Manufacturer's Specifications:
Power Output: 95 watts per channel, 8-ohm loads, 20 Hz to 20 kHz.
Rated THD: 0.1%.
Frequency Response: 1 Hz to 80 kHz, +0, -3 dB, at 1 to 95 watts rms.
Current Drive Capacity: 60 A, impulse current; 15 A, continuous current, no protection.
Reserve Energy Storage: 16 joules.
S/N Ratio: 90 dB, referred to rated output.
Input Impedance: 35 kilohms.
Input Sensitivity: 1.6 V for rated output, normal speaker output, non
Slew Rate: 70 V/µS.
Damping Factor: 250, from 20 Hz to 20 kHz.
Channel Separation: 70 dB. Minimum Output Load: 2 ohms.
Protection: A.C. line fuse, power supply rails fused separately for each channel, and output fuse; no V-I limiting.
Dimensions: 16 1/2 in. (41.9 cm) W x 8 in. (20.3 cm) H x 16 in. (40.6 cm) D; 19-in. EIA rack-mountable.
Weight: 55 lbs. (25 kg).
Price: $2,295.00.
Company Address: P.O. Box 2797, Livermore, Cal. 94550.
The Streets 950 is a solid-state power amplifier rated at 95 watts per channel into 8-ohm loads. Physically, the unit is large for a Class-AB design of its power rating. The large size is necessary to house the large heat-sinks and power supplies; this unit is unusual among power amplifiers in that it utilizes a fully regulated power supply for each channel.
The front subpanel is a beefy 0.225-inch thick piece of aluminum with another piece of 0.2-inch thick gold-anodized aluminum placed in front. A single, lighted power switch is the only control on the front panel.
On the rear panel are two gold-plated RCA signal-input connectors (each paralleled by a three-pin XLR female connector for potential professional use), three five-way binding posts per channel for speaker connections, a speaker fuse for each channel, the power-line fuse, a three wire a.c. power-cord connector, a cooling fan, a two-speed fan switch, and a pair of handles.
The unit is constructed around a U-shaped chassis that forms the bottom and sides. The sides of this piece have large rectangular cutouts that expose the heat-sink fins. The front subpanel and back panel are separate pieces that bolt on to the main chassis. Another U-shaped piece covers the top and sides, with slightly smaller openings in its sides rather than on the bottom chassis. The heat-sinks are quite large, having horizontal fins and some 1,400 square inches of radiating area in each sink. Outside air is drawn into the enclosure by the rear-panel tan and exits around the ends of the heat-sinks, where it passes across the fins and out the side openings.
A very large toroidal power transformer is bolted to the front subpanel. A full-wave bridge rectifier, mounted on the bottom of the main chassis, feeds two 25,000-µF, 75-V capacitors that form the basic ± 68.5 V unregulated supply for the input to the regulators. These large capacitors are mounted to the bottom in the center of the chassis. A large, thick copper plate is mounted on the top of the main filter capacitors to form a low-resistance ground connection for them and to connect the grounds of four more 6,400-µF, 60V capacitors that surround the main capacitors and are connected to the outputs of the voltage regulators.
Each channel's plus and minus voltage rails are separately fused and regulated, with the four fuses in two, dual fuse holders mounted atop the ground plane. These feed the regulator inputs via two thick wires in parallel for each connection, providing high current capacity without being unmanageably stiff. The regulator outputs come back to the four, 6,400-µF capacitors via insulated-braid buss wire. These four capacitors are all bypassed with 10-µF, 50-V film capacitors.
Mounted on each heat-sink are six output transistors, plus four additional power transistors used as series pass transistors for the regulators. A carrier p.c. board on the back surface of the heat-sinks serves to interconnect the power transistors and to mount emitter resistors, output chokes, etc. Two separate p.c. boards are mounted to each heat sink with standoffs. One is for the front-end and driver section of the power amplifier; the other is for the voltage regulator circuitry.
The Model 950 is beautifully constructed, and parts quality appears to be excellent.
Circuit Description
The overall circuit topology of the 950 is very similar to that of many other solid-state power amplifiers.
