Comments on Selected Audio Articles, Letters, & Circuit Ideas in Electronics World.> |
Updated: 4 Feb 2011
|
This two-part article is considered to have introduced the classic two-transistor RIAA input amplifier, which was widely used for many years afterwards. By modern standards, the configuration has many drawbacks, such as a lack of open-loop gain (which is particularily felt at the LF end of the audio band, where the RIAA equalisation gives the most boost) and poor load-driving capability. But you must remember that active devices were very expensive in those days, and you used as few as possible.
This article, together with a two-parter on Class-D efficiency published in 1967, are the earliest publications on the application of Class-D to audio amplification that I am aware of. I have always assumed that they were the inspiration for the Sinclair X10 and X20 Class-D amplifers, (though the circuitry is in fact rather different) but I was dead wrong, as on looking into it, the X-10 actually appeared in December 1964, before the Wireless World articles. These Class-D amplifiers had one thing in common. The linearity was dire.
This famous amplifier is notable for using an interstage transformer between the two-transistor voltage amplifier and the output stage; this allows the output to work like a complementary circuit (I think) even though the two output devices are both PNP. Interestingly, a letter from F Butler in the December issue indicated that even in 1966 the use of such a transformer was considered old-fashioned.
The preamplifier is notable for using a three-transistor input amplifier, though the other two stages (buffer, tone-control) use simple one-transistor amplifiers.
This was the first article on power amplifiers that I wrote, though I had been designing them for manufacture since 1975. This investigation into the concept of combining power FET output devices with bipolar drivers was done some years before I undertook my major investigation into the root causes of power amplifier distortion. (See Distortion in Power Amplifiers, Parts 1–8) In the throes of the design process, I realised with greater force than hitherto that the distortions in the small-signal part of a power amplifier were (a) far from negligible and (b) susceptible to analysis by a mixture of SPICE simulation and a few well-chosen experiments. I also determined that SPICE could be extremely useful in the analysis of output stages. The rest is history...
To prevent people being misled, I must point out that the final circuit falls somewhat short of the Blameless performance standards set by the Distortion in Power Amplifiers series. The input stage may look symmetrical, but in fact calculation shows it to be grotesquely unbalanced, with 16 uA flowing through the left transistor of the pair, and 580 uA through the right. This stage must have generated far more second-harmonic distortion than necessary, and with the benefit of hindsight I am not at all proud of it. This input Ic imbalance also means that the input stage transconductance is far too low, which precludes any emitter degeneration of the input pair, and this also explains why a dominant-pole capacitor as small as 15 pF is enough for stability.
I was not involved in selecting that lame title. At the time the editor seemed to feel compelled to put puns in the titles of articles if he could.
This is the letter I sent to EW at the time, which was duly published:
"I read Colin Wonfor's article on Class-A with interest. Having spent some time myself trying to make Class-A
amplifiers both tractable and as linear as they should be, I was surprised he did not adopt any of the technology
introduced in the Class-A (March 1994) and Trimodal (June, July 95) designs. No doubt there were good reasons, but it
would be nice to know what they were. Likewise it is a great pity that no performance details were given, to confirm that
the great cost and heat are really worthwhile.
My investigations showed that the use of power FETs in the Class-A output stages makes linearity worse. The Class-B
problem of a horribly jagged crossover region is no longer relevant, but the low device transconductance still causes
high distortion.
An automatic bias controller would have cost a few pence, and removed the need for a perilous quiescent-adjust preset that allows infinite current to flow at one end of its travel.
The instruction to trim the DC offset to less than 10mv with a preset seems rather strange when it is straightforward to make an amplifier that gives an offset of less than 25 mV with no adjustments.
It concerns me that there is no effective DC-offset protection given the large amount of power available. I note the output fuse (no value shown) but I would be interested to know how the constructor is going to select its value so it can protect loudspeakers without nuisance-blowing or introducing distortion. A 1 Amp slo-blo output fuse carrying 20W/8 Ohm generates 0.01% THD (third harmonic) at low frequencies, so fuse distortion could easily make the finer nuances of Class A linearity somewhat irrelevant.
It seems astonishing to use the constant-current (single-ended) mode for a large Class-A power amplifier as this doubles the already enormous heat dissipation. The largest version described releases 3 KW of heat, which will not be significantly reduced by playing music at full volume. Surely this will be an uneasy companion in summertime? The quiescent current for the so-called 300W/4R version is prescribed as 10.2A, so negative peaks cannot go beyond 4 x 10.2 or 40.8V below ground. This corresponds to a maximum sine output of only 208 W, so using supply rails as high as +/-75V appears to do nothing but increase the power dissipation. It allows no safety margin for loudspeakers that fall below a nominal 4 Ohm impedance. Perhaps there is a typo in the article for this huge emission of heat makes little sense to me.
At the risk of seeming discouraging, no information is given that makes me see this design as a significant advance in Class-A amplification."
The "new topology" is actually just adding an extra negative feedback path between the amplifier output and the VAS input; since it is well-known that global feedback is more effective than local or semi-local feedback, it is hard to see why this should improve linearity. The rather muddled text gives no clue on this point. No measurements are given that would enable us to see how well it works, if indeed it works at all.
However, to my surprise, the email (slightly edited to remove some of the criticism) was eventually published in the September 2009 issue, and indeed, once more in the November 2009 issue, which I think must have been an oversight.
