Ben Duncan brings the full might of modern technology to bear on the Soundcraft SA1000 Power Amplifier.

Soundcraft's brand new SA1000 is 3U high and 17" deep, into which it packs 465W stereo power into 4Ω, 288W into 8Ω; or 880W of mono, when driving 8Ω in the bridge mode. Altogether, there are four models in the SA series: the shortform data puts the SA1000 into perspective alongside the other three.

Rear View

Kicking off round the back for a change, the outputs are terminated with 2 sets of chunky Bulgin binding posts. These accept standard 4mm plugs, bare wires up to just 1mm², and at a pinch Hellermann insulated spades. It's a pinch, because access space will be restricted, once racked-up.

Input connections are easy: there's an 'A' gauge jack and an XLR, and either may be fed from balanced or un-balanced sources. For example, if a standard mono jack plug goes up the 'A' gauge jack socket, the input operates unbalanced as a matter of course because the mono jack's sleeve shorts the ring to ground, as required. Soundcraft's manual gives details on the whole cornucopia of wiring configurations, involving stereo versus mono jack plugs, and balanced versus unbalanced feeds. Focusing now on the input XLRs, it's at first sight disappointing to see them riveted in place. But wait: these are Neutrik types; the insert can be unscrewed and quickly replaced without bothering about the shell (which is unlikely to expire). It's also comforting to note that the XLR's polarity conforms to the UK/European standard, (pin 2 = hot). Comforting, because a lot of Soundcraft consoles have been wired pin 3 hot (obsolete since 1976) for the US market; except somehow, these 'pin 3 hot' variants still get loose in England.

Next stop is the groundlift; the signal ground and mains earth are brought out on adjacent 4mm terminals, where they can be optionally linked with a 5cm patch lead or length of tinned copper wire. Alas, the labelling isn't too hot, being a free translation from the Norwegian: 'Aud' means 'signal ground', while 'Chas.' refers to mains earth. As supplied, the SA1000 comes with a tinned copper link in place, but I wouldn't trust this in the long run, because the terminals will eventually shake loose, and the wire will oxidise, leading to all manner of erratic hums. If on installation, you find the system is quietest with the ground tied together, a short 4mm² patch lead is the surest choice for a firm and long term linkage.

The centre of the rear panel is occupied by a top quality fan, from Papst, the famous German 'luftwerke'. There's a facility for a clip-on filter, but this option is best skipped in studio racks, unless you're dead religious about cleaning the filter on a regular basis: a clogged-up airflow is noisy, useless for cooling, and strains the fan mechanism. Now just below the ground-link terminals, sits a fuseholder marked 'Fanfuse'. This is a tell-tale sign that the fan is energised separately from the power amp's guts. The significance of this is that it'll keep blowing cool air if the amplifier shuts itself down; or flames if one side blows up. This arrangement principally speeds recovery, if the sole calamity is overheating, and is good for stage or PA racks. Bridged operation is set with the usual recessed slide switch. Again, the legend is not altogether self-explanatory: 'STE' means normal stereo or tandem-mono operation, whereas 'Mono' means bridge-mode, and has nothing to do with the tandem input linking arrangement needed for PA work, as provided by makers like RSD or Rauch. On the plus side Soundcraft haven't forgotten to tell us which side of the amplifier is '+' in bridge-mode: that's when hooking across the red terminals on either side.

The SA1000's mains cable scores 100%: firstly it's sheathed in tough rubber, and secondly, the cross sectional area, at 2.5mm, is serious stuff, equating to a 20 Amp rating. Best of all, the budget cable-clamp has no problems gripping the cable: even if you swing all 60lbs of amplifier on it, it holds.

Control Panel

Looking from the front, the SA1000's aesthetics are undeniably brutal, as if it were designed for service in the SAS, Spetsnatz or some other unsavoury organisation. The massive, extruded rack handles look impressive, but the absence of a vertical member makes them useless for carrying the amp about. What's more, the edges aren't radiused: they're sharp, and likely to rip your hand apart if you slip.

