REVIEWS: 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 |

THE SOLITAIRE POWER AMPLIFIER

reviewed by Bascom King AUDIO magazine USA


I first encountered Kostas Metaxas at the 1992 Winter CES and was impressed by the man, by the range of products displayed in his company's demo room and by the sound of those products. Metaxas Audio Systems hails from Australia. They make a number of electronics pieces, including several preamps, three power amps and some interesting full-range electrostatic speakers. Other new products are in the making.

The Solitaire amplifier reviewed here is an unusual looking package rated at 130 watts per channel into 8-ohm loads and is said to be very wideband and fast. Some Metaxas literature claims a power output in excess of 100 watts per channel into any known speaker load, from d.c. to well over 500 kHz; I was reluctant to test that bandwidth spec until I had done the formal measurements and had my fill of listening!

The Solitaire is a near dual-mono design, with the common elements being the chassis and the power transformer. The overall layout ensures a very short signal path from input to output, including the current path from the power-transformer secondary windings and the rectifier and filter capacitors. The front-end circuitry consists of a complementary differential cascade arrangement feeding a complementary second stage that I usually call the "last voltage amplifier" (LVA). The differential input amplifiers don't use solid-state current sources but are connected to good old-fashioned resistors to the appropriate supply rails. The collectors of the LVA tied together through a bias-spreading regulator, feed a series of three Darlington-connected complementary emitter followers, of which the first two are drivers and the last is the actual output stage.

All the circuitry above, including the first output-stage driver is powered by complementary Darlington-connected capacitor-multiplier "regulators" fed from the main rectified power-supply rails. Capacitor multipliers divide the input d.c. voltage down by a small amount and bypass the lower arm of the divider with a large capacitor. This dividend and filtered voltage is fed to the base of an emitter follower, in this case a Darlington-connected one and the emitter is the multiplied output. This provides a low output impedance with the value of the capacitor connected from base to ground multiplied, in effect, by the beta of the transistor(s).

Back to the main signal path: A second complementary emitter follower drives four pairs of very fast (80-MHz gain-band-width product) output transistors. No Zobel (series RC) stabilising network across the output line to ground, or parallel RL output-buffering network in series with the hot output is used. The only stabilising elements, at least according to the schematic are a pair of small capacitors between the collectors and bases of the LVA transistors. A passive RC input filter reduces response above the designed bandwidth of the amplifier circuit itself. All devices are bipolar transistors. A claimed 11 dB of overall loop feedback is employed and an op-amp servo keeps overall d.c. offset to low values. Of course, the use of a d.c. servo negates the possibility of audio response to d.c., since servo circuits ultimately reduce the subsonic a.c. gain to a value less than what it is in the audio frequency range. Topologically, this circuit is a lot like other designs I've seen over the years, but to my knowledge, none of them were as fast as this design. Part of the reason for this wide power bandwidth is in the p.c. board design which borrows from r.f. and groundplane technology to maximise the speed of current delivery at high frequencies. A turn-on/turn-off time-delay and protection circuit operates a relay in series with the speaker terminals.

Another interesting circuit departure from convention is the use of a large number of small rectifier diodes in parallel instead of the customary single high-current bridge rectifier, which is slower.

Measurements

My first measurements were of gain (30.3 dB for the left channel, 30.2 dB for the right) and sensitivity (86.8 and 87.3 mV for the left and right channels respectively). Then I measured frequency response at an output level of 2.83V (equivalent to 1 watt into 8 ohms) for 8-ohm, 4-ohm and open circuit loads (Fig.1). As can be seen, the response is essentially unaffected by load.

Square-wave response for this amp is shown in Fig. 2. The top trace is for 200 kHz! I couldn't help using this much higher than usual square-wave frequency to illustrate that this amp is indeed fast. Measured rise and fall times were about 0.6 mS, giving an equivalent upper frequency response limit (+0, -3dB) of some 583 kHz. The little glitch on the negative-going transition is a fairly insignificant flaw; since its duration is some 200 to 300 nS, its energy is well out of the audio frequency band. With a more normal square-wave frequency of 10 or 20 kHz, the waveform would remain exponential even if I ran the level up quite a bit higher than the 10V peak to peak shown. (Remember, I didn't want to risk the possibility of rendering this amp unplayable as I hadn't fully assessed its sonic properties yet). In the middle trace of Fig.2, we have a 10-kHz square-wave frequency with an 8-ohm load paralleled by 2mF of capacitance.

Ringing here is typical of other transistor circuits and shows that even though the rise-time of the amp into a resistive load is blazing, it still can't deliver the current that fast into a high-frequency short (the capacitor) as evidenced by the slower rise-time in this trace. In the bottom trace, for 40Hz, there is some low-frequency roll-off (not much, mind you, but some), evidence that the response does not go down to d.c.

Figure 3 shows THD + N and SMPTE IM distortion as functions of power output and loading. (The 3-ampere fuses in the power-supply rails blew during my tests with 4-ohm loads, as the approximate average current per rail is about 3.8 amperes under those conditions. I therefore replaced them with 4-ampere fuses, despite the admonition in the owner's manual not to use fuse values larger than 3 amperes). Results are shown for the right channel, which was slightly higher in distortion than the left channel.

