Saturday, June 20, 2009

A Transformer-Coupled Booster Amplifier for the AN214

Due to it’s inherently low power output design, does addition of a transformer-coupled booster amplifier to an AN214-based amplifier aid or hinder it’s sound quality?

By: Vanessa Uy

Given it’s inherently low power output – between 4 to 5 watts into an 8-ohm load. It was therefore inevitable to find a way of increasing the “meager” output of the AN214 IC-based audio amplifier without sacrificing it’s inherently good sound quality (on the cheap?). I wonder how many ways were “engineered” during the past 30 + years or so history of this device. Surprisingly, a consensus was reached of choosing a “primitive” transformer-coupled PNP transistor-based design that probably dates from the 1950s.

The design configuration of the transformer-coupled booster amplifier used to increase the 5-watt power output of the AN214 IC-based power amplifier for all intents and purposes resembles that of a transformer-coupled 300B-based push-pull tube power amplifier circuit. Except that the 300B tubes (or valves as they say in Merry Old England) were replaced by MJ2955 PNP power transistors in a TO-3 package. From my point of view, this transistor-based transformer-coupled booster amplifier circuit was probably based on the first transistor-based audio power amplifier designs first published in the US-based audio electronics magazine called AUDIO. Probably those issues dated between January 1960 to December 1961 - i.e. the Golden Age of Stereo.

During my experimentation of the MJ2955-based transformer-coupled booster amplifier – which to all intents and purposes is a continuation of the experiment done by my older audio-buddies during the last 25 years or so – I did optimize the input transformer winding design. By eliminating stray capacitance of the input transformer / driver transformer winding and optimizing it for a near-perfect square wave performance and transfer function characteristics. I did manage to reduce the booster amplifier’s recommended negative feedback level, which only improved the sound quality to no end.

Compared to its 300B tube-based sibling sound quality wise, it was a close match. Although the 300B-based push-pull amplifier did manage to display irreproachable beauty where it excels. Like in the Middle-C region of the musical performance – i.e. upper bass and lower midrange region of the audio spectrum. And even though it was criticized during the 1980s that transistor-based transformer-coupled booster amplifiers are notorious for having high total harmonic distortion (THD) in comparison to a full complementary direct-coupled solid-state power amplifier. But if you wind the transformers properly by eliminating stray capacitance – especially the input / driver transformers, the inherently high THD (total harmonic distortion) of transformer-coupled solid-sate booster amplifiers can be dramatically reduced. Although high-quality audio frequency transformers – driver transformers and output transformers - that are wound at artisan levels to optimize their square-wave performance and optimize their transfer-function characteristics across the audio spectrum are magnitudes more expensive in comparison to matched pairs of high-power output transistors.

The bad news is – from a perfectionist audiophile’s perspective – a lot can go wrong, sound quality wise, in the actual construction of full complementary direct-coupled solid-state power amplifiers. Especially if the “engineer” doesn’t trust their own ears and chose to trust on measuring instruments instead could result in a sluggish sounding power amplifier with unlimited reserves of power - but one that is sorely lacking in the pace, rhythm, and timing department. That's why I really hate musically illiterate audio engineers. Looks like the old-school wisdom gained by audio engineers during the Golden Age of Stereo still applies well into the 21st Century. And it also provides a low-cost route to audio nirvana via the AN214 IC-based audio power amplifier route. Ain’t it fun to be an audio cheapskate?


  1. I assume that you've arrived in designing the best-sounding driver transformers and output transformers through a combination of trial-and-error sound quality assessment and with an aid of a cathode ray oscilloscope and a function generator's square-wave output. Or in your case it could have been a Windows PC - compatible virtual instrument package. Although there are some tweaking tips that you should need to know.
    When selecting the negative-feedback level of your booster amp for the AN214, make sure that you've optimize the windings of your input / driver transformer and the output transformer by making sure they are not producing a sharp supersonic treble peak manifested by the ringing in your square-wave tests - which could easily "fry" tweeters in a matter of minutes. Transformer ringing is primarily caused by leakage inductance and winding stray capacitance forming a resonant circuit - i.e. tank circuit. Though this can be easily prevented if you follow good transformer winding tips.
    Back when I was 10 - in 1982 - "ringy" transformers from a badly-designed AN214 transformer-coupled booster amplifier manifest itself as a high-pitched noise similar to that of an old Black and White TV's flyback transformer plus the horizontal sweep frequency of 15.75 KHz. At audible levels, this can easily destroy tweeters in a matter of a few minutes.

