Any THX-certified power amps sell at a premium over amps that would otherwise
seem comparable based on conventional specs. Is the premium based on what the
market will bear? Is it due to Lucasfilm licensing fees? Or does it cost more
to make a THX amplifier? Perhaps it's a combination of all three. I'll leave
it to you to decide, but, in point of fact, Lucasfilm does impose stringent
requirements for THX certification of power amps, of which the most serious
relates to power rating.
To meet Lucasfilm requirements, a power amp must put out a minimum of 28.28 V rms. With 8-ohm loads, that's 100 watts per channel. Although this would seem to be a relatively modest power rating nowadays, Lucasfilm insists that the same output voltage be delivered into lower impedances as well. For example, a two-channel THX power amp like the Soundstream DA-2 must put out a minimum of 28.28 V into 3.2-ohm loads. That's 250 watts per channel! Three- and six-channel THX amps get off a little easier as far as minimum load is concerned, but the bottom line remains the same: A THX power amplifier usually has a larger power supply than the average stereo amp.
Consider a two-channel power amplifier rated at 100 watts per channel into 8 ohms. In theory, a non-THX amp could get by with a power supply capable of delivering just 200 watts for the loads plus perhaps 50% more for circuit inefficiency. But a THX-certified amplifier of the same rating must have a supply capable of delivering 500 watts (with 3.2-ohm loads) plus an even wider safety mar gin for the greater loss in the output stage incurred when operating into a low impedance. I'm not saying here that well-designed, non-THX amplifiers shave things so closely; most deliver more power into lower impedances than they do into 8 ohms, which implies a sup ply capable of delivering that power. But most stereo amplifiers do not carry matters quite so far as Lucasfilm requires.
Beefy power supplies are costly, especially when designed in the conventional manner, i.e., a line-operated power transformer, rectifiers, and filters. To handle high power at the 60-Hz line frequency requires a massive power transformer to prevent core saturation and huge filter capacitors to maintain the supply voltage between successive charge cycles. Soundstream sidesteps the problem in the DA-2 by using a high-frequency switching power supply.
The DA-2's supply operates at approximately 60 kHz, so it recharges the filter capacitors about a thousand times more of ten than a line-operated, 60-Hz supply.
Microfarad for microfarad, the effective ness of the filter bank is multiplied accordingly, although the capacitors must be specially designed to handle high-level, high-frequency ripple currents. Further more, at 60 kHz a relatively small ferrite-core toroid can handle the flux required to supply adequate power without saturating.
Thanks to this switching-mode supply, Soundstream manages to cram a 200-watt-per-channel THX-certified stereo power amp into a package only PA inches high.
Three DA-2s would stack up at about 6 inches and power a six-channel THX home-theater system at 200 watts per channel into 8 ohms, 350 watts per channel into 4 ohms, and a whopping 500 watts per channel into 2-ohm loads. Further, each DA-2 can be bridged for mono operation at double the per-channel power into twice the impedance (400 watts with 16-ohm loads, 700 watts with 8-ohm loads, and 1,000 watts with 4-ohm loads).
Switching supplies have another advantage when used in high-power amplifiers: A relatively low turn-on surge.
When a conventionally powered amp is turned on, there's a huge in rush of current as the magnetic field builds up in the transformer and capacitors take an initial the large filter charge. The actual surge depends on the precise point in the line voltage cycle at g which the power switch makes contact but can be so large that it trips the circuit breaker. That's especially likely if multiple high-power amplifiers are plugged into the same line and turned on simultaneously.
----------
Fig. 1--Frequency response at 1 watt out.
Fig. 2--Noise vs. frequency.
Fig. 3-THD + N vs. frequency for 8-ohm loads in stereo (A) and mono (B), and for 4-ohm loads (C; see text).
-----------------
Switching-mode supplies can be designed to ramp up slowly to avoid the turn-on surge, as Soundstream has done. This is not to say that you can have three DA-2s plugged into the same house line, all delivering 500 watts per channel simultaneously; that's almost certain to flip any home breaker. But if you're using a 20-ampere line, you should be able to turn on three at once (they're rated for a 6-ampere turn-on surge), and once the amps are on, it's unlikely that you will require all channels to operate flat-out simultaneously over any appreciable time period.
