EAR 912 preamplifier Measurements
The EAR 912's phono stage offered gains of 50dB, 44dB, and 38dB in MM mode, depending on the position of the front-panel gain switch. The corresponding figures for MC operation were 19dB higher. The MM input impedance was 43k ohms at 20Hz, this increasing slightly to 50k ohms at 1kHz and 47k ohms at 20kHz. With the phono input set to MC and 40 ohms, I measured an input impedance of 422 ohms at 1kHz, this decreasing to 117 ohms at 20Hz and 378 ohms at 20kHz. The phono input preserved absolute polarity.
Fig.1 shows the 912's phono-stage frequency response, assessed at the main outputs. The RIAA correction appears to incorporate the IEC-recommended LF rolloff, but also features a slightly rising response above 5kHz. In this respect, it is less accurate than EAR's solid-state 324 phono preamplifier that AD reviewed in July 2004. Even when set to the highest gain, the 912's phono-stage signal/noise ratios were excellent. In MM mode, the A-weighted ratio was 81dB (ref. 1kHz at 5mV), this decreasing to a still good 67.7dB, wideband, unweighted. Due to the 912's use of a transformer to provide the additional gain required, the MC mode's S/N ratios were not appreciably different, at 79.7dB and 66.7dB, respectively (both figures referred to 1kHz at 500µV). This is a superbly quiet preamp.
Fig.1 EAR 912, RIAA error (right channel dashed, 1dB/vertical div.).
A downside of the very high gains available from the 912's phono stage is a reduced overload margin. The margins were acceptable at the lowest gain setting, ranging from 10dB at 20Hz and 20kHz to 21dB at 1kHz (all figures ref. 1kHz at 5mV). Each 6dB increase in gain, however, reduced the margin by the same 6dB. Owners of the 912 should set the phono-stage gain to the lowest level acceptable with their preferred cartridge. But with its gain set appropriately, the 912 offers quite low levels of both harmonic distortion (fig.2) and intermodulation distortion (fig.3).
Fig.2 EAR 912, MM mode, spectrum of 1kHz sinewave, DC–1kHz, at 1V into 8k ohms (linear frequency scale).
Fig.3 EAR 912, MM mode, HF intermodulation spectrum, DC–24kHz, 19+20kHz at 1V into 8k ohms (linear frequency scale).
The line stage offers a maximum voltage gain of 10.7dB, balanced input to balanced output. Peculiarly, the maximum gain for unbalanced operation was 2dB higher. Both inputs preserved absolute polarity; ie, were noninverting, and the XLR jacks appear to be wired with pin 2 hot. The balanced input impedance was 22k ohms at 20Hz and 1kHz, dropping slightly to 14.2k ohms at 20kHz, while the unbalanced input impedance was a constant 18k ohms across the audioband.
The line-stage output impedance was a suitably low 32 ohms balanced and 39 ohms unbalanced across most of the audioband, rising at 20kHz to still-low values of 65 ohms and 87 ohms, respectively. The front-panel VU meters were calibrated for an input of 775mV = 0dB (equivalent to a level of 1mW into 600 ohms), while the unity-gain setting of the Volume control was 2:00. The Volume control also showed excellent channel matching at all settings.
With the Volume control set to its maximum, the balanced frequency response into 100k ohms was flat to 30kHz, with a steep rolloff above that frequency (fig.4, top pair of traces). The increasing source impedance at high frequencies resulted in a slightly shelved-down top-octave output into the low 600-ohm load (fig.4, lower traces). The unbalanced response into 100k ohms (fig.5) was very different, with some ultrasonic peaking evident, though this should not have any subjective consequences. Channel separation was disappointing (fig.6), marred by capacitive coupling between the channels and resulting in just 28dB of separation at 20kHz, though 56dB was available at 1kHz.
Fig.4 EAR 912, Volume control at max, balanced frequency response at 1V into (from top to bottom at 200Hz): 100k ohms, 600 ohms (0.5dB/vertical div.).
Fig.5 EAR 912, Volume control at max, unbalanced frequency response at 1V into 100k ohms (0.5dB/vertical div.).
Fig.6 EAR 912, balanced channel separation (R–L dashed, 10dB/vertical div.).
Figs.7 and 8 show how the percentage of THD and noise in the 912's output varied with output voltage for balanced and unbalanced operation, respectively. The latter is actually slightly more linear, but both modes are beyond reproach at typical output levels. A maximum output of 9.5V (at 1% THD) is available from both modes into 100k ohms, and 8V into 600 ohms, both figures way more than enough for practical use. In fact, noting where the traces in these two graphs have their inflection points, it appears that the EAR 912's gain architecture is sensibly arranged to give the lowest distortion at the levels where the partnering power amplifier is close to being clipped.
Fig.7 EAR 912, balanced distortion (%) vs 1kHz output voltage into (from bottom to top at 1V): 100k ohms, 600 ohms.
Fig.8 EAR 912, unbalanced distortion (%) vs 1kHz output voltage into (from bottom to top at 1V): 100k, 10k, 1k ohms.
Its use of transformers does mean that the 912's line stage introduces rather more THD at very low frequencies than is usual (fig.9), though this is dominated by the subjectively benign second and third harmonics (fig.10). At higher frequencies all the harmonics drop considerably in level (fig.11), though a trace of 120Hz hum (at –84dB) could not be removed no matter how I adjusted the grounding between the 912 and my Audio Precision test set. Intermodulation distortion was also low in level (fig.12).
Fig.9 EAR 912, balanced THD+N (%) vs frequency at 1V into (from bottom to top): 100k ohms, 600 ohms.
Fig.10 EAR 912, balanced spectrum of 50Hz sinewave, DC–1kHz, at 1V into 100k ohms (linear frequency scale).
Fig.11 EAR 912, unbalanced spectrum of 1kHz sinewave, DC–1kHz, at 1V into 8k ohms (linear frequency scale).
Fig.12 EAR 912, unbalanced HF intermodulation spectrum, DC–24kHz, 19+20kHz at 1V into 8k ohms (linear frequency scale).
In general, the EAR 912's measured performance reveals the excellent audio engineering I have come to expect from Tim de Paravicini. But I was concerned by the disappointing channel separation and that (very faint) trace of hum in the output. It is fair to note that AD didn't note any hum in his system.)—John Atkinson