Nelson Pass On The Patents Of Pass
Pass's influence has now spanned three decades and shows no signs of stopping. Card-carrying audio junkies and the more-power-is-always-better crowd have long lusted after his megawatt beauties. The flea-power amp and open-baffle types drool over the Zen models of his First Watt line, and the true hard core, the DIY crowd, refer to him as "Papa" as they eagerly absorb his donated designs, wisdom, and general good humor.
Nelson Pass is many things, but most of all he's an engineer. He views the world as a puzzle to be solved, and immediately after figuring out how something works, he begins thinking about how to make it work better. He's approached audio design with this combination of curiosity and pragmatism, and the result has been a string of innovative, often brilliant designs, many of which have been based on technologies that Pass has patented. I tried to follow the evolution of his designs through these patents but was quickly buried in legalese, so I asked Pass to walk me through them chronologically. The result turned out to be a mini-history of his career as an audio designer.Brian Damkroger
1976US Patent 3995228: Active bias circuit for operating push-pull amplifiers in class-A mode: [Pass's first patent describes what's often referred to as "sliding bias," where the bias on the output devices varies with the signal to prevent them from shutting off, ergo avoiding crossover distortion. ]
There was a lot of interest in class-A operation, but there were practical issues, mainly their size and inefficiency. The maximum power output was theoretically limited to twice the bias current, and really more like half that. I modulated the bias to stretch the bend in the operating curve. Technically, this kept it operating in class-A, but at a lower bias level.
The design was successful; the amps sold like gangbusters and were copied immediately. Unfortunately, the approach ended up getting a bad reputation. Our first amplifier, the Threshold 800A, used about a 1:1 ratio, so that it idled at 150W for 150W output instead of 300W. That seemed reasonable and worked well, but some people took it to an extreme. Instead of idling at 150W, there were 200Wpc amps that idled at 10W, ran cool, and didn't have any heatsinksall claiming to be class-A. In retrospect, we could have sued the copycats, but we were young and foolish, and besides, I had another design waiting in the wings.
1978US Patent 4107619: Constant-voltage/constant-current high-fidelity amplifier: The patent application actually shows the circuit of the [Threshold] Stasis 1 as the example circuit. This is a kissing cousin to the Quad 405, injecting current into the output for error correction, but with way more current. We had a massive power supply and a huge bank of power devices that we used to support a smaller output-voltage source.
All distortions are variations in a device's characteristic with changes in voltage and current. Ideally, you want to lock the operating point of your gain device, the voltage and current, at a constant value, thereby eliminating distortion. Cascoding covers locking the voltage, so for the current, I devised a "current bootstrap," an external current source in parallel with the gain stage. It responded to the current going out and sourced current to the load, outside of the loop gain path. The result was to keep the output current very nearly constant, which dramatically lowered distortion, so we could operate without feedback. These were great amps. They sold for years and years. We licensed it to Nakamichi. They're still great amps.
1988US Patent 4752745: Opto-isolated bias circuit for operating push-pull amplifiers in class-A and class-AB modes: This one was optically coupling current sensing to the bias circuitry, which locked it nicely. Typically, bias drifts over time, and with changes in temperature. Even if it's perfect, the amp heats up and things shift. This was a way to look at the current and couple it back to the bias circuit for a constant bias. Nowadays, we just let the amp warm up and settle in. One hour is pretty good, two hours is better.
1990US Patent 4899387: Active low-frequency acoustic resonance suppressor: [This patent isn't for an amplifier circuit, but for a floorstanding device that canceled out problematic room resonances. You'd set it where a peak was occurring, usually in a corner. It would measure the sound, amplify it, and play it back out of phase, canceling out the resonancekind of like Bose noise-canceling headphones for your room.]
Oh! The [Phantom Acoustics] Shadow (footnote 1). You probably never heard of itit was an active acoustic absorber. It worked, but dealers didn't really know what to do with it. Plus, it cost a fortune. By the time they got done with the manufacturing [by this time Pass had left Threshold], I think it cost something like $2000 just to build it. I think to be successful, it would have had to retail for less than $1000.
1994US Patent 5343166: Efficient high-fidelity audio power amplifier: One of my favorite tricks is to look at what other people do and think creatively. For example, see what happens if you just swap the words voltage and current. There's a Panasonic patent, by Sano, Hirota, et al, where a little class-A amp has its power-supply ground driven by the output of a bigger amp that can swing a lot of voltage. If you switch voltage for current, you have an altogether different beast.
With this circuit, you can run a cascoded gain stage at low voltage, but at high current through a separate power supply. Then, you can bias the gain stages at enormous currents. Bipolar outputs usually have a sweet spot around 100mA or so, but MOSFETs just keep getting more linear with higher-bias currents. MOSFETs love current. For example, a 100W class-A amp would need a 5A peak output into 8 ohms, so you'd normally bias it at 2.5A. Here, you could bias it at 10 or 15A. The distortion is inversely proportional to bias, so 10 times the bias gives roughly 1/10 the distortion. There's no need for feedbackyou get great performance without any. It's a cute idea. Technics made a few [of their version] but it didn't seem to go very far. This is one that I've got, still waiting to be turned into product.
1994US Patent 5376899: Amplifier with gain stages coupled for differential error correction: This is Supersymmetry, which was the basis for the X series of amplifiers. It's a way of connecting two matched, balanced amplifiers to more effectively cancel out noise and distortion. One of the advantages of balanced circuits is that when the two halves are summed, noise and distortion that are in phase cancel out. This only works to the extent that the noise and distortion are identical in both halves. Rather than trying to totally eliminate noise and distortion, Supersymmetry works to make the noise and distortion in the two legs identical, which is comparatively easy. The two inputs of the balanced circuit are cross-coupled; the noise and distortion created in each half is fed through the other, so that they're more closely matched at the output, and effectively canceled. It's another way of making a very simple circuit work well enough to operate with less feedbackalthough it's a kind of feedback itself. You may notice that this amplifier was invented before the Aleph 1998 patent, but came out years later.
1998US Patent 5710522: Amplifier having an active current source: This the Aleph circuit, where the output device is biased to run in single-ended class-A mode. It draws current from the negative rail to the output. One of the reasons I started Pass Labs was because I wanted to go a new direction. When a company is successful, as Threshold was, it limits what you can do, how far you can diverge from successful products. If you do something different every year, no one will buy your products. Pass Labs was a clean slate for me.
There was a lot going on with single-ended designs, but it was mostly in the 12W region. I discovered that, with the Aleph circuit, I could get a reasonable amount of power before it transitioned into push-pull. The Aleph 0, which actually preceded this patent, was biased to produce 75W in single-ended class-A mode. It was a simple circuit with only three stages. The Aleph 3 was different. It had only two gain stages and was purely a single-ended design, with no transition to push-pull. There were no adjustments of any kind and it was impossible to break. Now, in the X series, we're again using a push-pull output stage that's biased by a single-ended constant-current source, similar to the Aleph 0.
Todaywhich is your favorite? It changes, depending on what I'm doing at a particular time, what else is in the system, what I'm listening to, how I'm feeling. Each one is a different experience. Plus, there are different situations where one type of amp is better suited than another. Yeah, I get asked that question pretty often. You know, I can't answer it. [laughs] They're like children...it's like asking which is your favorite child. Nelson Pass
Footnote 1: See Stereophile's December 1989 review.Ed.