Tuesday, May 13, 2014

Negative Feedback: The Bane of High Fidelity Audio Amplifiers?

Though it is still an indispensable necessity in solid state hi-fi power amplifier designs, should all hi-fi audio power amplifiers be designed to work with out negative feedback to achieve a more natural sound?

By: Ringo Bones

The zero negative feed back movement on hi-fi audio amplifier design first gained both widespread and worldwide notice during the hi-fi scene of the 1990s when many hi-fi enthusiasts, especially in the do-it-yourself front, noticed first hand that vacuum tube based audio amplifiers achieved their most natural sound when negative feed back was minimized to near at or reduced to zero. Although even now no solid state transistor hi-fi audio amplifier design has been able to work with minimal – never mind zero - negative feedback, should all hi-fi audio amplifiers, whether tube or solid state, be made to work without the use of negative feedback in the future in order to achieve a more natural sounding music playback result?

Historically, negative feedback in audio frequency power amplifiers was formulated to alleviate the problem of harmonic and frequency distortion that made voices unrecognizable and unintelligible during the early days of long distance telephony. The only way to solve the problem back then was the design of inherently linear output vacuum tubes that work without the aid of negative feedback – like the PX25, 2A3, the 300A and later on the 300B. But such very linear vacuum tubes are very expensive to produce – even now – thus the necessity of negative feedback. Harold Black who graduated from Worcester Polytechnic Institute back in 1921, he later took a post as a research scientist at Bell Labs was one of the more famous electronic engineers who conceived of the negative feedback due to his frequent contact with the electronic engineering academia at the time.

Though the British electronics engineer Paul Voight also deserve credit for inventing the concept of negative feedback for use in electronic audio amplifiers by patenting it back in January 29, 1924. Given the cost-effectiveness of negative feedback in making vacuum tube based audio amplifiers work linearly during the period, it eventually gained widespread acceptance and made the 300B the only widely manufactured directly heated triode vacuum tube for audio output power amplifier use that would work linearly without the aid of negative feedback. So the manufacture of inherently linear vacuum tubes becomes non-economic and later more modern vacuum tube designs with multi-element components in its glass envelope that where maximized for gain via negative feedback merely paved the way for the solid state transistor.

Given the runaway success of negative feedback, virtually all vacuum tubes still manufactured today intended for preamplifier use can be very hard to implement in a design that works without negative feedback. The only widely produced hi-fi audio preamplifier design to date that could work without negative feedback is the Art made by Conrad-Johnson using 6DJ8 / 6922 triode preamp tubes though critics said Conrad-Johnson could have used “inherently” more linear preamplifier vacuum tubes in the Art – like the 6SN7 / 6SL7 family of vacuum tubes given that 6DJ8 / 6922/ ECC88 vacuum tubes were originally designed for electronic measuring instrument use like 1950s era fast sampling oscilloscopes.

From an electronic engineering standpoint, hi-fi audio power amplifiers designed with negative feedback – whether thermionic vacuum tube or solid state bipolar transistor or MOSFET based – tend to have lower parts cost despite of the circuit complexity and tend to make the low frequencies / bass frequencies rhythmically taught and enables the power amplifier to drive a hi-fi loudspeaker with a widely-varying impedance curve. But from a musical standpoint, the time smear induced in the application of negative feedback tend to make the power amplifier to sound harmonically threadbare – i.e. that “gray electronic sound” – and tend to become rhythmically reticent especially in the vital midband section when most of the music happens.

Hi-fi solid state amplifiers – whether bipolar transistors or MOSFET based – tend to be very impossible to work and maintain overall stability and a total harmonic distortion below 1 percent when the overall negative feedback is reduced below 30-dB. Most thermionic vacuum tube power amplifiers – even ones operating in push-pull ones using more modern beam tetrode (EL34 and 6L6) and pentode tubes (7591A) can be made to work with as little as 5-dB of overall negative feedback though the bass frequencies may become softer and nebulously blobby.

Despite the caveats, the inherent naturalness of the midband of zero negative feedback single ended triode thermionic vacuum tube hi-fi power amplifiers is just too hard to ignore. Sadly, mainstream solid-state amplifier designs may not be able to achieve such superb sound quality before the year 2050 until the top consumer electronics manufacturing firms starts making solid state bipolar transistors and MOSFETS that are as linear and good sounding as the 300B directly heated triode thermionic vacuum tube and amplify and playback music without the help of negative feedback.

1 comment:

Arlynne Ann said...

I think the necessity of negative feedback - in both thermionic and solid-state electronic audio amplifiers is due to the wide bandwidth they must operate in - i.e. 20-Hz to 20,000-Hz audio bandwidth and sometimes a bit more. I heard audio systems built around active crossovers using integrated circuit operational amplifier devices with as much as 60-decibels of negative feedback with a sound quality rivaling that of an EL34 tube based amplifier or even better. The AN214 IC audio power amp had been known to rival in sound quality of an 2A3 based single ended triode amp with zero negative feedback. Reducing the bandwidth - or Q - of a typical audio amplifier (thermionic vacuum tube or solid-state bipolar transistor or MOSFET) needs to operate in via active crossover circuitry will definitely result in better sound quality despite high levels of negative feedback needed just to make them stable.