Control Layout
The DA-2 features both balanced and unbalanced inputs via XLR and gold-plated RCA jacks, respectively, at each end of the rear panel. Be tween each set of inputs is a push button that selects between them ("Bal/Unbal"); a second pushbutton, on the "Channel 2" input at the left of the back panel, selects be tween "Mono" (bridged) and "Stereo" operation.
Centered on the back panel is a circuit breaker, with an IEC three-wire jack for a removable line cord to its left and output connectors to its right. The latter are gold-plated, heavy-duty five-way binding posts spaced to accommodate "GR" type dual plugs. What's noteworthy about them, however, is that their through-holes will accommodate 6 gauge wire. Between the circuit breaker and the output connectors is an "Accessory Link" hookup, which permits the Soundstream C.2THX preamplifier (or any other preamplifier using a 5-V remote trigger signal) to turn on one or more DA-2 amplifiers.
The front panel has only a recessed power indicator and a "Power" switch that overrides the remote turn-on. (The switch is left in the out position when you are using the remote hookup.) When first plugged in (or when power is re stored), the Soundstream DA-2 goes through an initialization sequence that can take up to a minute. If the amp is left on standby, the turn-on delay is much shorter.
Except for a pattern of small ventilation slots at the far left and right, the top and bottom plates are solid-presumably to help contain the powerful 60-kHz signal generated by the supply. Heat from the output stage is dissipated by finned extrusions that extend to the left and right of the main chassis. Thermal sensors, mounted on each sink, shut the amplifier down in case of overheating. The switching power transistors appear to use the amplifier's bottom plate as a heat-sink.
The output transistors are clamped to the heat-sinks with a mounting bar rather ...
-----------------
SPECS
Power Output: 200 watts/channel, 20 Hz to 20 kHz, into 8-ohm loads, at less than 0.1% THD with both channels driven; at 1 kHz, 350 watts/channel into 4 ohms, 500 watts/channel into 2 ohms; maximum single-channel output at 1% THD and 1 kHz, 240 watts into 8 ohms, 400 watts into 4 ohms, and 500 watts into 2 ohms; bridged mono, 400 watts into 16 ohms, 700 watts into 8 ohms, and 1,000 watts into 4 ohms.
Frequency Response: 20 Hz to 20 kHz, +0,-0.3 dB; 10 Hz to 45 kHz, +0,-3.0 dB.
S/N: 110 dB, A-weighted, re: 200 watts into 8 ohms.
THD: Less than 0.1%, 20 Hz to 20 kHz, from output of 1 to 200 watts into 8-ohm loads.
Input Impedance: 11.5 kilohms.
Slew Rate: Greater than 30 V/µS.
Damping Factor: Greater than 500.
Average Power Requirements: 120 V a.c., 400 VA.
Dimensions: 17 1/2 in. W x 1 3/4 in. H x 15 in. D (44.5 cm x 4.5 cm x 38.1 cm).
Weight: 11 lbs. (5 kg).
Price: $1,195.
Company Address: 120 Blue Ravine Rd., Folsom, Cal. 95630.
-------------
Fig. 4--THD + N vs. output for stereo operation into 8 ohms (A), for bridged
mono into 8 ohms (B), and for 4-ohm loads (C; see text).
.... than being individually mounted with screws through the transistor tabs. This arrangement helps ensure better contact between the junction area of the device and the heat-sink.
Circuitry
The layout is neat, and component quality appears good. The power supply occupies its own board, which is centrally located within the case and shielded from the audio boards that flank it. Printed-circuit traces are wide where that's needed for cur rent handling, with point-to-point wiring used only where necessary. Four 1,200-uF, 180-V electrolytics serve as the main power-supply filter. Soundstream claims that because of the high frequency at which the power supply operates, these filters have the equivalent storage capacitance of a 220,000-µF bank at 60 Hz. The, main toroidal switching transformer appears to be hand-wound with a multifilar winding.
The power line has r.f. filters to reduce 60-kHz energy flow into the house wiring.
