Constellation Inspiration Stereo 1.0 power amplifier Measurements

Sidebar 3: Measurements

I measured the Constellation Inspiration Stereo 1.0 with my Audio Precision SYS2722 system (see the January 2008 As We See It"). Before performing any tests, I ran it at one-third its specified clipping power into 8 ohms for an hour. At the end of that time, both the top panel and the perforated side panels were warm, at 106.4°F (41.4°C); the gain at the speaker terminals was 25.1dB for both the balanced and single-ended inputs—slightly below the specified 26dB; and the output inverted absolute polarity with both inputs.

The Constellation amplifier's input impedance at low and middle frequencies was extremely high, at >1M ohm balanced and >310k ohms unbalanced. (I can't be more specific as the voltage drop method I use becomes increasingly inaccurate as the amplifier's input impedance increases.) The impedances dropped to a still-high 690k and 130k ohms, respectively, at the top of the audioband. I was puzzled by this, as the specified impedances are 20k ohms, balanced, and 10k ohms, unbalanced. Perhaps the circuit was changed after the specifications were published, but a very high input impedance is in no way a problem.

The output impedance was low, varying from 0.11 ohm at 20Hz and 1kHz to 0.12 ohm at 20kHz. Measured into our standard simulated loudspeaker, the Inspiration's frequency response varied by only ±0.1dB (fig.1, gray trace). Fig.1 also shows that the response was flat to 20kHz with all load impedances, but that the ultrasonic output increasingly rolled off above 20kHz as the load impedance dropped. Into 8 ohms (fig.1, blue and red traces) the output was down by just 1.3dB at 200kHz; a 10kHz squarewave in to that load was therefore reproduced with very short risetimes (fig.2), though a slight hint of overshoot was present on the waveform's leading edges.


Fig.1 Constellation Inspiration Stereo 1.0, frequency response at 2.83V into: simulated loudspeaker load (gray), 8 ohms (left channel blue, right red), 4 ohms (left cyan, right magenta), 2 ohms (green) (1dB/vertical div.).


Fig.2 Constellation Inspiration Stereo 1.0, small-signal, 10kHz squarewave into 8 ohms.

Channel separation (not shown) was excellent, at 115dB at 100Hz, though it decreased with increasing frequency, to 72dB at 20kHz. The Inspiration was also very quiet, its unweighted, wideband signal/noise ratio, ref. 2.83V into 8 ohms and assessed with the unbalanced input shorted to ground, measuring a superb 88dB (average of both channels). Restricting the measurement bandwidth to the audioband improved the ratio to 95.2dB, and A-weighting the measurement further improved it, to 98.1dB. Spectral analysis of the Stereo 1.0's low-frequency noise floor (fig.3) indicated that while some spuriae were present at the AC supply frequency and its harmonics, these were all at or below –105dB in the left channel (blue trace), and mostly lower than that in the right channel (red).


Fig.3 Constellation Inspiration Stereo 1.0, spectrum of 1kHz sinewave, DC–1kHz, at 1W into 8 ohms (linear frequency scale).

The Inspiration Stereo 1.0 is specified as delivering 200Wpc into 8 ohms or 400Wpc into 4 ohms (both equivalent to 23dBW) at 1% THD+noise. Figs. 4 and 5 respectively plot its percentage of THD+N against power into 8 and 4 ohms, and reveal that at 1% THD+N with both channels driven, the amplifier delivers more power than specified: 340W into 8 ohms (25.3dBW) and 480W into 4 ohms (23.8dBW). I tried repeating this test into 2 ohms, with just the left channel, but the Stereo 1.0 turned itself off before the THD+N reached 1%. When I turned it back on, only the right channel was operating, which suggests that one of the internal high-voltage fuses had blown. I could see no way to get inside the chassis to replace the fuse and the manual strongly advises leaving fuse replacement to the dealer.


Fig.4 Constellation Inspiration Stereo 1.0, distortion (%) vs 1kHz continuous output power into 8 ohms.


Fig.5 Constellation Inspiration Stereo 1.0, distortion (%) vs 1kHz continuous output power into 4 ohms.

