Bel Canto e.One Ref600M power amplifier Measurements

Sidebar 3: Measurements

I performed a full set of measurements on the Bel Canto Design e.One REF600M, using my Audio Precision SYS2722 system (see the January 2008 "As We See It"). As the REF600M uses a class-D output stage, ahead of the analyzer I used Audio Precision's auxiliary AUX-0025 passive low-pass filter, which eliminates noise above 200kHz that might otherwise overload the SYS2722's input circuitry.

The Bel Canto's voltage gain into 8 ohms measured 26.9dB from both its unbalanced and balanced inputs, though the latter result was obtained with the Audio Precision's balanced output floating with respect to ground. Grounding the XLR's pin-1 connection reduced the gain by almost 5dB and increased the noise level, so I floated the balanced ground for all the tests. Both balanced and unbalanced inputs preserved absolute polarity (ie, were non-inverting). The unbalanced input impedance measured close to the specified 100k ohms at low and middle frequencies, dropping to a still-high 58k ohms at 20kHz. The balanced input impedance was 77k ohms.

The REF600M's output impedance, including 6' of cable, was low, at 0.1 ohm from 20Hz to 20kHz. As a result, there was very little variation of its response with our standard simulated loudspeaker (fig.1, gray trace). This graph was taken with a balanced input; the unbalanced behavior was identical. Despite KR's conjecture, the amplifier's response is flat in the audioband, rolling off above 20kHz to reach –3dB at 40kHz. A 10kHz squarewave was reproduced with a very slight overshoot but no ringing (fig.2).


Fig.1 Bel Canto e.One REF600M, balanced frequency response at 2.83V into: simulated loudspeaker load (gray), 8 ohms (blue), 4 ohms (magenta), 2 ohms (red) (1dB/vertical div.).


Fig.2 Bel Canto e.One REF600M, small-signal, 10kHz squarewave into 8 ohms.

Although its class-D output stage produces 381mV of ultrasonic switching noise with a center frequency of 442kHz with no audio signal present, the Bel Canto was otherwise an extremely quiet amplifier. With the Audio Precision low-pass filter removing the switching noise, the unweighted, wideband signal/noise ratio (ref. 2.83V into 8 ohms with the input shorted to ground) was an extraordinary 109.9dB, which improved to 113dB when A-weighted. Spectral analysis of the amplifier's low-frequency noise floor while it reproduced a 1kHz tone into 8 ohms (fig.3) indicated that the only spuriae present were at 60Hz and its odd-order harmonics, though these all lie below –120dB and are thus negligible.


Fig.3 Bel Canto e.One REF600M, spectrum of 1kHz sinewave, DC–1kHz, at 1W into 8 ohms (linear frequency scale).

Fig.4 plots the percentage of THD+noise against power into 8 ohms. Distortion is extremely low up to 10W or so, and the REF600M clips (defined as 1% THD+N) at 340W (25.3dBW), exceeding the specified power of 300W into this load (24.8dB). The amplifier clipped at the specified 600W into 4 ohms (24.8dBW).


Fig.4 Bel Canto e.One REF600M, distortion (%) vs 1kHz continuous output power into 8 ohms.

How the Bel Canto's THD+N percentage varied with frequency at a level, 9V, where I could be sure I was examining actual distortion rather than noise, is shown in fig.5. Below 3kHz, the THD varies little with either frequency or load impedance, but does increase in the top octave. The 2 ohm trace (red) is a little irregular, apparently due to a slight history effect with the amplifier driving this very low impedance; that is, the measured percentage took a few seconds to stabilize. The distortion signature at this voltage into 8 ohms primarily consisted of low-order harmonics (fig.6), and the second and third harmonics were still predominant at high powers (fig.7). Even so, the third harmonic lay at just –86dB (0.005%). Intermodulation distortion was similarly very low in level, even at high power into 4 ohms (fig.8)


Fig.5 Bel Canto e.One REF600M, THD+N (%) vs frequency at 9V into: 8 ohms (blue), 4 ohms (magenta), 2 ohms (red).


Fig.6 Bel Canto e.One REF600M, 1kHz waveform at 10W into 8 ohms, 0.0049% THD+N (blue); distortion and noise waveform with fundamental notched out (red, not to scale).


Fig.7 Bel Canto e.One REF600M, spectrum of 50Hz sinewave, DC–1kHz, at 100W into 8 ohms (linear frequency scale).


Fig.8 Bel Canto e.One REF600M, HF intermodulation spectrum, DC–24kHz, 19+20kHz at 200W peak into 4 ohms (linear frequency scale).

