LABORATORY TEST REPORT

Graph 1 shows the distortion spectra in the Cambridge Audio DacMagic 200M’s analogue output when it is delivering maximum output voltage (3.9-volts) with a 1kHz digital signal at maximum level (0dB). This is a little unrealistic, because it would never happen outside an audio test laboratory environment, but you can see the performance of the DacMagic was excellent nonetheless.

There is a third harmonic component at –102dB (0.00079%), then a string of odd-order harmonics across the spectrum, but each one of these is around or lower than –110dB (0.00031%).

Graph 2 shows a more realistic test for distortion, with the same 1kHz test signal, but at a digital level of –10dB. This is about the maximum digital level that would be expected, as it leaves a comfortable 10dB of headroom for peaks. You can see that this time there’s only that single third harmonic distortion component in the DacMagic 200M’s output, and its level has dropped to –110dB (0.00031%) where it would be completely and totally inaudible. There is a small ninth-order harmonic visible, but it’s around 135dB down, or (0.00001%).

Also note in Graph 2 that the noise floor is sitting down at –140dB right across the audio spectrum, and although there appears to be some mains hum creeping in, it’s down at –128dB (0.00003%).

Newport Test Labs has in Graph 3 reduced the level of the 1kHz test signal to –20dB and you can see there’s been a corresponding reduction in the level of the single distortion (HDL3) component that is visible on the graph to –118dB (0.00012%). The tiny ninth-order component that was present at –135dB has disappeared below the noise floor. As for that noise floor, it’s a little bit noisier, but it’s still down at –140dB.

Graph 4 shows the DacMagic 200M’s output spectrum when it’s reproducing an undithered 1kHz digital signal at –60dB. The ‘grass’ on the noise floor is due to quantisation errors but it’s all still more than 120dB down. This test is only of academic interest, because all digital music is dithered in order to eliminate quantisation errors, the effect of which is shown in Graphs 5 and 6.

Graph 5 shows a 1kHz test signal at a level of –90dB that has not been dithered. You can see there are odd-order harmonic distortion components spread across the audio spectrum. The first four of these are between –110dB and –120dB down (0.00031% to 0.0001%) with all the higher-order harmonics more than 120dB down. You can see each harmonic is accompanied by quantisation error signals. Now look at Graph 6.

Graph 6 shows exactly the same test signal that was used for Graph 5, at exactly the same output level from the DacMagic 200M, but this time the test signal was dithered. As you can see, the dithering has removed all the harmonic distortion components, and also corrected all the quantisation errors.

The only penalty for doing this is a slight increase in noise, as you can see by comparing the bottoms of both graphs. However, because the noise floor is down at –140dB, this is of no concern.

Newport Test Labs has shown intermodulation distortion in Graph 7 using two test signals, one at 19kHz and the other at 20kHz, using the ‘Fast’ filter setting of the DacMagic 200M. You can see there are only two sidebands, one at 18kHz and the other at 21kHz, and both are around 118dB down (0.00012%). Importantly, there is no difference signal down at 1kHz. There are some spurious signals up around 40kHz, but these are too high in frequency and too low in level to be of any concern.

Graph 8 again looks at intermodulation distortion, using the same test signal, but this time with the ‘Slow’ filter setting of the DacMagic 200M. You can see that the sidebands at 18kHz and 21kHz have reduced a little in level, but there are now ‘images’ of the test signal up at 25kHz and 28kHz due to the operation of the delta-sigma oversampling filter. It’s important to note that this would not happen with music, which never contains signal levels at these frequencies high enough to introduce the imaging.

The frequency response of the Cambridge Audio DacMagic 200M is shown in Graph 9 for all three settings of the filter, using 16-bit/44.1kHz digital signals, which of course have a high-frequency limit of 20kHz. (In fact Newport Test Labs used ‘RedBook’ standard digital signals for all the tests

it made on the DacMagic 200M, input via the coaxial input.) Note that the vertical scaling of the graph is extreme, so the graph lines immediately above and below the traces are at +0.5dB and –0.5dB, so that from 20Hz out to 10kHz, all three filters are flat to within better than ±0.05dB. This is outstandingly good performance.

