Constellation Audio Performance Centaur Mono monoblock power amplifier Measurements

Sidebar 3: Measurements

Before performing any measurements, I ran one of the Performance Centaur Monos (serial no. 1101131901) for an hour at one-third its specified maximum power of 500W into 8 ohms, thermally the worst case for an amplifier with a class-B or -AB output stage. The amplifier was too heavy for me to lift it onto my test bench so it sat on the carpeted floor, with some clearance given by supporting it on Black Diamond cones. By the end of the hour, the top panel was warm, at 105°F (43.5°C), and the side-mounted heatsinks were warmer, at 114.5°F (45.8°C). With the amplifier cold, the THD+N at one-third power was 0.0329%. By the end of the hour, it had dropped to 0.0307%.

I performed a full set of measurements on serial number 1101131901 using Stereophile's loan sample of the top-of-the-line Audio Precision SYS2722 system (see and the January 2008 "As We See It") connected to the Centaur's balanced input. I repeated some of the tests using the unbalanced input, but did not test the amplifier via its Direct input, which is intended for use only with Constellation preamplifiers.

The voltage gain at 1kHz into 8 ohms was a little lower than average, at 25.8dB, but close to the specified 26dB. It was the same for both balanced and unbalanced drive, and both inputs preserved absolute polarity (ie, were non-inverting). The Centaur's input impedance is specified as a very high 100k ohms unbalanced and 200k ohms balanced, but my measurements were very much lower: 9.6k ohms unbalanced and 19k ohms balanced.

The output impedance was low for a bridged design, at 0.07 ohm including the cable at low and middle frequencies, rising slightly to just under 0.1 ohm at the top of the audioband. As a result, the modification of the Centaur's frequency response by the interaction between this source impedance and the impedance of our standard simulated loudspeaker (fig.1, gray trace) was less than ±0.1dB. However, this graph reveals that the Centaur's ultrasonic behavior depends on the load impedance. With the highest impedance I tried, 16 ohms (fig.1, green trace), the output rose by 0.8dB at 200kHz. The rise was 0.3dB into 8 ohms (blue), and into 4 ohms (magenta) and 2 ohms (red) the response increasingly rolled off above the audioband. Perhaps correlating with this rise at 200kHz, while the Centaur's reproduction of a 10kHz squarewave had very short risetimes (fig.2), there was also a very small amount of overshoot visible, with one damped cycle of ringing at a very high ultrasonic frequency. The 1kHz squarewave reproduction (not shown) was essentially perfect.


Fig.1 Constellation Performance Centaur, frequency response at 2.83V into: simulated loudspeaker load (gray), 16 ohms (green), 8 ohms (blue), 4 ohms (magenta), 2 ohms (red) (1dB/vertical div.).


Fig.2 Constellation Performance Centaur, small-signal 10kHz squarewave into 8 ohms.

Although MF noted that the Centaur was very quiet, the amplifier's wideband, unweighted signal/noise ratio (ref. 1W into 8 ohms), taken with the balanced input shorted, was slightly disappointing at 72.1dB, though it did improve to 81.2dB when A-weighted. Spectral analysis of the amplifier's low-frequency noise floor while it reproduced a 1kHz tone at 1W into 8 ohms (fig.3, blue trace) indicated that the residual full-wave rectified power-supply component at 120Hz lay at –83dB (0.007%), with the 240Hz harmonic at –87dB. The magnetically sourced component at 60Hz lay at –93dB, and all the other supply-related components lay at or below –95dB. I repeated the spectral analysis with the other sample of the Centaur, S/N 1101131902. The spectrum of this amplifier's noisefloor (fig.3, red trace) was very similar to that of the first sample (blue trace), but the odd harmonics of the 60Hz AC frequency were a little higher in level.


Fig.3 Constellation Performance Centaur, spectrum of 1kHz sinewave, DC–1kHz, at 1W into 8 ohms, samples 1101130901 (blue) and 1101131902 (red) (linear frequency scale).

Fig.4, which plots the percentage of THD+noise against output power into 8 ohms, indicates that the Centaur slightly exceeds its specified 500W into this load, clipping (defined as when the THD+N reaches 1%) at 520W (27.16dBW). The minimum THD+N is very low, at 0.0033%; the downward slope of the trace below 20W in this graph reveals that the distortion is actually below the noise floor at low powers. (The noise, being constant, becomes an increasing percentage of the signal level as the power is reduced.) The picture was similar into 4 ohms (fig.5), with the clipping power of 830W (26.18dBW) slightly greater than the specified 800W. The minimum THD+N was higher, however, at 0.0061% between 10 and 20W. The distortion was higher still into 2 ohms (fig.6), with the amplifier clipping at 1210W (24.8dBW), which is 0.8dB above the specified 1kW into this load. However, although I switched off the signal generator as soon as the Centaur clipped into 2 ohms, the amplifier wouldn't pass a signal after this test. Obviously, I had broken something. (The possibility of damage with this very stressful test is why I leave it until the end of testing.)


Fig.4 Constellation Performance Centaur, distortion (%) vs 1kHz continuous output power into 8 ohms.


Fig.5 Constellation Performance Centaur, distortion (%) vs 1kHz continuous output power into 4 ohms.


Fig.6 Constellation Performance Centaur, distortion (%) vs 1kHz continuous output power into 2 ohms.