Beginning with the channel attenuators, these are unsealed, but with a detent (Click stop) action. Looking now behind the panel, the pots are a standard type, but Soundcraft have taken the precaution of sleeving them in skin-tight plastic, to keep dust, grit and fumes from settling on the track. This makes sense when an enclosure is being blown (ie. fan cooled), and moreover, in view of the wiper's irregular movement, which aggravates the build up of grot over a period of years.

The illuminations comprise four recessed LEDs. Amongst these, there's just one green LED: marked 'Mono', it refers to bridged operation, according to the rear panel switch, as discussed. The remainder of the LEDs are red: two are channel clip indicators, the fourth, marked 'Temp' lights up if the amp overheats, and as such, it confirms the abrupt cessation of music. That's because the SA1000 plays safe, and disconnects its outputs when overheated. Additionally, its power supply switches off it the transformer overheats.

If you recall, there was no sign of a line fuse on the rear panel. Instead, Soundcraft hae opted for a resettable circuit breaker on the front panel, manifest as a dinky black button. Under overload conditions, the 'tit' pops out, and though slightly proud of the panel, and 'squashy' to the touch, it can't be tripped out by being pressed or poked.

SA1000: The Inside Story

Starting with topology, Soundcraft's first foray into power amplifiers embodies a radical (if bizarre) combination of MOSFET and bi-polar output devices. In essence, a pair of metal-can FETs drive twelve hefty bi-polar transistors. Granted that the pair of MOSFETs are the basis of an excellent output stage in themselves, hooking 12 bi-polar transistors across them makes little sense. But viewed from a fettered, if not foreign perspective, it's easier to get a good sound from massed ranks of bi-polars. Of course, this is a self-fulfilling prophecy, with the designer's opting for MOSFETs in metal cans, and cooling via an 'L-bracket' under these conditions, and when driven hard, any old bi-polar array will display a lower open-loop output impedance, to the possible benefit of the bass sound. Second, the hybrid combination does at least mollify the worst behaviour of bi-polar devices, namely their love of thermal runaway, alias accelerative pyrotechnics. In effect, the MOSFET devices sense the bi-polars' operating temperature (being mounted on the same L-bracket), leading to feedforward thermal compensation. To quote Soundcraft, one which 'anticipates thermal drift'. Up front, the input and pre-driver circuitry is also a radical departure from the norm. It's a hybrid assembly, comprising thick film and miniature surface mount components, packed onto a small, pluggable sub-PCB. Overall, the card's contents amount to a pair of fast, wideband analogue chips, better than an NE5534. In going to the trouble and expense of rolling-their-own IC, Soundcraft's rationale more likely relates to getting practical experience of surface-mount technology per se. Put another way, the hybrid card doesn't directly benefit the SA1000's cardinal parameters (size, weight, cost and sound quality), but is nevertheless a valid starting point for big changes in the future.

Inside, the SA1000 is initially impressive. Plus points include heavy gauge (2.5mm²) output wires and power feeds. Nodes (like the central ground point) are built onto chunky copper busses. Extensive use is made of latching Lucar connectors, that won't easily fall off, and bolt-down spade terminals. The single toroidal transformer is set inside a steel enclosure, which doubles as a heatsink for the bridge rectifier, and attenuates lateral magnetic fields, just to be sure that the adjacent channel cards and pots won't suffer hum pickup. For overseas use, mains voltage adjustment is evidently down to swopping over a Lucar blade, altering and/or cross-linking one of four terminations mounted in the top corner of the transformer's box. But the manual divulges no information on this, and furthermore, the terminals are themselves unlabelled.

Component quality is much higher than average; infamous names include ERO and Sprague capacitors, Spectrol-Reliance presets, and Motorola semiconductors. Access for service (a euphemism for repair) is also above average, but so it ought to be, with such a complex and heavyweight assembly! Most of the parts are visible with just the top cover plate removed. Also each channel can be removed wholesale. This includes not just the PCB, but one whole side of the amplifier. In other words, the extruded heatsink onto which the cards and power devices are built.