Fortunately, I had by then performed almost all of my tests, other than examining how the Constellation's THD+N percentage varied with frequency. I thus performed that test with just the right channel operating (fig.6), and at a level high enough that I could be certain I was examining distortion rather than noise: 20V, which is equivalent to 50W into 8 ohms, 100W into 4 ohms, and 200W into 2 ohms. Fig.6 reveals that the Constellation offered very low distortion into 8 ohms (blue trace), slightly higher distortion into 4 ohms (magenta), but about 10 times as much THD+N into 2 ohms (red). There is the usual rise in THD+N in the top two audio octaves, but still only to levels that are respectably low. As well as being very low in level, the distortion is predominantly the subjectively innocuous third harmonic (fig.7), though at very high powers into 4 ohms the third harmonic is joined by lower levels of second and fifth harmonics (fig.8). Tested with an equal mix of 19 and 20kHz tones, with the peak signal level reaching 200W into 4 ohms, the increase in THD+N seen in the top octaves in fig.6 results in quite a few intermodulation products (fig.9)—but of these, the highest in level are still at only –86dB (0.005%), and the potentially audible product at 1kHz lies almost 100dB down.


Fig.6 Constellation Inspiration Stereo 1.0, right channel, THD+N (%) vs frequency at 10V into: 8 ohms (blue), 4 ohms (magenta), 2 ohms (red).


Fig.7 Constellation Inspiration Stereo 1.0, 1kHz waveform at 50W into 8 ohms, 0.0029% THD+N (top); distortion and noise waveform with fundamental notched out (bottom, not to scale).


Fig.8 Constellation Inspiration Stereo 1.0, spectrum of 50Hz sinewave, DC–1kHz, at 200W into 4 ohms (linear frequency scale).


Fig.9 Constellation Inspiration Stereo 1.0, HF intermodulation spectrum, DC–30kHz, 19+20kHz at 200W peak into 4 ohms (linear frequency scale).

Overall, the Constellation Inspiration Stereo 1.0 offers quite respectable measured performance, and delivers considerably more power than its conservative specification claims.—John Atkinson

Constellation Audio
Suite 1, Level 6, 580 Street Kilda Road
Melbourne, Vic 3004
(805) 201-2610

georgehifi's picture

I agree, but! it can be a problem if left open circuit (nothing attached) or used with un-shielded interconnects which many whacko's do, as then it is very susceptible to RF break through.

Also regarding wattage advertised, purposely understating the 8ohm wattage, this is starting to become a disturbing smoke screen trend these days, advertised 200Wpc into 8 ohms or 400Wpc into 4 ohms make this amp look as though it can double into each having of impedance load making it look as though its real current pusher into nasty speaker loads like the Wilson Audio Alexia.
When as you tested it gives in fact 340W into 8 ohms and only 480W into 4 ohms. Which is not a very good current pusher, pretty weak in fact and may not drive the Alexia's .9ohm EPDR load in the bass very well at all.

In a way Larry Greenhill heard what the circuit/topology limitations are in the bass EG: not much current.
"However, the Constellation didn't go as low in the deep bass. This meant that dynamic contrasts involving deep bass, such as sustained pedal chords from pipe organs, were more emotionally involving with the No.334, shaking my listening room more. When I played the Toccata of Widor's Organ Symphony 5, as performed by Jonas Nordwall and recorded by John Atkinson (24/88.2 AIFF file), the No.334 delivered more thunderous deep-pedal notes that pressurized the room"

Even though the ML334 is only a 125w amp it can push more current into the bass of his Revels than the 340w Constellation can.

Cheers George

dce22's picture

"Even though the ML334 is only a 125w amp it can push more current into the bass of his Revels than the 340w Constellation can."

Not really?
ML334 can do 240W squeaky clean sine to maxedout clipped 510W squarewave in 4 ohm,

The difference in low frequency performance comes from the subsonic filter that distorts the signal in Constellation.

CG's picture


First, I hope Larry Greenhill wasn't sitting in the listening position for hours during that 90 dB break in period. That's OSHA limit sound levels. Interestingly, with that listening position and those speakers with their efficiency performance, that sound level only requires about 3 (three!) watts in the bass region. That's assuming more or less monophonic source material in the bass and the measured roughly 4 Ohm speaker impedance down there.

I don't think any voltage source amplifier can "push" more power into any load resistance than another, based on Ohm's Law and basic physics. More current equals more volts (and vice versa). That's assuming that you're not into amplifier current limiting, of course. Since the Constellation can deliver 340 watts into 4 Ohms with relatively low distortion, you're not into current limiting up to that level. If one presumes that the (mostly) undistorted power output is limited by voltage clipping at higher load resistances (like 8 Ohms) and by current limiting at lower impedances (maybe even 4 Ohms), it only takes high school math and physics to determine the current limits of the amplifier.

So, do you think the lower bass impact is what's shown in figure 1? That is, the amplifier amplitude response being down by 3 dB at 20 Hz? Seems likely to me...