Before this test, I plotted how the THD+N varied with frequency at a level, 18V, where I could be sure that I was looking at actual distortion rather than noise. The results are shown in fig.7. The THD rises slightly at the top of the audioband, but is very low into both 8 ohms (blue trace) and 4 ohms (magenta). The amplifier is less comfortable with 2 ohms (red), however. Again, to be sure I was looking at distortion rather than noise, I examined the waveform of the THD+N residue at a high level, 50W into 8 ohms (fig.8). The distortion at this level was 0.005% and was predominantly third harmonic in nature. Even at high power, 200W, into 4 ohms (fig.9), the subjectively innocuous third harmonic remained the highest in level, at –72dB (0.03%), with the second harmonic at –90dB (0.003%) and all other harmonics and power-supply–related components at or below –100dB. Intermodulation distortion at high power into 4 ohms was also extremely low (fig.10).


Fig.7 Constellation Performance Centaur, THD+N (%) vs frequency at 18V into: 8 ohms (blue), 4 ohms (magenta), 2 ohms (red).


Fig.8 Constellation Performance Centaur, 1kHz waveform at 50W into 8 ohms, 0.005% THD+N (top); distortion and noise waveform with fundamental notched out (bottom, not to scale).


Fig.9 Constellation Performance Centaur, spectrum of 50Hz sinewave, DC–1kHz, at 200W into 4 ohms (linear frequency scale).


Fig.10 Constellation Performance Centaur, HF intermodulation spectrum, DC–24kHz, 19+20kHz at 200W peak into 8 ohms (linear frequency scale).

Constellation Audio's Performance Centaur Mono is a powerhouse of an amplifier capable of delivering very high power with very low distortion into 4 and 8 ohms. While it was less comfortable with 2 ohms, it will have no problem driving low-impedance speakers with aplomb. Although I broke it with sustained delivery exceeding 1kW into 2 ohms (footnote 1), there isn't a speaker made that would not be the first to expire at this power level! Color me impressed.—John Atkinson

Footnote 1: When he received the amplifier back after the review had been published, Peter Madnick let me know that an internal fuse had blown, that the amplifier wasn't damaged.
Constellation Audio
3533 Old Conejo Road, Suite 107
Newbury Park, CA 91320
(805) 201-2610

xsipower's picture

Hello Mr. Atkinson,

I was wondering if you were told what failed on the first unit you were testing at 2 ohms. Was it a fuse? There are four small aluminum panels on the back of the amplifier when removed allow access to ATO (automotive type) fuses.

I'm also surprised the signal to noise measured so poorly. John Curl who designed the input side of the amplifier is known for his superb low noise small signal amplifier designs. See pictures of the inside of the Centaur (can be found by searching the web) they show a rats nets of wires crisscrossing around. Some of these wires come quite close to the transformer.

I understand from the article that this is a "budget" version of the flagship Hercules. It uses the same circuits, but with surface mount components and assembly done in China. Unfortunately the build quality inside the unit does not present itself as a piece of high end audio, especially at the price it's going for. I would say that the metal work of the case is amazing. They must spend a good percentage of the material cost on machining (probably done on a 5 or 6 axis machine).

I think the peaking in the high frequency response is because they do not use an output Zobel network in their design. From the picture of the amplifier module you can see that the output is directly fed from the Mosfets to the output terminals via bus bars that pass under the PCB. The output does not go through a high frequency filter (LRC network). This can make the amplifier susceptible to oscillation with capacitive loads and external ingress of RF signals.

I’m not against five or six figure amplifiers, but I am concerned with:

  1. Poor Signal to Noise for an amplifier of this price and pedigree.
  2. Input impedance way low (way out of specification).
  3. Iffy with loads less that 4 ohms ( specifications say 1000W into 2ohms)
  4. Innards build/wiring quality not up to the price being paid.
  5. Not a big deal, but it concerns me that it has a peaking response in the infra-frequencies.
  6. Probably paying a lot for the glitzy aluminum panels, rather than the meat inside.

Other than that I’m happy that Mr. Fremer enjoyed the units.


John Atkinson's picture

xsipower wrote:
I was wondering if you were told what failed on the first unit you were testing at 2 ohms. Was it a fuse?

Yes, as I write in the footnote to the measurements section in this Web reprint, Peter Madnick told me that a fuse had blown.

xsipower wrote:
I'm also surprised the signal to noise measured so poorly....

Me too, which is why I repeated the noisefloor measurement on both samples. But the noisefloor is still low enough not to be a factor in the amplifier's sound quality, I feel.

John Atkinson

Editor, Stereophile

Rick Tomaszewicz's picture

...for being so gimlet-eyed in reviewing ultra high-end gear.  This means we (the 99% - or 99.9%, as another commentor corrected me) can also rely on you for calling out underperforming affordable gear which has a more relaxed price/expectation ratio.

Perhaps your next gig should be evaluating, with the same thoroughness, large government programs prior to their launch!  

Vadim S's picture

Hello Mr. Fremer,

Thank you for the interesting reviews about some of the world's best power amplifiers.

Could you compare the Centaur monoblock performance to the performance of the Soulution 710 stereo power amplifier, at least concerning their basic sonic character? It seems like they share many qualities in common (basing on your reviews of the bespoke components).

Thank you in advance,