The signal input is coupled to the first stage through a 10-µF Wondercap polypropylene capacitor. A series resistor and shunt capacitor serve as a first-order, low-pass filter to limit the bandwidth into the first stage. A pot for overall d.c. offset zeroing is connected to the ground end of the input d.c. return resistor.
The first stage is a complementary, bipolar differential amplifier with current-source diodes and emitter-degeneration resistors. Output of the first stage is direct-coupled to the second stage, which is a complementary, Darlington connected, common-emitter amplifier. The collectors of this second stage are tied together through a two-transistor bias regulator which has its first transistor mounted on the heat sink for temperature sensing. Resistors from each collector of the second stage to ground control the open-loop gain and provide a lower source impedance to the output stage.
The output stage is a triple, complementary emitter-follower arrangement, with the first and second transistor pairs mounted on the front-end circuit board and the output transistors on the heat-sinks.
Overall negative feedback comes back to the inverting input of the input stage and is fully d.c.-coupled, without the usual shunt feedback-path series capacitor. This means that the low-frequency response only has one roll-off provided by the input network.
As previously mentioned, there are three binding posts per channel for output connections; one red for plus output, one black for normal speaker return, and a third, in white, for an alternate speaker return that provides some negative current feedback. The shunt feedback resistor returns to the white output terminal, which is connected to the black terminal through a piece of wire that serves as a current sampling resistor. The black terminal is returned to power-supply signal ground (more on this under "Measurements"). The idling current is set fairly high in this design-some 0.45 A, which causes an idling power dissipation of about 44 watts per channel. This allows for a theoretical Class-A power output of 3 to 4 watts per channel into 8-ohm loads before going into Class-AB operation.
The power-supply regulators are relatively conventional and consist of op-amp error amplifiers driving an emitter follower, followed by two paralleled Darlington power transistors that act as the series pass regulating elements.
Regulated output voltage is ±49 V d.c.
Measurements
The Model 950 was first run for one hour at 33% of rated power, or about 32 watts per channel into 8-ohm loads with a 1-kHz test signal. With the cooling fan on low speed, the heat-sinks got very hot but the amplifier didn't thermally shut down. With the fan on high speed, heat-sink temperature w2s at a lower, more desirable temperature.
Voltage gain was measured and found to be 17.3 x , or 24.8 dB into 8-ohm loads. The IHF sensitivity for 1 watt into 8-ohm loads at 1 kHz was 0.164 V. The IHF signal-to-noise ratio, which is A-weighted noise referred to 1-watt output into 8 ohms, measured-90 dB for the left channel and -92 dB for the right.
Figure 1 shows THD plus noise versus frequency and power output for 4and 8-ohm loads. The results for the right channel are plotted; it had slightly higher distortion than the left. Figure 2 shows THD versus power output at 1 kHz, with 400-Hz and 80-kHz filters switched in, and also shows SMPTE-IM distortion.
Crosstalk versus frequency was measured by driving one channel and measuring the resultant leakage in the other channel, with the unused input terminated by a 1-kilohm resistor and not connected to any external ground. It was found to be better than-90 dB from 20 Hz to 3 kHz, rising to -85 dB at 10 kHz, -81.4 dB at 20 kHz and peaking to-35.6 dB at 55 kHz. Connecting a clip lead between input grounds, which is more like using the amp in a normal stereo hookup, caused the crosstalk to deteriorate to -69.5 dB at 20 Hz, -66.0 dB at 1 kHz,-51.8 dB at 10 kHz, -46.7 dB at 20 kHz, and -19.2 dB at 55 kHz. These figures are for the right-to-left direction. The left-to-right direction was some 3 or 4 dB better. The peaking of the crosstalk at 55 kHz is symptomatic of a slight peak in the 950's frequency response, which also shows up as slight square-wave ringing and, of course, in the actual frequency response.
Figure 3 illustrates square-wave responses for the Model 950. The top trace is for 10 kHz into 8 ohms. The middle trace shows the effect of adding a 2-µF capacitor across the 8-ohm load. The bottom trace is for a 40-Hz frequency and shows very little tilt, indicative of excellent, extended low-frequency response.
