**Sidebar 3: Measurements**
Before performing any measurements, I ran one of the Ypsilon Aelius amplifiers (serial no.31) for an hour at one-third its specified maximum power of 67W into 8 ohms, thermally the worst case for an amplifier with a class-B output stage. At the end of that period, the heatsinks were moderately hot at 44.2°C (111.5°F), while the chassis was cooler at 40°C (103.9°F).

I then measured the amplifier, using *Stereophile*'s loan sample of the top-of-the-line Audio Precision SYS2722 system (see www.ap.com and the January 2008 "As We See It"). I first performed a complete set of tests using the Siemens NOS C3g input tube, whose sound Michael Fremer preferred, then repeated some tests using the Electro-Harmonix 6C45PiEH, Ypsilon's stock input tube for sales outside the US. I used the Aelius's unbalanced, RCA input; the balanced input appears actually to be single-ended, connecting pin 2 of the XLR jack in parallel with the unbalanced input.

The Aelius's voltage gain at 1kHz was the same with either input tube: 30.2dB into 8 ohms, which is slightly higher than specified. The Aelius preserved absolute polarity (*ie*, was non-inverting), and the input impedance was very slightly lower than the specified 47k ohms, at 45k ohms at low and middle frequencies, dropping slightly to 37k ohms at 20kHz.

The output impedance was high for a solid-state design, at 0.44 ohm at 20Hz and 1kHz, 0.47 ohm at 20kHz. As a result, the modulation of the amplifier's frequency response by the Ohm's Law interaction between this impedance and the impedance of our standard simulated loudspeaker was a moderate ±0.3dB (fig.1, gray trace). The audioband response in this graph is flat, though with a very slight downward tilt into 2 ohms (red trace), and steep rolloffs above 50kHz and below 30Hz into all loads. A 10kHz squarewave (fig.2) was reproduced with short risetimes but a critically damped overshoot, the latter presumably due to the interstage coupling transformer in the input tube's plate circuit. Fig.1 was taken with the C3g tube; repeating the measurement into 8 ohms with the 6C45 tube gave the response shown in fig.3. The ultrasonic rolloff is identical, reaching –3dB at 90kHz. However, there is now a small rise in response below 100Hz, peaking at +0.65dB at 17Hz. I doubt that this will be audible as extra bass, given how little energy music has below 40Hz, but it's possible that the additional group delay from this peak contributed to some people's preferring this tube.

Fig.1 Ypsilon Aelius, C3g input tube, frequency response at 2.83V into: simulated loudspeaker load (gray), 8 ohms (blue), 4 ohms (magenta), 2 ohms (red) (0.25dB/vertical div.).

Fig.2 Ypsilon Aelius, C3g input tube, small-signal 10kHz squarewave into 8 ohms.

Fig.3 Ypsilon Aelius, 6C45 input tube, frequency response at 2.83V into 8 ohms (red) (0.25dB/vertical div.).

I commend Ypsilon for including a ground-lift switch on the Aelius, though I found the lowest level of noise during the measurements with the ground connected with this switch. The noise levels with the two different tubes were basically identical; though there was slightly more 240Hz component with the 6C45 (fig.4, blue trace), this was not even close to an extent that would be audible. The wideband, unweighted signal/noise ratio with the input shorted was 75.8dB ref. 2.83V or 1W into 8 ohms, this improving to 89.3dB when A-weighted.

Fig.4 Ypsilon Aelius, spectrum of 1kHz sinewave, DC–1kHz, at 100W into 8 ohms with: C3g tube (red), 6C45 tube (blue) (linear frequency scale).

The lack of global negative feedback in the Aelius's unusual circuit—the input tube feeds the interstage transformer, which has two antiphase secondaries to feed the six matched pairs of output MOSFETs, and that's it!—means that the percentage of THD+noise steadily increases with the output power. Figs. 5, 6, and 7 show how this percentage changes into 8, 4, and 2 ohms, respectively. Below 1W or so into any load, where the measured distortion is very low, the percentage is dominated by noise. The actual THD then rises slowly before reaching clipping. Defining clipping as when THD+N reaches 1%, these three graphs indicate that the Aelius clips at 233W into 8 ohms (23.7dBW), 455W into 4 ohms (23.6dBW), and 444W into 2 ohms (20.45dBW). The first two powers exceed the specified maximum output power by 0.5dB or so; the third is 0.55dB lower, but this is most likely due to the fact that I don't hold the wall AC voltage constant for this test.

Fig.5 Ypsilon Aelius, C3g input tube, distortion (%) *vs* 1kHz continuous output power into 8 ohms.

Fig.6 Ypsilon Aelius, C3g input tube, distortion (%) *vs* 1kHz continuous output power into 4 ohms.

Fig.7 Ypsilon Aelius, C3g input tube, distortion (%) *vs* 1kHz continuous output power into 2 ohms.

Fig.8 shows how the Aelius's THD+N changed with frequency at a level where I could be sure I was looking at actual distortion: 5V, which is equivalent to just over 3W into 8 ohms. The distortion is respectably low between 100Hz and 5kHz, even into 2 ohms (red trace). It rises at low frequencies, presumably due to the onset of saturation in the coupling transformer's core, as well as in the top octaves, but not to any alarming level into the higher impedances.

Fig.8 Ypsilon Aelius, C3g input tube, THD+N (%) *vs* frequency at 2.83V into: 8 ohms (blue), 4 ohms (magenta), 2 ohms (red).

Perhaps more important, the nature of the distortion is almost entirely second harmonic (fig.9), which tends to be subjectively innocuous, with the third harmonic more than 20dB lower in level (fig.10). Figs. 5–10 were taken with the C3g tube. Fig.11 repeats the spectral analysis of fig.10, but with the C3g replaced by the 6C45. This is not as linear a tube as the C3g, the second harmonic increasing in level by 8dB, though the third harmonic lies at the same –81dB (0.009%). It's possible that this increased level of second-harmonic distortion also contributed to the feeling of warmth in the sound with this tube.

Fig.9 Ypsilon Aelius, C3g input tube, 1kHz waveform at 22W into 8 ohms, 0.154% THD+N (top); distortion and noise waveform with fundamental notched out (bottom, not to scale).

Fig.10 Ypsilon Aelius, C3g input tube, spectrum of 1kHz sinewave, DC–10kHz, at 15W into 8 ohms (linear frequency scale).

Fig.11 Ypsilon Aelius, 6C45 input tube, spectrum of 1kHz sinewave, DC–10kHz, at 15W into 8 ohms (linear frequency scale).

Finally, the increasing amount of distortion at the top of the audioband gives rise to a 1kHz difference product noticeably high in level under the worst-case condition of an equal mix of 19 and 20kHz tones, which is just below visible clipping on the 'scope into 4 ohms (fig.12). But the higher-order intermodulation products in this graph are only moderately high in level, and the midrange is commendably clean.

Fig.12 Ypsilon Aelius, C3g input tube, HF intermodulation spectrum, DC–24kHz, 19+20kHz at 200W peak into 4 ohms (linear frequency scale).

Ypsilon's Aelius is a most unusual amplifier, offering measured performance that in many ways is typical of a classic tube amplifier, but with an ability to drive low impedances usually associated with solid-state designs. Its sound quality will very much depend on the input tube fitted. Like MF, I preferred the NOS Siemens C3g.—**John Atkinson**