Having removed either channel, an unusually methodical checkout is feasible, thanks to Soundcraft's unique and thoughtful design: by removing straps (ie. link wires), the input and drive stages can be tested in isolation. This is a big change from conventional power amp cards, because with the DC levels (ie. tickover currents) being mutually dependent, cause and effect are frequently difficult to untangle. At the same time, the SA1000's individual DC bias circuitry adds considerably to the component count, which itself can prejudice reliability.

Turning to the L-brackets, the upper row of bi-polar power devices are easy to replace, being screwed in place, but the devices on the lower tier aren't accessible without first removing one of the L-brackets. In turn, these actually hold the heatsink together, which of course, is one whole side of the amplifier chassis. It's made up of three separate extrusions. Now although the mating surfaces appear to be machined dead flat, the extrusion tolerances are poor. The upshot is that even on our brand new factory sample, the brackets don't make contact with the heatsink sections over their full span. This leads to dodgy hot spots in operation, and an alignment nightmare if the assembly is ever dismantled. And the gaps can only get bigger if subjected to heavy G-forces, on the road. Equally, I can't see this bothering a careful studio owner.

Before we replace the lid, let's not forget the dollops of red gunge, indicative of screws which won't come undone. Further, the PCBs are solidly supported and printed legend identifies every small component.

Performance Tests

For the first time ever (in the UK, at least!), we've taken the liberty of running up the SA1000 on the world's most advanced test set, the Audio Precision. Introduced in November 85, it's designed expressly for professional audio measurements, and being designed by a group of ex-Tektronix engineers, you can be sure that the analogue and digital electronics attains unprecedented standards of accuracy. Thankfully, the machine skips knobs and dials altogether. Instead, the measurement parameters are set up and controlled by an IBM type PC. For example, Figure 1 shows the VDU panel whilst checking distortion on the SA1000. Reading left to right, Column 1 displays the generator settings, (waveshape, level, frequency, output mode, sweep increments and reference levels in dBr). Likewise, the centre panel describes what's coming out of the SA1000 (distortion level, power, output frequency, phase shift, and then the sensing parameters). On the right meanwhile, the panel presents the sweep and scaling parameters, leading to a performance graph, and thereafter, some hard copy - graphical printouts in other words).

Before going into the curves in detail, I must caution readers that our results differ from Soundcraft's own specifications. This is not unexpected. Anyone involved in the design, test, and evaluation of audio equipment knows that paper specifications only align with one's own measurements when Sun is conjunct Mercury in Virgo, or something like that. Across the board, the reasons for differences are essentially fourfold.

Firstly, the measurement conditions are different. For example, if the mains voltage isn't exactly 240 volts, then the amplifier's distortion will begin to rise steeply at a lower wattage than anticipated. Or maybe the source impedance of the signal generator is 600Ω, instead of 50. Or there's a ground loop at large.

Secondly, the test equipment used by the manufacturer (or reviewer) is out of alignment, or defective, or it's not being used properly.

Thirdly, the manufacturer could be quoting optimistic figures based on a handful of pre-production amplifiers, whereas in real life, production tolerances take the edge off the wonderful figures. Or something had to be changed to facilitate production, and this disrupts the printed specification.

Fourthly, the manufacturer's spec is a load of bull.

Granted that Soundcraft are a reputable manufacturer, and that we know every one of their products is individually tested, we may only conjecture that their brochure copywriters are out of touch with reality, or else the amplifier we tested is just a pre-production model, and that the full scale production run is liable to be tidied up. Finally, I must stress that if we'd tested any of the other amplifiers reviewed over the past two years in H&SR with the Audio Precision, there's no doubt that the results could not have been any less embarrassing, with the possible exception of Yamaha.

Lets kick off with Power Bandwidth. This defines the frequency response of the complete amplifier under real life conditions, ie. when loaded, with 4.3Ω in this instance. The x-axis plots the gain, in 0.2dB increments per division, with frequency on the y-axis, of course. Without too much squinting, we can discern that it's 33.35dB in the mid-band, (at 1 kHz). Looking right, it's dropped off to just over 33.0dB at 20kHz, that's -0.35dB down. If the power output is 504W in the mid-band, this spells just 465W at 20kHz, and likewise at 30Hz. Oddly, the SA1000 specification avoids any mention of power bandwidth.