Total harmonic distortion, as a function of frequency and power for 8-ohm loads, is shown in Fig. 4. Even though the bandwidth of this amplifier is so high, distortion does start to rise in the audio range, as can be seen in the figure. The distortion a lower power levels was dominated by line harmonics rather than distortion per se; accordingly, I used a 400-Hz high-pass filter to cut off the measurements. Spectrum analysis (not shown) revealed a second-harmonic component of about 0.0025% in the right channel, with the rest of the spectrum mostly made up of line-harmonic components. The left channel, however, had much lower noise, so more discrete distortion components could be seen, including a decreasing series down to the sixth harmonic.

Out put noise levels are enumerated in Table 1. Noise levels are dominated by line-harmonic (hum) components at a fundamental frequency of 120Hz. The waveshape of these hum components looks like the effect of charging current pulses in the main filter or smaller bypass capacitors tied to the main supply line. In my opinion, the left channel was only marginally quiet and the noise in the right channel was outright too much.

Crosstalk between channels was found to be down more than 90 dB over the audio range in the left-to-right direction, with the exception of the aforementioned hum components (which aren't crosstalk per se but do show up in the measurement). In the right-to-left direction, crosstalk was again better than 90 dB down, including the hum components, but started to rise at about 1 kHz, crossing over the -90dB point at 3 kHz and ending up at -83 dB at 20 kHz.

Damping factor for both channels is plotted in Fig. 5. The higher damping factor is for the left channel. Notice the dip at 120 Hz for the right channel. That is the 120 Hz fundamental hum component showing up in this measurement. The basic technique in measuring damping factor is to inject 1 ampere of current into the measured channel's output terminals, with the signal input jack terminated by 1 kilohm, and then measure the voltage appearing across the output terminals. Since the injected current is 1 ampere, the measured voltage across the output is directly convertible to ohms of output impedance. However, any noise or hum in the output of this measured but undriven channel can raise the voltage measured in this test, contaminating the results. Since damping factor is really 8 ohms divided by the measured output impedance, a small rise in level at hum frequencies will appear as an apparent dip in damping factor.

In dynamic power testing, the Solitaire's dynamic clipping power was found to be 150 watts into 8 ohms, for a dynamic clipping headroom of 0.62 dB. Into 4-ohm loads, the dynamic clipping power at the visual onset of clipping was 288 watts. Steady-state clipping power at the visual onset of clipping was 134 and 220 watts into 8- and 4-ohm loads, respectively. This yields a clipping headroom of 0.13 dB for 8-ohm loading.

The Solitaire's d.c. offset measured less than 1 mV in each channel. The a.c. line current at idle was 0.72 ampere, indicating a moderate idling power dissipation that gets the heat-sinks warm to the touch. Power-supply rails were + 55.3 V for a 120- V a.c. line input.

After a thorough listening evaluation, I returned the amp to my lab to look a little closer into its ultrasonic power capabilities. Driving a 20-kHz signal into clipping with 8-ohm loads showed some signs of "sticking". A sweep of distortion versus power output with 8-ohm loads and a 200-kHz test signal (now shown) revealed that distortion stayed under about 0.6% for both channels up to 100 watts, and visual onset of clipping occurred at about 120 watts. At 400 kHz, again with 8-ohm loading, I got about 100 watts per channel at the onset of clipping. Finally, at 500 kHz, I could squeeze about 80 watts per channel out, through the waveform was starting to look more triangular than sinusoidal. Although this doesn't quite meet the Solitaire's spec for high-frequency power, what the amp does do is pretty damned impressive, if you ask me! Definitely not something to try on most other solid-state amps!

Use and Listening Tests

Ancillary equipment used to evaluate the sonic properties of the Metaxas Solitaire power amp included the following: An Oracle turntable fitted with a Well Tempered Arm and Spectral Audio MCR-1 Select moving-coil cartridge, a Krell Digital MD-1 CD transport feeding a PS Audio UltraLink or Counterpoint DA10 D/A converter, a Nakamichi 250 cassette recorder and an ST-7 tuner and a Technics 1500 open-reel recorder. Preamplifiers used included a Counterpoint SA-5000 and First Sound Reference II. Other power amplifiers used were a Crown Macro Reference and a pair of Quicksilver M135 mono tune prototypes. Loudspeakers used were Win Research SM-10 monitors, early Genesis Technologies two-way prototypes and Scientific Fidelity Joules.

The Solitaire is yet another solid-state amplifier that I liked from the first time I heard it in my system. It passed very musical and unharsh sounds through to the speakers. Its sound is characterised by exquisite spatial presentation, solid dynamics, great transparency and a tonality that is a little soft-sounding in the high frequencies. This amp is lyrical and quick sounding. Some of my favourite software that might be a bit edgy on other otherwise good amps sounded smooth and less irritating on the Solitaire yet had great definition.

I had two nit-picks about the Solitaire: It made a very audible pop some 5 or 10 S after the power switch was turned off, and the hum in the right channel was definitely audible at the loudspeaker and just discernible at the listening position when no signal was going through the system. In practice, however, this hum level is not going to be a problem except with very efficient loudspeakers; it didn't really bother me.

Overall though, I enjoyed having the Metaxas Solitaire in my system and listened to a lot of music through it. I forgive its trivial technical flaws for its wonderful music transference. A BHK thumbs-up for this one.

Bascom H. King

 


home!