Croft Acoustics Phono Integrated integrated amplifier Measurements
Sidebar 3: Measurements
I measured the Croft Phono Integrated using Stereophile's loan sample of the top-of-the-line Audio Precision SYS2722 system (see the January 2008 "As We See It" and www.ap.com). Before performing any measurements on an amplifier, I run it for an hour with both channels driven at 1/3 its rated power into 8 ohms; this is the level at which the maximum amount of power is dissipated in the output devices with a class-A/B topology. At the end of the first 30 minutes running at 1/3 the specified 45Wpc into 8 ohms, the top panel over the Phono Integrated's output stage was very hot, at 118.4°F (48°C), though it got no hotter for the remaining 30 minutes of preconditioning.
Looking first at the phono input, I examined its performance at the Line Out jacks, which appear to be placed ahead of the volume control but after the Mute switch. These offer just under unity gain, with a very high output impedance: almost 10k ohms at low and middle frequencies, dropping slightly to 7.5k ohms at the top of the audioband. This suggests that there is no output buffer stage ahead of these jacks, merely a high-value series resistor. Signal polarity was preserved from these output jacks for both phono and line input signals. The phono input appeared to be optimized for moving-magnet cartridges, with a gain at 1kHz of 40.9dB and an input impedance that varied from 49k ohms at 20Hz to 44k ohms at 1kHz and 36.5k ohms at 20kHz.
The Croft's phono-input RIAA error is shown in fig.1. The two channels match superblyy well; however, the phono stage suffers from a severe rolloff in the treble, reaching 3dB at 10kHz and 6dB at 20kHz. The low frequencies are boosted a little, reaching +1.8dB at 20Hz. Phono-input channel separation at high frequencies was poor, at <35dB RL and <41dB LR above 10kHz. The separation did increase at lower frequencies, reaching 56dB RL and 60dB LR at 1kHz. Phono noise was relatively low, with unweighted, wideband signal/noise ratios (measured with the input shorted and ref. 1kHz at 5mV) of 52.9dB in the left channel, 50.7dB in the right. These ratios respectively improved to 55.4 and 56.3dB when the measurement bandwidth was restricted to the audioband, and to 75.5dB when A-weighted.
The phono-stage overload margin was good at high and middle frequencies, at almost 20dB ref. 1kHz at 5mV. But the margin was very poor at low frequencies: just 1.8dB at 20Hz. The phono stage's distortion signature was predominantly the subjectively innocuous second harmonic (fig.2). When it came to high-frequency intermodulation, however, the Croft's phono stage did poorly, an equal mix of 19 and 20kHz tones at 50mV giving a second-order difference product at 34dB (1.5%, fig.3).
Turning to the Croft's performance via its line inputs, assessed at the speaker terminals, the Phono Integrated offered a maximum gain into 8 ohms at 1kHz of 34dB, the two channels differing by just 0.08dB. The input impedance was very high at almost 80k ohms at low and middle frequencies, dropping inconsequentially to 51k ohms at 20kHz. The amplifier as a whole inverted absolute polarity at the speaker terminals.
The output impedance was very high for a solid-state design, at 2.15 ohms at 20Hz and 1kHz, dropping slightly to 2 ohms at 20kHz. As a result, the variation in the Croft's frequency response due to the Ohm's law interaction of this impedance with that of our standard simulated loudspeaker was relatively large, at ±1.2dB (fig.4, gray trace). This graph also indicates that while the Croft has a wide small-signal bandwidth, there are the beginnings of an ultrasonic peak, suggesting an incipient instability. However, while the traces in this graph were taken with the twin unganged volume controls at their maximumnote the excellent channel matchingturning down these controls to 6:00 both eliminated the response peak and reduced the bandwidth to 3dB at 50kHz (fig.5). This graph was taken with the settings of the volume controls visually matched; note the channel imbalance of 0.7dB. This was typically the best balance I could achieve by matching the controls' positions by eye. Art Dudley likes this volume-control arrangement; I find it ergonomically unsettling.
With its volume controls at 6:00, the Croft's reproduction of 10kHz and 1khz squarewaves was excellent (figs.6 & 7, respectively). Channel separation at 1kHz (not shown) was moderate, at 63dB RL and 78dB LR, decreasing due to the usual capacitive coupling between channels to 38dB RL and 52dB LR. The unweighted, wideband S/N ratios, ref. 1W into 8 ohms (2.83V), taken with the line input shorted but the volume controls set to their maxima, were good at 68.4dB, improving to 82.9dB with the measurement bandwidth restricted to the audioband, and 84.6dB when A-weighted. Fig.8 shows the spectrum of the Croft's output while it drove a 1kHz tone at 1W into 8 ohms. You can see that, other than the full-wave rectified spurious tone at 120Hz, all the power-supplyrelated spuriae lie below 90dB.
The Croft's output stage is not very linear, and is actually reminiscent of a tube amplifier. Figs.9 and 10 show how the THD+noise percentage changes with output power, both channels driven, into 8 and 4 ohms, respectively. The THD is already >0.2% at just 100mW, and rises with increasing power. Into 8 ohms (fig.9), the THD almost reaches our usual definition of clipping at just 8W, but drops a little above that power. It crosses the 1% line at 33W (15.2dBW), and the amplifier reaches its specified power of 45W into 8 ohms (16.5dBW) only at 3% THD. The picture is worse into 4 ohms (fig.10), the THD reaching 1% at just 700mW, and reaching 55W (14.4dBW) at 3% THD. One complicating factor with this measurement was that, into 4 ohms, the output with continuous drive slowly declined with time, the implication being that the Croft's power supply begins to collapse with sustained high levels of current delivery.
Fig.11 shows how the THD+N percentage changed with frequency at 2.83V into 8 ohms (blue and red traces) and 4 ohms (cyan, magenta). Though the THD didn't change at different frequencies, it is high overall. Yes, it predominantly consists of the subjectively innocuous third harmonic (fig.12), but at this level, 1.23% at 1W into 4 ohms, the distortion will certainly be audible with pure tones, if not music. (Try the test tones with various levels of second, third, and seventh harmonic that I included on Test CD 2, Stereophile STPH004-2.)
At low frequencies and an 8 ohm load, the second harmonic is the highest in both channels (fig.13), but at higher powers, the third and fifth predominate (fig.14). Intermodulation distortion is similarly high, even at a level well below clipping (fig.15).
Without any circuit board, the three-dimensional, hard-wired layout of the Croft's circuit is a thing of wonder. However, I must admit to being puzzled why Art Dudley liked the sound of the Croft Phono Integrated as much as he did. To me, it seems, at best, inadequately engineered, and at its worstthat nonflat RIAA response, the high levels of harmonic and intermodulation distortionjust plain inadequate. I therefore asked Stephen Mejias to take a listen to the Croft. His thoughts on its sound appear in this issue's "Follow-Up" section.John Atkinson