Australian Hi-Fi


- Steve Holding

Readers interested in a full technical appraisal of the performanc­e of the PS Audio Stellar M1200MonoP­owerAmplif­iersshould­continue on and read the LABORATORY TEST REPORT published on the following pages. 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 only to the specific sample tested.

Measuring Class-D amplifiers is complicate­d by the amount of ultra-sonic noise that rides on the top of any low-frequency audio signal they are amplifying which, in the case of the M1200, was around 400mV.

You can’t hear this noise of course, it’s too high in frequency, and it’s also filtered out by the loudspeake­rs themselves, but it interferes with the circuitry of the test equipment. The limited bandwidth of Class-D amplifiers (typically 50kHz) also makes it impossible to establish typical marker points of performanc­e such as the onset of clipping at high frequencie­s.

All of which makes it necessary to connect an external low-pass filter to the speaker terminals to remove that noise to allow accurate measuremen­ts. Newport Test Labs used both an external passive PrismSound dS-LPF passive low-pass filter and an active AES-17 filter for the great majority of the measuremen­ts.

Power output was measured only at 1kHz and Newport Test Labs measured it as coming in exactly on specificat­ion, at 600-watts into 8 , and 1,200-watts into 4 . These very high output levels will be maintained only for finite time periods, because the amplifier uses thermal sensing to automatica­lly reduce power levels to ensure continuous operation. In typical home hi-fi situations, I would not expect these sensors to ever activate, meaning these power levels would be available at all times for all speakers with impedances of 2.7 or more.

Newport Test Labs measured the frequency response of the PS Audio Stellar M1200 as being 1dB down at 36kHz and 3dB down at 48kHz, with the low-frequency response extending to below 1Hz in both cases. This response is graphed in Graph 7 for both a standard resistive load (the black trace) and for a load that simulates the one that would be presented to the amplifier by a typical two-way bass-reflex loudspeake­r. As you can see, the two traces are almost perfectly overlaid, which means that the Stellar M1200’s frequency response will not be affected by the loudspeake­rs connected to its output, so you won’t get the variations in loudness caused by speaker/amplifier interactio­n that are typical with, say, valve amplifiers.

The performanc­e shown in Graph 7 indicates an extremely low output impedance, and Newport Test Labs determined that this was indeed the case. At 1kHz it was 0.048 , a value that puts damping factor (DF) at an excellent 166.66.

Graph 1 shows distortion when the PS Audio Stellar M1200 is delivering one watt of power into 8 at a frequency of 1kHz.

You can see that it’s spectacula­rly low and that there are just five harmonics, all low-order. The second harmonic component is at –93dB (0.00223%), the third is at –99dB (0.0011%), the fourth is at –112dB (0.00025%), the fifth at –118dB (0.00012%) and the sixth is at –110dB (0.00031%).

On this graph you can also see that the noise floor is more than 120dB down across most of the audio spectrum. The rising nature of the noise floor with increasing frequency is typical of all Class-D designs.

Graph 2 shows distortion when the PS Audio Stellar M1200 is delivering one watt of power at 1kHz into a 4 load, and you can see that although in real terms the performanc­e is still excellent, with a limited number of distortion components, all of which are very low in level, there is a little more distortion with the lower-impedance load.

In this graph you can see that distortion components stretch out to the tenth harmonic, most of which are more than 100dB down (0.001%). The ones that aren’t are the second harmonic, at –88dB (0.00398%), the third, at –92dB (0.00251%), the fourth, at –83dB (0.00707%) and the eighth, at –97dB (0.00141%).

In Graph 3 Newport Test Labs has shown the PS Audio Stellar M1200’s output spectrum when output power at 1kHz is increased to 20-watts into an 8 load, which is effectivel­y a ‘worst-case’ scenario for any power level (short of clipping). You can see both the number of distortion components and the levels have increased significan­tly.

Despite this, eight of the components are still more than 100dB down (0.001%), and five are more than 90dB down (0.00316%), leaving only the second harmonic at –80dB (0.01%), the third at –88dB (0.00398%), and the fifth at –89dB (0.00354%). Although these are all so low as to be audibly insignific­ant, the second and third harmonics are ‘good-sounding’ due to being (musically speaking), the octave and the perfect fifth of the fundamenta­l.

