole CA-S10, THD+N (%)vs frequency at 4V into (from bottom to top): 2, 4, 8, 16 ohms, with series sixth-order, 30kHz low-pass filter.

The bottom trace in fig.7 shows the residual waveform after the 1kHz sinewave (top trace) has been notched out. All it shows is the ultrasonic switching noise. Repeating this test with the 30kHz low-pass filter in circuit gave the result shown in fig.8. Actual harmonic distortion appears to be very low, but a brief burst of ringing can be seen at every zero-crossing point of the waveform. This is unusual, and gets worse with increasing frequency. The Flying Mole is very linear at low frequencies, as can be seen in fig.9, which shows the spectrum of the amplifier's output while it reproduces a 50Hz tone at 39W into 4 ohms. The second harmonic is the highest in level, but it lies at just –96dB (0.0015%). But when the signal frequency is raised to 1kHz, not only does the second harmonic rise to –77dB (0.014%), there is now a regular series of higher-order harmonics present (fig.10). Though these are all well below –80dB (0.01%), their presence is disturbing.

Fig.7 Flying Mole CA-S10, 1kHz waveform at 2W into 8 ohms (top), 1.14% THD+N; distortion and noise waveform with fundamental notched out (bottom, not to scale).

Fig.8 Flying Mole CA-S10, 1kHz waveform at 2W into 4 ohms (top), 0.137% THD+N; distortion and noise waveform with fundamental notched out and with series sixth-order, 30kHz low-pass filter (bottom, not to scale).

Fig.9 Flying Mole CA-S10, spectrum of 50Hz sinewave, DC–1kHz, at 39W into 4 ohms (linear frequency scale).

Fig.10 Flying Mole CA-S10, spectrum of 1kHz sinewave, DC–10kHz, at 39W into 4 ohms (linear frequency scale).

The poor high-frequency linearity contributes to a relatively high level of high-order intermodulation products even at a relatively low power level (fig.11), though the second-order difference product at 1kHz is low, at –87dB (0.007%).

Fig.11 Flying Mole CA-S10, HF intermodulation spectrum, DC–24kHz, 19+20kHz at 15.5W peak into 8 ohms (linear frequency scale).

Although its measured performance at low and midrange frequencies is excellent, the Flying Mole CA-S10 did less well at high frequencies. And while the class-D PS Audio GCC-100 with which RD compared the CA-S10 also had some linearity problems at high frequencies, its output signal was overall very much cleaner than the Flying Mole's. I am not surprised, therefore, that RD ultimately found he could not recommend the CA-S10.—John Atkinson

Flying Mole Electronics Corporation
3592 Rosemead Blvd., #509
Rosemead, CA 91770
(626) 374-7734
Share | |

Enter your username.
Enter the password that accompanies your username.