Above 10kHz the three frequency responses diverge. The black trace shows the DacMagic 200M’s frequency response with the ‘Short Delay’ filter, and you can see that it extends right out to 20kHz, still within ±0.05dB. The same is true of the ‘Slow’ setting of the filter, though although it, too, is only 0.05dB down at 20kHz, it rolls off a little earlier than the ‘Short Delay’ filter to reach the same down-point.

The frequency response of the ‘Fast’ filter is shown as the green trace on Graph 9 and you can see that although it appears to roll off steeply above about 12kHz, the so-called ‘steepness’ is a function of the graph scaling, because the trace is just 1.7dB down at 20kHz, which would put the normalised frequency response of the DacMagic 200M with this filter in circuit at 20Hz to 20kHz ±0.85dB.

Graphs 10 and 11 show the effect of the ‘Fast’ and ‘Slow’ filters on the response above 20kHz (for a 44.1kHz/16bit test signal). Graph 10 shows the fast filter rolls the high-frequency response off so quickly that it’s around 80dB down at 24kHz.

Graph 11 shows the effect of the ‘Slow’ filter and you can see the response is only 10dB down at 24kHz and 15dB down at 29kHz and 50dB down at

34kHz.

Graph 12 shows the effect of the

‘Fast’ filter when the DacMagic200M is reproducing a 20kHz signal at maximum level (0dB).

As with the IMD tests, this is a signal that would never occur when playing music, it’s included merely to show the operational characteristics of the DAC and filter. You can see the 20kHz signal just to the left of the graph’s centre. There are no sub-harmonics further to the left, and only three sampling-related artefacts to the right, one more than 110dB down and the other two more than 120dB down. Graph 13 shows the output using the same test signal, but the DacMagic’s ‘Slow’ filter.

You can see there’s a sub-harmonic at around 16kHz that’s 120dB down.

To the right of the test signal is a mirror image at 24kHz. To the right of this are several sampling-related signals, all of which are more than 120dB down.

Graph 14 shows the output of the DacMagic 200M using the same test signal as in Graphs 12 and 13, but this time with the ‘Short Delay’ (Minimum Phase) filter. There are no subharmonics in the audio band, and only a vestigial mirror image at 25kHz that’s 104dB down. There are only two other sampling-related artefacts, both higher than 35kHz in frequency, one of which is at

–112dB and the other at –119dB.

Graph 15 shows the DacMagic 200M’s reproduction of a 1kHz square wave using the fast filter. The time-reversed ringing on the wave demonstrates that Cambridge Audio is using a standard delta-sigma oversampling DAC.

Graph 16 shows the DacMagic 200M’s reproduction of a single impulse. You can see that the typical pre-ringing introduced by the delta-sigma decoding and that the pulse is positive-going, which means the DacMagic 200M preserves phase, rather than inverting it, like some DACs.

Newport Test Labs’ testing revealed the Cambridge DacMagic 200M has superb channel separation, with the laboratory reporting figures of 151dB at 20Hz, 141dB at 1kHz and 116dB at 20kHz. Along with this the laboratory also reported excellent inter-channel phase characteristics — 0.01 degrees of error at 20Hz and 1kHz and j94ust 0.12 degrees at 20kHz. Channel balance was also excellent (0.1027dB at 1kHz).

Wideband signal-to-noise ratios were outstanding, as you’ve probably already guessed from the noise floors revealed in the graphs used to show distortion. Newport Test Labs measured it as 115dB unweighted, and 121dB A-weighted.

Rather unusually these days, the DacMagic 200M has de-emphasis correction built in, which means you’ll get accurate high-frequency response with not only modern digital recordings, but also with old ones made back in the 80s. The circuit was very accurate, too, as you can see from the error figures in the tabulated chart of test results.

Linearity error was close to perfect at all tested levels, and actually perfect at –60dB and –80.59dB, as you can see from the tabulated results. Again, this is outstanding performance.

Power consumption is a low 7.95-watts when the DacMagic 200M is operating, and 1.46 when it’s in Standby mode. I was a little intrigued by this, because the DacMagic 200M is powered by an external wall-mounted switch-mode power supply, so when you turn the DacMagic 200M to standby, the power supply is still on, hence the 1.46-watt consumption in Standby mode is almost entirely due to the power supply, not the DAC itself.

Overall, the Cambridge Audio DacMagic 200M returned stunningly good levels of performance in all the tests performed by Newport Test Labs, performance that is made all the more stunning because all these great results were returned with standard 16-bit/44.1kHz test signals.