(Twist on usual disclaimer: I've never seen or heard any Constellation product. I don't think I've seen or heard any Levinson product since Harmon purchased Madrigal. Point is, I don't have a horse in this race. Wait - I have met people who are hearing impaired by exposure to too loud sound for too long periods. So, I'll stick to my one-sided bias of not playing music too loudly.)

dc_bruce's picture

Mr. Atkinson's measurements show this amplifier's amplitude response is about 2.5 dB down at 20 Hz. That seems to be audible, based on the review. A number of full range speakers will be flat in-room to 25 Hz or below, so this kind of response droop is more than academic.

Perhaps this amplifier's virtues are best appreciated with a stand-mount speaker, most of which fail to reproduce much below 40 Hz.

Still, for $11,000, I think a buyer would reasonably expect flat response to 20 Hz, on the off chance that he or she might some day replace that bass-shy stand mount speaker with something bigger and more powerful in the bass.

georgehifi's picture

dce22 "Not really?
ML334 can do 240W squeaky clean sine to maxedout clipped 510W squarewave in 4 ohm,"

This show the ML334 can do the doubling into half the impedance far better that the Constellation can, which means it's far better at supplying current.

Cheers George

dce22's picture

"Doubling into half the impedance far better that the Constellation"

Yes ML334 can drive lower resistance without brownout but by the time your in 400 watt you are crunching deep into the voltage rail its all gone wrong, while Constellation is squeaky clean up to 410 watts and bearable at 510 watt while ML is fully flatlined,

So the current is not a problem into these speakers its the distortion ML is pristine into the lower frequency while Constellation distorts the low end creating that light sound

Example - Figure 1 and Figure 6 blue line

ML334 has lower voltage rail more stability at 2ohm, 1ohm load,
on these 3,5 ohm Revels Constellation amp has 3db more dynamic range that will never come to play normal levels are maxing out at 10 Watt with some peaks at 100 Watts

Its easy to fall into the trap of
It sound like this because of that (Current ability in this case)

Myself did the same when designing a tube amp for a friend,
trying to fix a specific sound quality issue focusing on one fancy thing that seems logical when infact its something less sexy and more simple.


georgehifi's picture

These ML2's at only 25w per channel will drive "to a given volume" the Wilson Alexia better than the Constellation can, and even the ML334 as well. As they will stay flat and not act like a tone control like the Constellation and to a lesser degree the ML334 will.

This is why I'm so against manufacturers understating the 8ohm wattage just so it looks like it's doubling into 4ohms, to give the impression that their amp has massive current ability.

Cheers George

dce22's picture

Constellation amp from the fig 4 and 5 looks like is designed to double from 200 watt 8ohm to 400 Watt 4 ohm but the designer upped the voltage rail to give it more dynamic range into 8 ohms,
Nelson Pass does the same thing in his Pass Labs Amps.

This is ok when you see the same shape in 8,4,2 ohms just shifted in the THDvsWatt curve like on this amp on PassLabs and on ML334,
alright there is no curve for 2ohms on this one.

If the 2ohm curve starts to bend up early then it will wimp out on Alexia's

On 4 ohm load it is actually cleaner than ML334 because ML will crush the signal into the rail and the current window will close
(maximum at 100% distortion is 510 Watts),
Constellation will continue to deliver clean current up to 410 Watts and 3% distortion at 510 Watt so it can do more current into 4 ohm even if it has weaker Transformer then ML334.

Specs dont tell jack, audio manufacturers need to print a graph so you can see if it starts to distort up more (gain accounted of course) into low resistant loads or stay idential, does not matter if it double up on power if distortion spectrum is stable aka identical into 8,4,2 ohms then it will sound fine.

The messed up tone control is indeed what damages the sound.

georgehifi's picture

The end result is this amp was never going to get the best out of the Revel Ultima Salon2 impedance curve, the ML334 did a better job. All you have to do is look at the remarks from the reviewer (below) and then draw a conclusion from the tests results.

Larry Greenhill
"However, the Constellation didn't go as low in the deep bass. This meant that dynamic contrasts involving deep bass, such as sustained pedal chords from pipe organs, were more emotionally involving with the No.334, shaking my listening room more.
The conga drum that begins "Hotel California," from the Eagles' Hell Freezes Over was well defined and tight through the Stereo 1.0, but more 3D and room-shaking through the No.334.
The heartbeat that opens "Breathe (in the Air)," from Pink Floyd's The Dark Side of the Moon was a well-defined, dense, solid, pitch-perfect pulse with the Stereo 1.0, but not as ominous or as room-shakingly massive as with the No.334."

Cheers George

dce22's picture

Constellation can push more current into Revel Ultima Salon2 then ML334 its Ohm's law
If you turn up the volume so that 11 Amps flows to the Revel's

Constellation Will sound a bit distorted
ML 334 Will sound like screeching noise (more current capability blocked by voltage rail to delever less)

ML will sound better at 1 ohm speakers,
you need voltage so that current flows thru the wire
Constellation has more voltage and at Ultima Salon2 curve will not runout of current at clipping.

