All amplifier measurements are performed independently by BHK Labs. All measurement data and graphical information displayed below are the property of the SoundStage! Network and Schneider Publishing Inc. Reproduction in any format is not permitted.
Note: Measurements were made at 120V AC line voltage with both channels being driven. Measurements made on right channel digitally fed via the AES/EBU input at a 24/96 sample rate. Unless otherwise noted, the Audio Precision Aux 0025 external low-pass filter was used to keep high-frequency spuria from contaminating the Audio Precision SYS 2722 measuring instrument.
See "Additional data" section (these are manufacturer-supplied specs)
- Output power: 190W @ 8 ohms, 240W @ 6 ohms
- This amplifier does not invert polarity through the digital or analog inputs.
- AC-line current draw at idle:
- Gain set +20 & 0, AES/EBU input, 96k Fs: 59W, 0.96PF, 0.5A
- Gain set -20 & below, AES/EBU input, 96k Fs: 30W, 0.94PF, 0.27A
- Gain: output voltage divided by input voltage, analog line input, gain set to 0dB: 13.0X, 22.3dB
- Input sensitivity:
- For 1W output into 8 ohms, analog line input, gain set to +30dB (max): 6.9mV
- For 1W output into 8 ohms, digital input, gain set to +30dB (max): -52.2 dBFS
- Output impedance @ 50Hz: 0.0024 ohm
- Input impedance @ 1kHz: 14.2k ohms
- Output noise, digital input, digital input level at 0, 8-ohm load, Lch/Rch:
- Gain at -30dB, wideband: 123/130uV, -87.2/-86.7dBW
- Gain at -30dB, A weighted: 40/40uV, -97.0/-97.0dBW
- Gain at 0dB, wideband: 137/143uV, -86.3/-85.9dBW
- Gain at 0dB, A weighted: 42/39uV, -96.6/-97.2dBW
- Gain at +30dB, wideband: 174/180uV, -94.2/-83.9dBW
- Gain at +30dB, A weighted: 74/71 uV, -91.6/-92.0dBW
- Output noise, analog input, input termination 1k ohm, 8-ohm load, Lch/Rch:
- Gain at -30dB, wideband: 124/118uV, -87.2/-87.6dBW
- Gain at -30dB, A weighted: 40/45uV, -97.0/-96.0dBW
- Gain at 0dB, wideband: 296/298uV, -79.6/-79.6dBW
- Gain at 0dB, A weighted: 154/160uV, -85.3/-85.0dBW
- Gain at +30dB, wideband (note: values in mV): 8.4/8.6mV, -50.5/-50.3dBW
- Gain at +30dB, A weighted (note: values in mV): 4.6/5.0mV, -55.8/-55.0dBW
The Devialet D-Premier is unique -- a volume-controlled power DAC that accepts both digital and analog inputs. Its uniqueness is in how it generates its output, being a combination of a low-powered, class-A analog output stage and a digital-switching section. The class-A stage controls the output voltage, and the switching section adds the necessary current to supply the output power. Also of note is the power-factor-corrected power supply, which measures close to unity. This is a good thing, as it makes the incoming AC line current sinusoidal rather than the usual 120Hz, 2-3ms rectifier-charging pulses of conventional capacitor input power supplies. The result is less crap on one’s AC power line, and less messing up of the sound of the other connected gear.
This D-Premier is extremely well protected against various things that might damage it or the load. As a consequence, it was difficult, if not impossible, to produce curves of power output vs. distortion that went into clipping below loads of 8 ohms, as is usual with other, more conventional amps that have been measured. Therefore, the usual measured output powers at 1% and 10% distortion are not shown in the additional data. A Devialet publication, "Advanced Practical Information," indicates that the D-Premier’s short-term RMS power output is doubled each time the load is halved, to a maximum total power of 600Wpc.
Another observation was that the D-Premier’s distortion and noise floor was pretty much the same at the 192kHz sample rate, so we used a 96kHz sample rate for most of the measurements. The output noise, measured without the Aux-0025 filter, varied from 5 to 12mV over the gain range of ±30 for the digital and analog inputs.