No schematic was available for reference during my testing, but it seems apparent that the DA-2 employs output coils for protection against wayward loads. Soundstream claims, however, not to use current limiting in the output stage, relying instead on a circuit that senses the load and shuts down the system if the load impedance drops below 0.5 ohm.
The output stage operates in what is now known as Class H. For low to medium signal levels, the output devices are powered from relatively low-voltage rails; as the signal level rises and approaches the low-voltage rails, the voltage is in creased to stay above the signal until, ultimately, full supply voltage is reached and the amplifier clips.
This technique provides greater efficiency in the output stage during normal operation, reducing heat generation.
Soundstream prefers to keep the details of its Harmonic Phase Correction circuit close to its chest, but I was able to gather that the system is intended to keep distortion-generated harmonics in phase with the fundamental over a wider portion of the audio band than is typical with other designs.
This is said to mimic the phase relation ships that occur in the harmonic structure of musical instruments and thereby render distortion less audible.
Measurements
Switching supplies are two-edged swords; they eliminate the need for a large transformer and filter bank, but they potentially generate huge amounts of high-frequency noise, which must be kept from contaminating the music directly or via intermodulation. The DA-2 employs input filters in each channel to reduce these effects. The filters have an initial slope of 18 dB per octave and a cutoff frequency of 46 kHz. In the audio band, response is +0.0,-0.2 dB from about 18 Hz to 20 kHz, which meets THX requirements. Actual response is shown in Fig. 1 for both stereo and bridged operation. These curves were taken on the left channel, using the unbalanced input, but the response from the balanced input was the same, and the result for the right channel was so close to the left's that there was no point plotting it.
A small amount of 55-kHz switching noise can be seen in Fig. 2, but it's more than 85 dB below 1 watt (except in bridged mode) and, in my opinion, negligible.
However, I did find it impossible to listen to weak FM stations on my lab tuner when operating the DA-2 near it. In fairness, the problem was mainly on weak stations, and I was using an indoor dipole. If you're in a stronger reception area and/or use a good outdoor antenna and a shielded downlead, you'll undoubtedly be in better shape.
Figure 3 shows THD + N versus frequency. Figure 3A depicts stereo operation with 8-ohm loads at output levels of 1, 10, and 200 watts per channel. At low levels, right-channel THD + N (dashed curve) is below left-channel distortion (solid curve).
But the roles reverse at rated power, where left-channel distortion reaches a maximum of 0.03% at 8.5 kHz, while right-channel distortion continues to rise to 0.05% at 20 kHz-still only half that allowed by Sound stream's specification. At clipping, the DA 2 delivered 235 watts per channel (23.7 dBW) into 8 ohms and, with the IHF tone burst, managed 290 watts (24.6 dBW) of dynamic power a side into 8 ohms, for a dynamic headroom of 1.6 dB. Comparable results for 4-ohm loads were 455 watts per channel and 1.1 dB, respectively, and for bridged mono operation were 925 watts and 1.2 dB. However, the line voltage may have dipped during the 20-mS tone burst without registering on my slow-responding live-voltage meter, in which case this data may understate the DA-2's capability.
Bridged for mono with an 8-ohm load (Fig. 3B), THD + N just tops 0.15% at 500 watts, the maximum output at which I was able to maintain a 120-V line. The jagged nature of the 600-watt curve reflects the onset of clipping as the Audio Precision test gear attempted to maintain 600 watts with a line voltage that had sagged below the reference. I calculate that, had it been possible to maintain a 120-V power source, output at clipping for these conditions would have been 820 watts (29.1 dBW).
Similarly, it was impossible to maintain the correct line voltage with stereo operation
Fig. 5-Interchannel crosstalk for balanced input at 10 watts out (A) and
for unbalanced output at 10 and 100 watts (B).
... into 4-ohm loads at Soundstream's rated power of 350 watts per channel. Figure 3C therefore shows the results in stereo at 1, 10, and 300 watts per channel, where the line voltage could be maintained (again, with solid curves for the left channel and dashed curves for the right), but the 350-watt results are in mono, to maintain the proper line voltage. I expect the mono curve would be representative of two-channel operation for those whose power company is more generous than mine. The THD + N, which remains under 0.08% under all conditions, easily betters Soundstream's claim, and maximum out put level (450 watts, or 26.5 dBW) exceeds spec by a decent margin.
