Pass Labs X1000 monoblock power amplifier
Pass's Supersymmetry Balanced Single-Ended class-A topology was actually granted a US patent in 1994. Using a balanced gain-stage, the distortion and noise are made absolutely symmetric at the two outputs. The signal is amplified, while noise and distortion are canceled. In a nutshell, this is achieved by precise matching of the two halves of the circuit and "cross-coupling" their internal drive signals. (See the "Supersymmetry Explained" Sidebar for details.)
Symmetrical building blocks
The X1000's power supply begins with a pair of big-boy toroidal transformers, individually rated at 1500VA continuous. From the paperwork, I gather that under actual conditions they'll do about 2000VA continuous per transformer, and at least half that again for short periods. Each transformer has its own circuit breaker and rectifier bridge. The breaker current is set at 20 amps, although the breakers will allow transients on the order of 300A to pass. To avoid a huge inrush of current during charge-up, all four transformer primaries are equipped with inrush suppressors, limiting current to 100A or so.
The transformers are potted in steel cans to minimize mechanical and magnetic noise, and the balance of the amplifier is crafted mostly of aluminum. There are eight computer-grade capacitors (large, old-style cans) at 25,000µF and 75V, making up over 240,000µF of storage capacitance (700 Joules of stored energy). These are used to create the unregulated output rails, at ±75V at 30 amps. A smaller transformer generates the ±99V rail voltages for the front-end circuitry. The voltage rails are passively filtered to produce a ripple voltage of only a few millivolts, "none of which shows up at the output." In his humorous and very engaging manual, Nelson Pass goes on to explain that the extra front-end supply lowers the system's distortion and noise and allows it to "swing the output stage rail-to-rail with losses on the order of only a volt or so, extracting every last possible watt."
The "balanced single-ended" front-end gain stage comprises a differential pair of MOSFETs with local negative feedback. They're biased by a constant-current source and cascoded for best effect. This stage develops the voltage gain and presents it to a large bank of follower MOSFET power transistors. "Originally it was assumed we would have to enclose this output stage in a Supersymmetry loop to get the performance we wanted, but ultimately we found that we could operate it without feedback as long as we put a healthy bias current through it. For the X1000, that's about 500W of dissipation per channel."
The 80 (!) output power MOSFETs are actually HEXFETs from International Rectifier in TO-3 metal cans, each taken from the same lot code (indicating the same wafer) and "hyper-matched" to 0.5% gate voltages. "The output stage can sustain transients of about 12,000W, but it's never allowed to dissipate more than 2400W for any instant, even into a dead short." Most of the front-end transistors, the current sources, and the cascode devices are rated at 200V and 150W, but are run at a low 2.5W each. "The speed-critical gain devices in the front-end—the actual balanced pair of transistors—are rated at 20W, and we run them at 1W each on the same heatsink for perfect thermal tracking."
The circuit is completely DC-coupled, with no capacitors in the signal path. Two film capacitors stabilize the constant-current sources and two more caps filter the Zener diode voltage references, "but that's it!"
Power from the supply caps to the output transistors is conducted through formed sheets of aluminum with a large surface area, the output devices bolted directly to them. Wiring from the transistors to the speaker terminals consists of 24 parallel runs of 10-gauge copper cable. As you'd expect, all PC boards are double-sided, with plated-through holes and double-thickness copper traces.
Each monoblock idles at about 600W, or 1200W for the pair—about the same as a pro-size hair dryer. The X1000 is biased for class-A operation up to 128W peak, beyond which it drops into class-AB. The amplifier typically draws about 5A (continuous RMS) from the wall during normal use, reflecting the idle current running through its output stage. If you're driving a low-impedance load hard it'll draw more, but both channels can be run on a single 20A, 120V AC line "without problems in most cases."
If you're truly power-mad, a number of X1000s can be arrayed to produce up to 2000W per chassis. Two of them can deliver 4000W into 8 or 2 ohms, four can deliver 8000W into 4 ohms, and 16 will do 32,000W into 1 ohm! The mind boggles. "As a practical matter, we would expect peak output levels on the order of 64,000W for such arrays." Yes, as a practical matter. Ahem. Your dealer or Pass can supply you with the interface adapter(s) needed to run the amps in series or parallel.
Each amplifier packs a detachable 20A IEC power-cord with a plug that's larger than the standard 15A item found on many components these days. The IEC-320 power cord (hot blade vertical, neutral blade horizontal) requires a modern, high-current NEMA 5-20 wall socket, so you might have to call an electrician. The amplifier's chassis is connected to ground only through the power cord. It is not attached to the circuit or to the amplifier signal ground. "Under no circumstances should you defeat the ground connection of the power cord. For your safety, the chassis of the amplifier should be earth-grounded. We aren't kidding." He isn't kidding.
The X1000 accepts only a balanced input with pin 1 ground, pin 2 positive, and pin 3 negative. "This is the first amplifier that truly demands balanced operation to obtain high performance. If you don't have a true balanced source, you won't break the amplifier, but you're not getting anywhere near the performance you paid for." The rear panel houses a power breaker switch, two pairs of hefty paralleled output binding posts, and a 5-way connector for remote turn-on.
Flicking the breaker switch on the back panel lights a small blue LED in the meter and keeps the front-end warmed up and ready for use. When the front panel's Operate button is pressed, the eerie blue meter light is activated and the indicator runs up to something just below half-scale. As the amp warms up, the meter drops back a bit; it shows the level of bias current passing through the output stage. No relay, no delay; you're off, then you're on. If the meters bounce at a moderate or low listening level, then you're probably driving a very low impedance load, or perhaps a short. "If your speakers are not known to be very low impedance, say less than 4 ohms, then you should check this out." Yes, do. Basically, the X1000 operates in pure class-A at current levels that don't bounce the meters. Then this gem: "Do everybody a favor and try not to have shorted output cables. It happens accidentally all the time, and the amplifier is designed to survive, but I wouldn't bet the farm on it." He's a riot.
It took about an hour of warm-up for the amp to sound its considerable best. For the record, the blue-eyed beasts got mighty toasty after an evening of frolicking-good musical fun.
The X1000 sounded extraordinary in our system. Yes, the fanatical attention to detail and simplicity of design made for a very refined sound. But what continually struck me was the sense of endless power that seemed rooted way down in mother earth's core. It was essentially different from what I'm accustomed to hearing from smaller amps—those that manage to keep their composure while running high, wide, and hard. The X1000's authority in the bass was comprehensive yet effortless. The tremendous solidity and controlled power delivery made for a stable foundation upon which the music built.