The curve in Figure 3 displays distortion at 0.5dB below Clip. I'm pleased to report that the SA1000's clip LEDs are absolutely true indicators of overload, coming on more or less exactly ½dB before the waveform visibly crunches, and this is regardless of load impedance, mains voltage droop, or the drive frequency. Nevertheless, we cross-checked the -0.5dB point against the tell-tale 0dB point, where distortion curve for any feedback system turns on its axis, and shoots up like a rocket! Looking now at the curve itself, forget the figures for a moment, and instead, notice the shape: flat below 1 kHz, then winding up gradually at HF. Again, this is normal for any amplifier with feedback: that's to say very nearly every product in pro-audio. The gap between 40 and 70Hz is a data glitch, thanks to some 50Hz mains hum. So rather than printing out nonsense, the Audio Precision leaves us to interpolate. In this instance, we can reasonably assume a straight-line interpolation between these frequencies. Turning now to the brochure's figures, Soundcraft quote '<0.05% between 5Hz and 50kHz, at the rated power into 4Ω.' Our own measurement conditions differ favourably in that our load is slightly higher, at 4.30, and the test power is 503 watts (see the middle column in Figure 1), versus the 550W of rated power per channel. Again, with a little squinting, you'll see that the distortion curve crosses the 0.05% threshold at 10kHz, so we disagree with the SA1000's spec above 10kHz. In all fairness though, even the 2nd harmonic of 15kHz is unlikely to be directly audible: at 30kHz, you'd need to be a dog or bat to wince at it!

Figure 4 displays the Phase Response. Soundcraft quote 4° at 20Hz, and omit to give a figure at the top end. But so much is understandable, because phase shift is much more of an issue below 700Hz, were our stereo/binaural perception switches over to being phase sensitive. This time, the curve reads +20° at 20 Hz,and -25° at 20kHz. Neither figure is especially reprehensible, and we can only assume that the SA1000 spec was supposed to read 24°, because even with 100% direct-coupling, it's impractical to get LF phase shift much below 5° at 20Hz.

Figure 5 completes the set: this time, the balanced input's Common Mode Rejection (CMR) is under scrutiny. Once again, the general shape of the curve is a good sign of validity: CMR decays at -6dB per octave at HF, as the feedback runs out. The glitch at 15Hz is unexplained, but may relate to ¹/_f noise, the burbly, bubbly sound of random electron clusters. Given more time, one should ideally run 3 or 4 plots on top of one other, to develop an average or 'trend' curve. In the SA1000's spec, CMR is given blandly as 'greater than -70dB', which is nonsense, insofar as nature has always contrived to make CMR curves which vary with frequency. In fact, the curve displays a CMR of -60dB below 500Hz, and around -30dB at 20kHz. This represents an attenuation of 1000 fold at low frequencies, about 100 fold (-40dB) at 3kHz, and still a useful 30 times at HF. Also, the Audio Precision common mode measurement technique extracts true CMR, and doesn't, for instance, include the effect of RF filtration above 20kHz. It follows that whilst Soundcraft's figures simply aren't true, the performance is nevertheless consistent with the 'industry standard', and quite satisfactory in the real world.

The SA1000's HF response is average, viz. -0.1 dB down at 10kHz, and -3dB at 65kHz. What's more, Soundcraft's Frequency Response specification (-0.2dB at 20kHz) is, in fact, a power bandwidth measurement. So they did give it, after all! The DC offset voltage on both channels' outputs is low, at around 10mV, and more to the point, after running at high temperatures, it didn't drift upwards.