The distortion curves of Figs. 4A, 4B, and 4C depict THD + N versus output using test frequencies of 20 Hz, 1 kHz, and 20 kHz. Because of the difficulty in maintaining proper line voltage at the power levels that this amplifier is capable of producing, I ran the 4-ohm curves with one channel driven and repeated the 1-kHz measurement with both channels operating.
The anomalies in the shape of these curves around 45 watts (stereo operation into 8 ohms), 150 watts (mono operation into 8 ohms), and 75 watts (stereo operation into 4 ohms) probably correspond to the point at which the output stage switches from the low-voltage rails to the high-voltage rails.
Needless to say, the transition could be more seamless and may account for some of the characteristics I noted during my listening tests, but the distortion through the transition point is still relatively modest.
I also noticed what appears to be power-supply switching noise riding on the positive-going portion of the signal at particular output levels, which also may correspond to the anomalous region. Soundstream is convinced that these anomalies are an artifact of the measurement process, caused by switching noise entering the ground lines when one device (the Audio Precision test equipment) is connected to both in put and output. I take extreme pre cautions to prevent ground loops in my test setup, but I cannot guarantee that there isn't enough stray capacitance between the Audio Precision's input and output that some high-frequency switching noise wouldn't leak through. If Soundstream is correct and the problem is one of leakage, it should not occur when driving loudspeakers.
The final set of curves, Figs. 5A and 5B, shows crosstalk versus frequency. The data is quite good from the balanced input but rather odd from the unbalanced input, where left-to-right crosstalk is substantially greater than right-to-left crosstalk and, furthermore, varies with signal level.
THE DA-2'S CLARITY AND IMAGE DEPTH THROUGH THE MIDRANGE WERE OUTSTANDING.
Sensitivity (100 mV for 0 dBW from the unbalanced or balanced input) is right on the THX spec; sensitivity for bridged mono is 51 mV. Output impedance is, at 0.22 ohm or less across the audio band, within a gnat's whisker (and experimental error) of THX requirements; the resultant damping factor is 310. Channel balance is near perfect (0.04 dB); input impedance (10.7 kilohms, balanced or unbalanced) is fine. The A-weighted noise was -90.6 dBW for the un balanced input, -91.2 dBW for the balanced input, and -82.2 dBW for bridged operation with the unbalanced input; from a technical standpoint, better A-weighted noise figures than these could be expected.
Use and Listening Tests
Despite my demurral about measured noise, the Soundstream DA-2 was certainly not noisy in the listening room. At normal listening distance, no trace of electronic noise could be heard, and, even with my ear close to the speaker, electronic noise was barely discernible.
The immediate characteristic that struck me when listening to the DA-2 amplifier was its remarkably solid and powerful bass.
Although its response is no flatter or more extended than that of other good power amps, it seemed as if it were plumbing depths yet unfathomed; the results were very gratifying.
The second quality to which I reacted was the DA-2's outstanding clarity and image depth through the midrange. This was particularly noticeable at low and moderate listening levels, both on solo classical instruments (harpsichord, piano, and violin) and on voice. On louder passages, however, the sound thickened perceptibly on choral passages and became more strident when reproducing stringed instruments, whether struck, plucked, or bowed. Interestingly, the low bass remained solid and clean at all levels, and, when really pumping out the watts, the amplifier seemed to regain its aplomb and sound better over the entire frequency range than it did at the middling level.
The Soundstream DA-2 packs a mighty wallop into a minuscule package. Under lab-test conditions, it gets warm and draws substantial current from the power line, but in normal use, it's efficient and runs cool as a cucumber. I really do believe you could stack three of these amps, power them from a single house line, and provide a few kilowatts of on-demand, short-term power to your home theater setup as conditions demand. That's a lot to say for this mighty mini.
-Edward J. Foster
(Source: Audio magazine, Nov. 1994)
Also see:
Snell Type B Speaker (Sept. 1992)
Vandersteen 2Ci Speaker (Equip. Profile, Jun. 1992)
= = = =