Notice: Undefined variable: raticle_ordinal in /usr/home/dlamkins/public_html/lamkins-guitar/struct-links.phpi on line 25
http://lamkins-guitar.com/music/article/class-illusion
David Lamkins picked up his first guitar a long time ago. As best he can recall the year was 1967: the year of the Summer of Love. Four decades later David has conjured up an amalgam of folk, rock and jazz solo guitar music for the occasional intimate Portland audience.
location: Portland, OR USA

Facets: amplifiers, technology, terminology, @musings info
Loading...

The "Class-A" illusion

While I don't dispute the technical definitions of the operational classes, I believe that they have negligible predictive or descriptive value when it comes to guitar amps. When amp classes were defined (a long time ago, in the `30s or `40s) everyone concerned assumed that the goal was linear amplification, free of distortion insofar as the technology allowed. I don't believe that operational classes are useful in describing guitar amps because the goal is for their power stages to be operating nonlinearly during normal operation.

Suppose you have a guitar amp that's built to operate in Class A (single-ended or push-pull - the argument applies to both) up to the limit of its unclipped power output. If you push the amp harder, distortion will rise dramatically as the output stage clips and the power output will continue to increase. At this point the amp is not operating in Class A because an output tube is either saturated (unable to conduct any more current) or cut off (not conducting at all) for some portion of the cycle. The important distinction is that the cutoff and saturation points are symmetrical, otherwise the amp would be - by definition - Class AB. Our clipping Class A amp is still technically Class A because it will deliver full unclipped power with the output tube(s) conducting for a full 360 degrees.

Now consider a Class AB push-pull amp. Full unclipped power will be more than the true Class A amp of comparable topology; letting one tube cut off for a portion of the cycle allows the output stage to swing more voltage into the load. The key difference between Class AB and Class A is that conduction occurs for less than 360 degrees even when the amp is still operating linearly.

This is where the A vs. AB distinction becomes a matter of degree. Traditionally, Class AB amps were designed to maximize available power while still providing acceptably low distortion while operating in their linear range. They had to be biased hot enough to reduce crossover distortion (remembering always that low distortion was an implict goal), but not so hot as to approach the lower power output of Class A operation.

What if you're running a push-pull amp designed for true Class A operation, but the output tubes have aged symmetrically and are unable to pull their full rated output current? That case is simple: the full output before clipping will be reduced. The amp will still be running Class A.

OK, next case: Consider the same amp (designed for true Class A operation), but now the output tubes are mismatched such that the amp delivers full unclipped power with one tube not conducting for a full 360 degrees? Clearly you'll get an uptick in second-order distortion as the power increases above the point where both tubes are still conducting. Our mismatched tube is cutting off earlier than it should, yet can still push full rated output before clipping. The other tube is conducting for a full 360 degrees. This corresponds better to the definition of Class AB than it does to Class A. The operating class definitions say nothing about symmetry, although it's implied by the desire for minimal distortion. But symmetry is not required by definition, so our Class A amp with a weak tube is really - by definition - a Class AB amp.

Now consider an ideal hot-biased Class AB amp running matched tubes. The hotter the bias, the closer it will approach Class A operation while still producing unclipped output. At some point, any differences between 360-degree conduction and some lesser coverage of the full cycle will have an inconsequential effect on the perceived tone and behavior of the amp. Remember that we're still looking at unclipped output. The difference between this case and the case in the preceeding paragraph is that third-order distortion will increase as the output power exceeds the point where both tubes are not conducting for a full 360 degrees, because the output transformer tends to cancel second-order distortion due to the symmetry of the push-pull circuit. We're still not clipping the output.

So where do you draw the line? Do we say that a Class A amplifier is only running Class A so long as it's running perfectly matched tubes? If not, then why not admit that a hot-biased Class-AB amp is "close enough" to Class A at some point? What if the difference was down to twenty degrees of non-conduction? How about ten degrees? Five? One? The only detectable difference between a perfectly balanced hot-biased Class AB amp and an unbalanced Class A amp is in the amount of second-order vs third-order distortion products as the amplifier approaches full unclipped power output.

The problem with the class definitions is a matter of intent. No one ever considered a hot-biased Class AB amp at the time the class definitions were conceived. Such an amp would have been rejected as a poor design in those days. The entire reason for the existence of a Class AB amp was to increase the output power available from a pair of tubes without unduly compromising low distortion. If your intent was to minimize distortion you'd build a Class A amp. To maximize power you'd build a Class AB amp and set the bias just high enough to give the negative feedback a fighting change to eliminate crossover distortion.

If you've followed my arguments so far, you'll see that I'm advocating that there's no important difference between Class A and hot-biased Class AB guitar amps. The distinctions, considering that guitar amps provide little if any facility to balance the output stage, are a matter of degrees (pun intended). Sonically, there's too little difference between the cases to be noticeable, especially considering that a guitar amp is intended not as a linear amplification device, but rather as a non-linear tone-modification device.

Within the closed universe of guitar amplification, "Class A" has become a marketing term implying the presence of a cathode-biased output stage. As such, it's a useful distinction. A cathode-biased guitar amp exhibits different behavior than a fixed-bias guitar amp.

The technical definition of Class A operation should not, IMO, be applied to guitar amps at all because

  1. the intent of guitar amplifiers is not linear signal reproduction, upon which the definitions of operational classes are predicated, and
  2. any sonic differences between Class A and Class AB (applying the technical, not marketing definition) power stages are swamped by other distortions (preamp, speakers) in the overall system and therefore have no predictive or descriptive value.
January 13 2006 04:29:21 GMT