The 1-watt frequency response into 8 ohms and 2-watt response into 4 ohms (both at 2.83-V output) are shown in Fig. 4.
Damping factor versus frequency, using the normal red and black output terminals, is plotted in Fig. 5. The effect of negative current feedback is to raise output impedance and reduce damping factor. When using the red and white output terminals, the output impedance at 1 kHz rose from about 0.046 to 0.145 ohm on the right channel and 0.038 to 0.132 ohm on the left. This corresponds to damping factors of 55 and 60 under these conditions.
Rise- and fall-times at ±5 V output were 3µS into 8 ohms and 4µS into 4 ohms. Furthermore, these response times were essentially constant right up to clipping, which is an excellent (and rare) characteristic for a power amplifier.
Dynamic headroom measured 0.22 dB coming from a burst onset of clipping power of 100 watts in relation to rated power of 95 watts into 8 ohms. Steady-state power at onset of clipping was also 100 watts, giving a clipping headroom of 0.22 dB. The small value of these numbers and their identity for dynamic and clipping conditions are due to the regulated power supply used in the 950. In a new test to determine some idea of peak output current obtainable from the 950, I set up a signal consisting of one cycle of 100 Hz, repeating at a 100-mS rate, and fed this into the amp, which was terminated in a 1-ohm load.
Under these conditions, with one channel driven (I didn't have another 1-ohm power resistor), the 950 produced about ±30 A, peak, before clipping. I'm impressed.
In a final test of the power-supply regulators, I found that the 950 would put out 100 watts per channel at a.c. line voltages down to 103 V. Below 103 V, the regulators become saturated, and their output voltage and obtainable output power begin to drop.
Fig. 1--Total harmonic distortion plus noise vs. frequency and power.
Fig. 2--Total harmonic distortion plus noise, and SMPTE IM, vs. power output.
Fig. 3--Square-wave responses at 10 kHz into 8 ohms (top); same, with 2-uF
capacitor across the load (middle), and at 40 Hz into 8 ohms (bottom).
Fig. 4--Frequency response at 1-watt output into 8 ohms and at 2-watt output
into 4 ohms.
Fig. 5--Damping factor vs. frequency.
Use and Listening Tests
Equipment used to evaluate the Streets 950 included an Infinity air-bearing turntable and arm with a Koetsu EMC-1B "Black" cartridge, GC/BHK and Audio Research SP-10 reference preamps, Quicksilver Mono tube and MLAS ML-9 solid-state power amps, Infinity RS IIA speakers, and Stax SR-X/Mk3 headphones.
The 950 was first listened to on the RS IIA speakers, using the Audio Research preamp. Bass definition and impact were excellent. The wallop certainly belied the amplifier's power rating. Compared to the Quicksilver and ML-9 amplifiers, there was more textural detail revealed, which made low-level sounds stand out more noticeably. Along with this effect was some additional brightness and a small bit of higher frequency edginess. But spatial replication and localization were judged to be less precise than with the other amps used, and, as beguiling as this increased apparent detail resolution was, the overall effect was a bit unmusical, with the other power amps giving a more believable overall musical presentation.
When I used the GC/BHK tube reference preamp to drive the 950 into Stax phones, the sound had excellent bass definition and quality. Mid- and high-frequency definition were good, again giving the feeling of being too good, resulting in a slight sense of artificiality. Overall feeling of space in the reproduction was one of less size than with the better tube power amps that I've used with this preamp and headphones.
The unit functioned flawlessly in testing and use and had no glitches or operational faults.
In summary, I feel the Streets Model 950 is a very well constructed and attractive power amplifier that should be reliable when driving difficult speaker loads. Regarding my comments on its sonic properties, I continue to urge prospective buyers to listen to whatever I've reviewed in as many situations (combinations of components and rooms) as possible.
-Bascom H. King
( Audio magazine, Jan. 1985)
Also see:
Sumo Model Nine Amplifier (Equip. Profile, Dec. 1983)
Sumo Athena Preamp (Aug. 1989)
= = = =