  2. I try to avoid - as much as possible - digital-based electronic test and measurement gear, including digital multitesters, because they tend to add their own "digital" noise into the system's steady-state condition you are trying to measure.
    For my purposed, I'm using a purpose-built (as in an electronic kit) Precision Function Generator based on the long discontinued XR2206 Monolithic Function Generator IC, and a good old cathode ray oscilloscope. The XR2206-based function generator - when calibrated and tweaked properly - can output a sine wave with only 0.01% total harmonic distortion (THD) - especially when you use audiophile grade Alps potentiometer. It can also generate square and triangular waves of various frequencies up to 100 KHz. The 1 KHz square-wave setting is very useful in assessing the phase shifts of the transformers that I self-wound, or whether they are prone to ringing.
    I tend to over-engineer the transformers I use in my MJ2955 transistor-based transformer-coupled booster amplifier. Like Cary and Conrad-Johnson used to do. The input or interstage coupling transformer of my MJ2955 booster amp uses the same core size as the one used in a push-pull amp with a 50 watt power output - except the winding configurations of course - to make it have an excellent bass performance.

  3. First of all, I think that the XR2206 multifunction generator on-a-chip is still in production. A number of websites on tips on "tweaking" this IC still appear on Internet searches.
    Using a generously-over-sized transformers for both the input and output transformers of the MJ2955 PNP transistor-based transformer coupled booster amplifier for the AN214 IC-based amplifier does tend to improve the bass frequency response / low-frequency bandwidth. It also enables the amp to handle more powerful low-frequency signals before transformer core saturation sets in. And transformer core material also plays a role in sound quality. If you can still manage to find one, a 78.5% Nickel Permalloy transformer core - very popular during the 1950s - produces a very gorgeous midrange frequency response. But overwhelmingly, today's audio-frequency transformers are made out of 4.25% Silicon Steel / A-Grade Silicon Steel. Using inter-stage transformers with generously-over-sized cores lowers overall measured distortion and makes your Saint Säens organ listening experience more interesting.
    The Square-Wave signals required for measuring inter-stage transformers and output transformers are of such a sufficiently large magnitude that "stray digital noise" level becomes a non-issue. Though using a kit-based multifunction generator like the XR2206 IC-based function generator can be "sexy" from an audio enthusiasts perspective.
    Have you tried using personal computer-based virtual instrumentation? I am currently using a Keithley Metrabyte DAS 1800 multifunction board with 12-bit DAC used as the signal source and the 12-bit ADC used to convert the signal output of the component under test into something the PC understands - i.e. Using the National Instruments' Lab Windows / CVI virtual instrumentation software. Lab Windows is a visual programing tool that uses library functions to generate compiled C code. So an extensive knowledge of C+ is a must. Luckily this virtual instrument came free of charge with a second-hand PC that I bought 5 years ago.
    P.S. Does the transformer-coupled booster amplifier for the AN214 IC-based amplifier remind you of the uranium-238 booster jacket devised by Dr. Edward Teller and Dr. Stanislaw Ulam for their H-Bomb?

  4. PNP transistor based amplifiers such as you described probably became popular during the 1950s. Back then they relied heavily on germanium PNP transistors because silicon types were still difficult to make. These amps looked quite like tube amplifiers (or valve amplifiers as they say in Merry Old England) of the time - in terms of interstage and transformer layout. But according to first-hand experience of audiophiles back in the 1950s, these amplifiers tend to offer very poor performance compared to their tube based counterparts.
    Long story short I inherited one described in this blog - ie. AN214 with MJ2955 transformer-coupled booster amp. Unlike the one described in 1950s era audio journals, this one did sound as good as you described - like that of an 300b push-pull amp in fact. Did one of your audio-buddies manage to manufacture such amp them? Please drop me a line on transformer winding details - ie. true center-tapped wound audio transformers that sound as good as you described.

  5. Good audio transformers that I used in my set-up are about impossible to produce using our current technology of large-scale integration integrated circuit manufacturing methods. Hand-wound is probably the only way to go, so don't be surprised that tube / valve amps these days that sport a well-designed audio transformer normally retail at Barrett Light Fifty prices - i.e. around 3,000 US dollars MSRP.
    I am somewhat weary about giving out detailed tips on 1950s-era transformer winding tips in our post-9 / 11 world because these tech tips - in conjunction with the main driver magnets used in better hi-fi loudspeakers like B&W DM601 can be used to create centrifuges for purifying uranium-235 to 99.999% (six nines) purity. I'm still currently checking out other on-line audio transformer winding tips not yet being shut-down by the US National Security Agency or the US Central Intelligence Agency. If they are off the hook, then I'll be posting what the other sites are posting without fear of being "digitally redacted".

  6. Back in the good old days of thermonuclear weapons proliferation, Dr. Stanislaw Ulam - when asked by the press whether he worked with Dr. Edward Teller with the H-Bomb - Dr. Ulam softly replied: "Dr. Teller worked with me."
    Transformer-coupled solid-state amplifiers that had been produced commercially are few and far between. One of the few that I know of is the McIntosh MA6800 amplifier - which is equipped with "autoformers" - i.e. center-tapped transformers without any secondary windings. These do essentially the same job as the output transformers of a tube / valve amplifier by providing exactly the right load impedance for the output stage to drive whatever the impedance of the load or speaker that is connected. This , McIntosh claims, means that the output stage is always running into the optimum load (good impedance match) for the highest efficiency - also improving performance and reliability.