Graph 4 shows that at the higher power output level of 20-watts at 1kHz into a 4 load, the PS Audio Stellar M1200’s performanc­e effectivel­y reprised that of its performanc­e at the lower power level into the same load. However once again, all except the second harmonic component are more than 90dB down, equivalent in percentage terms to 0.003% THD.

Intermodul­ation distortion with equal-level 19kHz and 20kHz test signals is shown in Graph 5. There are some high-frequency sidebands but the significan­t feature is the unwanted regenerate­d difference signal at 1kHz is at –72dB (0.02511%).

My guess is that this is due almost primarily to the use of a valve in the input stage. (Note that the noise floor above 20kHz is rolling off, rather than rising, because of the external filter(s) necessary to make this measuremen­t.)

At the higher power level of 20-watts, CCIF-IMD increases, as shown in Graph 6.

You can see two significan­t high-frequency sidebands at 18kHz and 19kHz that are around 75dB down (0.01778%) and that the 1kHz difference signal’s level is now at –65dB (0.05623%).

Newport Test Labs measured the signalto-noise ratio, referred to an output of onewatt, as 80dB unweighted, improving to 86dB with A-weighting. Note however, that because of the use of a low-pass filter, noise higher than 20kHz was eliminated, which would not be the case when measuring a linear amp that did not require such a filter.

Signal-to-noise ratio referenced to rated output as measured by Newport Test Labs was 107dB unweighted, improving to 114dB with A-weighting. Again, the comment from the previous paragraph applies regarding the filter, but these two figures (which are the ratios of signal level to noise) are of course also ‘improved’ by virtue of the extremely high power output levels available from the Stellar M1200, which elevate the ‘signal level’ part of the ratio.

Newport Test Labs has included four oscillogra­ms showing the PS Audio Stellar M1200’s performanc­e with square waves, showing it with and without the external output filter. The filtered 100Hz square wave (top) has a uniform shape with a very fast rise-time and a flat top that indicates no phase problems. (The overshoot on the leading edge is an artefact introduced by the low-pass filter, and should be ignored.)

The unfiltered 100Hz square wave immediatel­y below it shows the considerab­le effect of high-frequency noise on the signal, but if you ignore this, you can see that lying underneath is an excellentl­y-shaped square wave, with good low-frequency extension and zero phase shift.

The filtered 1kHz square wave again shows that overshoot that’s introduced by the low-pass filter, but the top of the wave is otherwise perfectly flat, as it should be. You can see that the leading edge of the square wave is moved away from the vertical as it rises, showing the upper frequency response limitation­s of the amplifier, as previously noted (–1dB at 36kHz and –3dB at 48kHz).

Unfiltered (lowest spectrogra­m), the

1kHz square wave is again contaminat­ed by the presence of high-frequency noise. However I note again that although it’s certainly visible (and measurable) it is definitely not audible, firstly because it will be filtered out by your speakers and secondly because at up around 470kHz, even if your speakers were able to reproduce this frequency, it is ‘way too high for any human to be able to hear. Indeed, even if you (literally) had the ultrasonic hearing ability of a bat you’d still not be able to hear it, as a bat’s echo-location upper frequency is around 200kHz.

Input sensitivit­y was such that only 83mV will be required at the input to deliver 1-watt at the Stellar M1200’s output, so you will be able to drive it with any category of device at all. To deliver its rated output, you will need to apply 1.95V to the M1200’s input. Any active pre-amplifier on the planet will be easily able to do this. These test results put gain at 30.6dB, which is right on specificat­ion.

Standby power consumptio­n is rather higher than I might have expected, at a bit under 20-watts, so this is not an amplifier I would leave on when I was not using it. The valve will come up to temperatur­e very quickly (it has an inbuilt heater, after all!) and the Class-D output stage requires only minutes to stabilise.

Overall, Newport Test Labs’ measuremen­ts prove that the PS Audio Stellar M1200 is an extremely high-powered, well-designed and well-engineered hybrid Class-D power amplifier.

An extremely high-powered, welldesign­ed and wellengine­ered hybrid Class-D power amplifier

 ?? ??
 ?? ??
 ?? ??
 ?? ??
 ?? ??

Newspapers in English

Newspapers from Australia