Rogue Audio Titan Atlas Magnum power amplifier Measurements
Sidebar 3: Measurements
To take these measurements, I used 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). The sample Erick Lichte shipped to me for measurement had its output transformer set for a 4 ohm load and was fitted with KT90 output tubes. Before performing any tests, I made sure each output tube was correctly set to 35mA bias current, using the top-panel meter. I then repeated the testing using KT120 tubes; as these were too fat to fit through the holes in the top panel, I removed the panel. (I'm told that in subsequent production Rogue will enlarge these holes.) The KT120s were each biased to 35mA for the tests; I then increased the bias current to 45mA, which EL felt gave the best sound quality, and repeated some of the tests. Finally, I hooked up just the left-channel output to the 8 ohm transformer tapI lacked the necessary right-angle nut driver to hook up the right channel's 8 ohm tapand again repeated some of the tests.
The Rogue's voltage gain into 8 ohms from the 4 ohm tap was lower than the norm, at 24.8dB, with both KT90 and KT120 tubes, and the amplifier preserved absolute polarity (ie, was non-inverting). Commendably, the channel gains were closely matched, within 0.1dB. As with the original Atlas, the gain into 8 ohms was the same from the 8 ohm tap, but when I investigated this more closely, it appeared that the slightly higher unloaded gain from this tap compared with the 4 ohm tap was offset by the slightly higher source impedance: 0.485 vs 0.375 ohm. These figures were taken at 1kHz and were not affected by the output tube used. The source impedance increased at the frequency extremes, to 0.8 ohm at 20Hz and 0.9 ohm at 20kHz, 4 ohm tap; and 1 ohm at 20Hz, 1.2 ohms at 20kHz, 8 ohm tap. The input impedance was usefully high, at well above 200k ohms at all audio frequencies.
Because of the moderate source impedance from the 4 ohm tap, the change in frequency response with our standard simulated loudspeaker was mild, at ±0.4dB (fig.1, gray trace). The Titan Atlas Magnum's response rolls off slightly in the extreme bass and more quickly above 30kHz, with a faster rolloff into low impedancesthe response is flat to 30kHz into 8 ohms (fig.1, blue and red traces), but 1.8dB at 20kHz into 2 ohms (green trace). Fig.1 was taken with KT90 tubes; the response with KT120s was identical. You can see that the upper-frequency rolloffs in this graph are disturbed by a small discontinuity. This is due to an ultrasonic resonance in the output transformer; though well damped, it does give rise to some slight overshoot and a couple of cycles of ringing in the amplifier's reproduction of a 10kHz squarewave (fig.2). This resonance has a clearer effect on the response from the 8 ohm tap (fig.3), but it is well down in level.
Channel separation (not shown) was good, any crosstalk below 1kHz or so lying below the noise floor and still around 70dB in both directions at 20kHz. The two channels differed in their noise performance, the unweighted, wideband signal/noise ratios (ref. 2.83V into 8 ohms) measuring 75.2dB in the right channel but only 67.4dB in the left. A-weighting the measurement increased the S/N ratios to 85.6dB left and 88.0dB right. I experimented with every possible means of connecting the signal grounds between the Audio Precision test set and the amplifier, but nothing improved these measured ratios. Investigating the reason for the difference, I found a higher amount of 60Hz hum and its odd-order harmonics in the left channel than in the right, which may well have something to do with the fact that the left-channel circuitry is immediately adjacent to the AC power transformer, whereas the right-channel circuitry is next to the two output transformers.
Fig.4 shows how the percentage of THD+noise in the output of the Rogue when fitted with KT90s varies with output power into loads ranging from 2 ohms (top) to 16 ohms (bottom). Below clipping, the actual distortion rises in a linear manner with increasing power. This is both different from how the original Atlas performed and suggests a reduction in overall loop negative feedback with the more recent design. The distortion is respectably low at low powers, especially into higher impedances, but with clipping defined as 1% THD, the Titan Atlas Magnum's 4 ohm tap only just meets its specified power output of 90Wpc into 4 ohms (16.5dBW), and exceeds that power into 2 ohms only if the definition of clipping is relaxed to 3% THD, when the amplifier delivers 117W (14.7dBW). Into 8 ohms, the amplifier clips at 60Wpc (17.7dBW), the lower power being due to the mismatch between load and tap.
Fig.5 repeats the measurements with the amplifier fitted with KT120 output tubes, each biased to the same 35mA. By comparing figs. 5 and 4, you can see that while the KT120s don't allow the amplifier to deliver any more power at clippingexcept into 2 ohms, when it delivers 120W (14.7dBW) at 1% THDthe Titan Atlas Magnum is more linear below clipping with KT120s than with KT90s.
Figs. 6 and 7 show how the small-signal THD+N varies with frequency into loads varying from 2 to 8 ohms with KT90 and KT120 tubes, respectively. Midband distortion is respectably low, but there is a major rise in THD in the top three audio octaves, especially into lower impedances.
Fortunately, the distortion comprises mainly low-order harmonics (fig.8), but AC-supply spuriae are also present (fig.9). However, as with the original Atlas, the amplifier's lack of linearity at high frequencies meant that it performed poorly on the high-frequency intermodulation test. Even at low powers (fig.10), the 1kHz difference component lay at a high 46dB (0.5%). Fig.10 also dramatically reveals the difference in the noise floor between the channels, the left (blue trace) having low-level power-supplyrelated spuriae extending up to the 24kHz limit of this graph, whereas the right channel (red) is much cleaner. It is important to note, however, that the AC supply-related spuriae in the left channel are still very low in level in absolute terms.
Erick Lichte very much liked the sound of Rogue Audio's Titan Atlas Magnum, especially with KT120 tubes. Though I must admit to some disappointment with the amplifier's lack of high-frequency linearity, in this respect it is no worse than the original EL34-equipped Atlas, which Fred Kaplan reviewed in March 2007.John Atkinson