Since the effect is proportional to input resistance, I extrapolate that the wholly unnecessary 10K input resistor would have given 0.015% THD at 1 kHz, which I suggest is approaching audible proportions, and a distinctly excessive 0.10% at 10 kHz. Oh dear.
Could this be the same chap that perpetrated the appalling Earache article in Mar 2009? See above. There can't be many people in the audio world called Kaguongo.
"30W High Fidelity Amplifier" by Arthur Bailey, published Wireless World, May 1968, p94
Arthur Bailey produced a new design that dispensed with an output transformer. It was a single-rail rail design with output capacitor, but apart from that looked fairly modern with a fully complementary EF output stage and VI overload limiting. The input stage only had one transistor though, not a differential pair, and must have intoduced a lot of unnecessary second-harmonic distortion.
"Output Transistor Protection in AF Amplifiers" by Arthur Bailey, published Wireless World, Jun 1968, p154
One of the first expositions of what is now the standard approach to VI-limiting.
"Low-cost 15W Amplifier" by Hardcastle & Lane, published Wireless World, Oct 1969, p456
This design was the first amplifier in WW to have a differential pair input stage, complete with the same value of emitter degeneration resistors that I prefer myself, though the tail was only a simple resistor rather than a current-source. Despite the input pair, it was a single-rail rail design with an output capacitor, which seems to be missing the point rather. This capacitor had a cut-to-the-bone value of only 800 uF, which with an 8 Ohm load gives -3 dB at 25 Hz. This implies that the capacitor would generate quite serious distortion, though whether this effect was known about at the time is doubtful. However- probably the vital point here is that there are separate DC and AC feedback paths, and the AC feedback is taken from outside the output capacitor, hopefully nullifying its non-linearity. The published THD graphs certainly show a rise at LF, but it is nothing like as steep or severe as I would expect from uncontrolled capacitor distortion.
1970-79
Dynamic Range versus Ambient Noise" by George Izzard O'Veering, published Wireless World, Apr 1970, p198
If the name of the author doesn't give it away, you might ponder the fact that this article was published in the April issue. This is (to my knowledge) the only April Fool article that ever appeared in WW. Having said that, the amplifier design, which was enormously powerful for its day, appears at a first look to be quite sound.
"Versatile Tone Control Circuit" by P B Hutchinson, published Wireless World, Nov 1970, p538
The only real interest in this article is a personal one. It was the first opamp circuit that I built, using a 709 IC. The circuit proved horribly prone to HF instability, and as I struggled with it it was not long before a slip of the probe destroyed the (rather expensive) opamp. 709s had no short-circuit protection and this was a real drawback to experimenting with them. The tone-control circuit itself might be versatile, but its control-setting for a given response was highly non-intuitive and profoundly un-ergonomic. So far as I am aware, this tone-control concept sank without trace and has never re-surfaced.
"Audio Preamplifer with No TID" by Yuri Miloslavskij, published Wireless World, Aug 1979, p58
"Audio Preamplifer with No Performance" is more like it. An uber-crude design that gets everything wrong; wholly passive RIAA equalisation, transistor biasing that depends critically on device beta, and a low +12V supply rail that makes non-linearity even worse. Mr Miloslavskij was with the Institute of Constructional Physics, Moscow, so possibly this design was intended to demoralise Western audio enthusiasts and so bring down Capitalism.
1980-89
1990-99
Douglas Self "Sound Mosfet Design" published Wireless World, Sep 1990, p760
The voltage-amplifier stage could be improved in linearity by adding another transistor within the Cdom local feedback loop.
Colin Wonfor "Class A to 300W" published Electronics World Mar 1999, p188
2000-2009
Russel Breden "A New 100W Class B Topology" published Electronics World June 2000
COMMENTS ON EW/WW LETTERS
The Input Filter Distortion Debate (started Aug 2000)
I never bothered to get involved in this debate, because it was about inaudible phase-shifts at silly frequencies. What escaped me at the time (and everybody else, it would appear) is that the tiny shunt capacitor was wholly irrelevant; the crashing error was putting a 10K resistor in series with the amplifier input. Such a source resistance increases both the distortion and worsens the hum performance of the typical power amplifier with bipolar input devices. By 2002, I had realised just how harmful this effect could be, and I wrote an article on the subject that was published in EW in May 2003. I found that a input resistor of 3K9 degraded the THD of a Blameless amplifier at 1 kHz from 0.0018% to 0.0060% at 50W/8R. At 10 kHz and 50W/8R the degradation was from 0.013% to 0.042% with 3K9.
The Class-A Imagineering Debate (started Sept 2004)
Graham Maynard's interminable "Class-A Imagineering" series quite rightly received a lot of very negative comments. I never got involved myself, something I now regret. I should have stood up to be counted. However, it would have been hard to improve on Phil Denniss's merciless dissection. I particularly like the phrase "Faith-based audio".
COMMENTS ON EW/WW CIRCUIT IDEAS
150W AB amplifier by Harrison Kaguongo (Jan 2007)
This design at first looks pretty conventional, but it has an enigmatic transistor (Q17 in Fig 3) whose function is mysterious because it is never mentioned in the text. The design has been simulated but not measured so it may not work at all in reality.