Chart 1 shows the frequency response of the D-Premier with varying loads. The Devialet’s output impedance is so low that no difference can be seen at the scale we usually use in testing analog amplifiers. Also of great significance is that the D-Premier lacks an output low-pass filter, as is necessary in almost all other switching amplifiers; as a consequence, the high-frequency response is not load dependent -- an interesting plus among many of this design.
Chart 1A is a plot of the D-Premier’s frequency response as a function of the incoming sample rate, at 44.1, 96, and 192kHz. (Note: This plot is exactly what one sees for regular D/A converters used to decode signals from the digital outputs of CD transports and other digital sources to produce analog outputs.) The frequency response for analog inputs is similar to that shown in Chart 1, as the sampling rate for the analog input is also 96kHz. Not shown is the low-frequency response, which was flat to below 10Hz at all sample rates. The pulse and squarewave response shape, with its symmetrical ringing, is indicative of FIR filters being used. This plot is done without the Aux-0025 low-pass filter, to allow the full bandwidth of the D-Premier to be measured.
Chart 2 illustrates how the D-Premier’s total harmonic distortion plus noise (THD+N) vs. power varies for 1kHz and SMPTE IM test signals and amplifier output load for 8- and 4-ohm loads. Amount of distortion is noise dominated up to perhaps 10-20W and then rises as distortion, per se, at higher power up to the power outputs shown on the chart. The amount of THD+N with the analog inputs was roughly twice as much. Looking at the D-Premier as a high-voltage-output D/A converter, I plotted the THD+N amplitude not as a percentage of reading, but as dB down from full scale as a function of decreasing input level below 0dBFS. I and others commonly do this to reveal any glitches in distortion level at various input levels, and also to easily illustrate the noise floor of the device when its input levels get way below where distortion, per se, occurs. This is shown in Chart 2A. A major aspect of this curve is that the noise floor is at about -115dBFS -- one of the lowest I have measured for a D/A converter over the years. However, the analog inputs were not so quiet, with a noise floor closer to -105dBFS.
Chart 3 shows the D-Premier’s THD+N as a function of frequency for 4-ohm loading at several different power levels. The apparent increase in distortion at high frequencies is reasonably low. Again, the Devialet’s protection circuitry prevented the taking of any measurements near the maximum amount of power the amp can deliver with music signals.
The damping factor vs. frequency, shown in Chart 4, is very high, and remains high to a far higher frequency than is typical of analog power amplifiers.
Chart 5 plots a spectrum of the harmonic distortion and noise residue of a 10W, 1kHz test signal. The magnitude of the AC-line harmonics is relatively low, except for a prominent output at 120Hz. Signal harmonics are low in amplitude, the second, third, and fifth harmonics being the most significant.
I listened to this amplifier in my system quite a bit, and found it to be most revealing, clear, and musical. I wish I owned it!
Chart 1 - Frequency response of output voltage as a function of output loading
Red line = open circuit
Magenta line = 8-ohm load
Blue line = 4-ohm load
(Note that the curves are so close together, it is not possible to see the different colors.)
Additional: Chart 1A
Chart 2 - Distortion as a function of power output and output loading
(Line up at 10W to determine lines)
Top line = 8-ohm SMPTE IM distortion
Second line = 4-ohm SMPTE IM distortion
Third line = 4-ohm THD+N
Bottom line = 8-ohm THD+N
Additional: Chart 2A
THD+N vs. decreasing input level in dB down from full scale; 0 dBFS = 35.8V output
Chart 3 - Distortion as a function of power output and frequency
Red line = 2W
Magenta line = 20W
Blue line = 60W
Cyan line = 150W
Chart 4 - Damping factor as a function of frequency
Damping factor = output impedance divided into 8
Chart 5 - Distortion and noise spectrum
1kHz signal at 10W into a 4-ohm load