Bel Canto Design's e.One REF600M amplifier packs a huge amount of power into a small package, and offers impressively low levels of distortion and noise.—John Atkinson

Bel Canto Design
221 N. First Street
Minneapolis, MN 55401
(612) 317-4550

georgehifi's picture

JA you never used to use this bench test filtering with Class D's, and in the past we got to see just what was coming out of these Class D speaker outputs.
But now with the Audio Precision's auxiliary AUX-0025 passive low-pass filter we're left screen shots and measurements of something that not representable of the product being tested.
Were the Class D manufactures complaining that the 1khz or 10khz square waves shots were looking more like a timber band-saw than a proper square wave?
In my view not a valid representation of the product on test.

Cheers George

John Atkinson's picture
georgehifi wrote:
JA you never used to use this bench test filtering with Class D's...

I have been using this filter for 10 years.

georgehifi wrote:
... and in the past we got to see just what was coming out of these Class D speaker outputs.

I still occasionally show class-D amplifier squarewaves with and without the low-pass filter. But without the filter, all you see is a wave with a large amount of RF noise on the tops and bottoms. As I have explained in the past, removing this noise with the filter allows me to see overshoot and ringing, which I regard as significant. Note that I always note the level and center frequency of a class-D amplifier's switching noise.

georgehifi wrote:
But now with the Audio Precision's auxiliary AUX-0025 passive low-pass filter we're left screen shots and measurements of something that not [representative] of the product being tested.

The AUX-0025 filter is necessary to prevent the Audio Precision analyzer's input circuitry being driven into slew-rate limiting by the class-D amplifier's ultrasonic switching noise.

georgehifi wrote:
Were the Class D manufactures complaining that the 1khz or 10khz square waves shots were looking more like a timber band-saw than a proper square wave?

No. And so what if they were?

georgehifi wrote:
In my view not a valid representation of the product on test.

And by expressing that opinion, all you are doing is revealing your own lack of insight, I am afraid.

John Atkinson
Editor, Stereophile

georgehifi's picture

Well JA I still like to see what's trying to ultrasonically fry my tweeters/ears. Especially some Class D at turn on that squeal like a stuffed pig for a second and have a very high level of HF switching noise hiss who's artifacts can be heard from the tweeter.

One of our well known reviewers that I know used his Wilson Watt 8's, reviewed a Nuforce 9se V3 for a few weeks, while reviewing them both had the turn on squeal and tweeter hiss, which Nuforce said was normal.

After the 2 week review, complaints were that the top end of his speakers were sounding a bit hard, under my instructions he got new tweeter diaphragms for me to insert (he's a reviewer, not very good with his hands) when I remove both old tweeter diaphragms I was shocked to see that the coils were blued with heat damage and the lacquer was bubbled because of either the squelch of the turn on, or the 800khz fundamentals of artefacts of the switching noise hiss that we can hear from the tweeter.

I get asked to recommend amps including Class D
This is why I like to see what's coming out of the speaker terminals of Class-D amps without it being hidden from view in test results, before recommending them and not through lack of insight.

BTW from a quick look and there must be others, it is a bit selective when this filter is applied, without as early as last year, not as of 10 years ago.

Please JA go back to showing both with and without the filter, not just with.

Cheers George

dumbo's picture

There must be some way to see the presence of these HF anomalies (if they exist) and the use of the filter on the measurement equipment could hide the possibility of this behaviour being present? Is it possible for HF noise at the upper limits of a Class D Amps bandwidth to find its way back down into the audible range via reflections and or load instabilities at those Freqs?

Do the Hypex based amps use output filters at all before the speaker output terminals? Do most none Class D Amps use output filters in general?

georgehifi's picture

"Do most none Class D Amps use output filters in general?"

Some linear amps have HF output filters if they're subject to oscillating, but I have found that if the linear amp is unconditionally stable there is no need for them

And yes all Class D's I know have a gentle (so it won't burn out) passive high wattage output filter, if not they would fry your speakers in a micro second.

But this filter can't filter out all the switching noise without totally effecting the audio band, this is why you'll see the remminance of that noise as the dreaded sawtooth ringing noise all across the top of a test square wave, some much more than others.

The Audio Precision's auxiliary AUX-0025 passive low-pass filter that JA uses goes between the class d amp speaker outputs and the measurement device, which eliminates most all of the switching noise, so you don't see it or measure it anymore. If it were left in place and speakers attached after it instead of the measurement device, it would sound very different to what it would if it were not there, and may well burn out very quickly if the level is turned up, because when used as stated it's only used with a small signal level.

PS: Once the switching frequency is >5 x higher than it is now, then you'll see that switching noise/sawtooth dissapear, and JA won't need to use that filter to hide it, but technology is't up there yet for that to happen, Technics I have have seen have made a very expensive >$20K class D poweramp that has double the switching noise to 1.5mhz!! this is a step in the right direction, 5mhz would be much better. Once this happens, our treasured big, hot, heavy, inefficient linear amps, will become door stops or boat anchors.

Cheers George