Last of all, we checked the thermal performance, with fan cooling, under steady drive conditions. With the SA1000 driving 130W of pink noise into an 8 ohm load (that's -3dB below clip, where the waste heat dissipation is generally highest, pro rata at around 50%), the power device bases rose to 85°c, while the external fins settled at 75°c. The 10°c difference represents a thermal resistance of around (^10°/_65w) = 0.15°c per watt, which is good. Under pulse conditions though, we could anticipate a figure nearer to 0.2 or ^0.25°c/_watt, typical of L-brackets. In conjunction with the thermal cutout, this is almost bound to prematurely curtail the power output under live rock'n' roll conditions. Again, in fairness, Soundcraft recommend subsidary fans in the rack to counteract this.

Listening Check

When we first fired up the SA1000, the sound was unacceptably thin and hard. With sound systems experience, this immediately suggests a polarity (or phase) error. Swapping over the polarity at just one of the outputs confirmed this: the bass end sprang into life. When a power amplifier has unbalanced inputs, this is usually down to checking the speaker leads' polarity, but whenever balanced inputs are involved, the possibilities for reversed polarity are multiplied. After checking the speaker's polarity (watch the cone move forward when the +ve terminal of a 1½ volt battery is applied to the speaker's red or + terminal), and checking that the connections to pins 2 and 3 weren't swapped over on either of the two input leads, we ended up reversing one of the output connections, concluding that one of the SA1000's balanced inputs is (internally) wired back-to-front. Beyond this circumstantial detail, the sound quality was promising, though it retained a sparse 'edge' in the mid-band on our Celestion SL6 reference.

We then cross-checked the SA1000 with Andy Dransfield and Bram, at 'The Chapel', which (incidentally), boasts a 24-channel Soundcraft desk. Again, there were unusual and unaccustomed 'problems' with speaker polarity. One of the Yamaha S20X nearfield monitors was initially wired out of phase, but this time, the SA1000 worked best with correct phasing. Even so, the difference was less marked than usual, and Andy judged the sound as hollow, cold and hard, even clinical. To an extent, the amplifier was merely emphasising the darker side of Japanese monitors. Nevertheless, the SA1000 distinctly seemed to emphasise the high mid, and possessed peculiar low mid-range qualities. Also, the fan cooling was even noisier than, quote 'The Lexicon or the 3M (multitrack), and we couldn't switch it off!'

On the plus side, everyone noticed the SA1000's punchiness; it had no trouble relating to the dynamics of first generation Rockabilly playback (Cotti and the Alleycats were in The Chapel at the time). These characteristics were confirmed on the main Eastmill monitors, albeit somewhat less strongly. This reinforces the fact that amplifiers can be almost as coloured as monitors themselves, and the combination needs careful choice. On this basis, the SA1000 would go down best with monitors possessing complimentary colouration, say models from ATC, Electrovoice or Tannoy, contrast the SL6 or JBLs, or Japanese monitors.

Summary

Despite the problems we encountered, the SA1000 shows promise as a useful studio monitoring amplifier, principally in view of its pulse power rating, and ability to drive into low impedances (ie. impedance dips) for short periods, without limiting. Equally, it's unsuited for any serious stage/PA work (pending some re-design work), on at least three counts. First, the output shuts down and distorts badly whenever the mains voltage marginally exceeds 240 volts, whereas surges up to 256 volts are both legal, and to be expected in UK touring. Second, the strength-to-weight ratio is poor, meaning the SA1000 would quickly fall apart on the road. For example, the unit weighs 60lbs, yet the vital rear support brackets are made of thin sheet steel, and are joined to the chassis with just a pair of tiny screws. Third, there's no facility for linking or daisy chaining the inputs, which creates some headaches in the racking-up department.

Against this, the SA1000 comes with a 19 page manual (albeit typewritten), and is well presented on the inside. Perhaps the clue to this dichotomy lies in the 'design by committee' syndrome, since the press release reads 'with the combined efforts of an international design team from Scandinavia, the UK and the US' I will leave you to judge for yourself.

The SA1000 costs £977.50, the SA600 £765.90, the SA200 £610.00 and the SA160 £603.75 including VAT.

Further information is available from Soundcraft Electronics, (Contact Details).