Current capability in this case does not limit the low frequency sound quality its the circuit design is it subsonic filter or is it overactive DC servo you need circuit schematic to tell the tests say that is something like that nothing to do with current.

You are attributing the cause of the problems to something unrelated.


Edit : By "push current" i mean capable of current flow without bending out of shape, distorting in irregular fashon different that higher impedance.

georgehifi's picture

Sorry, you can't see the forest through the trees, I'm not talking loudness WATTS for rave parties!! If I want that I'll get a 1000w Crown. I'm talking the abilty to best drive (to a certain volume) the review and test comments back it up.

Would you think a pair of 25w ML2 monoblocks would drive the Revells better (to a given volume) than the Constellation can, yes, because they can do current better, just like the ML334 can.
Yes these are only 25w into 8ohms

Cheers George

dce22's picture

Not denying that 25 watts good amp can sound better,
you are mixing current capability with distortion, if you design in same front end distortion mechanism that is in Constellation into ML334 it will sound alike

Fig. 6 blue line 12Watts 8Ohm you can see the distotion mechanism in action

Does not matter the load impedance, current, loud or quiet it will sound the same.

Trying to tell you that Constellation can do more current/power more dynamic range into the revels and the sound is not connected to your misguided opinion that its because loads or currents.


dce22's picture

Take ML 334


and change the transformer to the same core just change the winding to output higher voltage and you will get


The same amp....

does that means that now the ML 334 sounds different, not really.

Pass Labs does this Constellation does this not doubling on power is not precursor to bad sound you need to look at the distorion curve does it change into higher load if the shape is the same just shifted all is good.

georgehifi's picture

"you are mixing current capability with distortion,"

No you sorry are mixing high loudness distortion with quality normal listen level that has better everything dynamics included. And the ML2's and ML334 will do it better. End of story.

Cheers George

dce22's picture

If the amp is more capable at high power it will not brake a sweat at normal listening level.

ML2 25/50/100 Watt vs Constellation 300/400/200* Watts

*Let's say for the sake of argument that it can do at least 200 at 2ohm

Costellation has better dynamics any way you slice it, its all in signal distortion.

You are behaving like a conspiracy theorists even in the light of clear evidence you are clinging on that the earth is flat.

Just trying to help people get a better reading of the Measurement graphs not to change your mind really, people that does not know much can easy be steered in a wrong way and snowball your lie that the heavy amp with lots of metal that goes 5/10/20/40 watts at 8/4/2/1 ohm is more dynamic than Class D ncore KW amp,
and it sounds better because its big and heavy and uses lots of power not because it distorts less or in a different manner.

No hard feelings or anything personal i just think you are not informed enough recommending Bob Cordell's book "Designing Audio Power Amplifiers".


georgehifi's picture

Sorry you can understand what I said therefore you can't see the forest through the trees.

The proof to you is, the 25w ML2 will sound and drive better "to a given level" the .9 ohm load of the Alexia and also to a degree the Revells here than the 380w Constellation can. As the Constellation will act like a bit of a tone control because of the Alexia's impedance variations from 20hz to 20khz. Where both the ML's will stay flatter. End of story.

Cheers George

dce22's picture

"ML2 will sound and drive better "to a given level" the .9 ohm load of the Alexia"

Yes you are correct

"As the Constellation will act like a bit of a tone control because of the Alexia's impedance variations from 20hz to 20khz. Where both the ML's will stay flatter."

Yes you are correct and on that "Tone Control" Constellation has another "Tone Control" on the frontend of the amp that creates the bass discrepancy between the ML334 and Constellation on Revel's because the Output stage has no problems driving the Revel Ultima Salon2 as good as ML334

Constellation output stage is better at driving the Revel Ultima Salon2 than ML334 by virtue of higher output voltage

ML334 is voltage limited on Ultima Salon2
Constellation is current limited on Alexia

The Measurments confirm that low frequency impact on Constellation is caused by the amp front end distortion not the output stage so more transistors and bigger transformer to drive 1ohm at kilowatts will not change the bass performance of the amp it will still sound the same.

You get what im saying.


georgehifi's picture

a 25w ML2 will sound and drive better "to a given level" the .9 ohm load of the Alexia than the Constellation, so will the ML334.
I don't care about party levels, it was all about picking the right speaker to do an amp review with, and in this case it wasn't, as the reviewers comments and tests bore out.

"Yes you are correct"

That all I wanted to say. By

Cheers George