Conversations with Peter Baxandall:
Part 2
Opinion from a distinguished British engineer/audiophile.
IN THE LAST ISSUE (4/79, p. 17), Mr. Baxandall was discussing those areas in audio which he considers most in need of work. He said he thinks the sound systems found in average homes need to be better quality and simpler rather than full of novelty and gadgetry. -ED.
TAA: What do you consider the home audio system's weakest link?
PJB: It's customary to say the weakest link in the audio system is the loudspeaker. Sometimes I feel this is said without thinking, because though it's probably true that when everything is being done up to the very best standards that are today achievable, the loudspeaker is really the thing that's furthest from perfection, I think that when one actually listens to practical systems, very frequently one finds the loudspeaker is nothing like the weakest link. One gets badly balanced program sources, poor recordings, pick-ups or even sometimes amplifiers that make nasty noises, so it's certainly not always true that the loudspeaker's the weakest link.
Another thing, unfortunately commercial pressures always move people to achieve the same result more cheaply. In tape recording, or with cassettes manufacturers try to achieve at 3 3/4" the quality that people once thought they could only get at 7 1/2." So nearly everybody goes over to 3 1/4 " and quarter track, or cassettes. Hence, rather than achieving something better, which ought to be the result of improved tapes and tape heads and general techniques, what they really do is to achieve about the same somewhat mediocre results more cheaply than before. I think the dbx© system, though excellent in itself, is in danger of doing this. It makes it possible to put more minutes of music on one side of a disc, and if that's carried to extremes you'll end up with the same slightly mediocre result, only more of it on each disc.
TAA: Yes, as you say, the commercial pressures are there.
To change the subject, what are your views on quadraphonic sound?
PJB: I'm not in a very good position to comment because I haven't really done anything much in the field. The question ties in with an earlier one in which I was talking about headphones and dummy head stereo. I think if the right thing is done with the head phone technique, it ought to achieve satisfactory spatial representation without needing full quadraphonic techniques.
I set up four loudspeakers for the B.B.C.'s two or three experimental quadraphonic programs using both the Radio Three stereo transmitter and the Radio Four one, which usually carry separate pro grams. This was four fully separate channels with no encoding technique. Some of the program items were a bit gimmicky, you know, spears thrown across the room or something. But they also included one or two recordings made quadraphonically in the Albert Hall.
Oh, it certainly added something that was quite pleasing and fascinating. It didn't leave me with the feeling "My goodness, I really must go in for surround sound. " Certainly four-channel recordings of the type that give one the illusion of being in the middle of the orchestra are not what we want for anything in the nature of classical music. I wonder whether the primary appeal of quadraphonics may not be to people interested in the pop type of presentation. Of course, in recent years the B.B.C. have been doing a great deal of work on quadraphony behind the scenes.
We are now having fairly regular program items in Matrix H quadraphonic form. I have not listened as I haven't a Matrix H decoder. I'm rather in a halfway state-I haven't made up my own mind about quadraphonics yet.
My fear about disc quad is that, with CD4 at least, it will result in a degradation of the general quality of the discs. It stretches the ordinary disc format to the limit to put high frequency carriers on for additional channels. It has been demonstrated to work quite impressively, of course, but one questions what it will be like in the hands of the average user three years after he's bought the equipment.
TAA: Are headphones going to give us all the dimensions?
PJB: I have done one or two experiments in that direction. I'd like to know a lot more about it. When you go to a concert you only have two ears and all the audio information you get from your surroundings comes from those two ears. Of course the thing you can do at concerts which you can't effectively do with headphones or even loudspeaker reproduction is to gain additional information by making unconscious little head movements. And visually you are aided when you see someone get up on the stage and start singing. On the occasions when I've attached a separate loudspeaker to the television set to get better quality for some music broadcast, if I put that speaker several feet from the set, when the soloist becomes visible on the screen, it's surprising how I imagine the set is where the sound comes from. Without that visual clue we suppose the loudspeaker is the source.
The mind is quite a complicated thing, isn't it-it's influenced by all these factors.
TAA: Do you think time delay techniques have applications in audio?
PJB: There are certain definite applications in audio where one would like to be able to put in easy and economical time delays-especially in speech reinforcement systems. In the high quality audio field I haven't given a great deal of thought to detailed applications, but certainly quite a long time ago Peter Walker had a scheme for stereophonic reproduction in a room in which he fed a long horizontal electrostatic loudspeaker system with stereo signals from the two ends with time delay-producing two wave fronts that were at different angles.(7) There are all sorts of possibilities in controlling wavefronts and their directions by use of loudspeakers in association with time delay devices. So the short answer is yes! TAA: How about amplifier distortion?
PJB: My view is probably rather unusual. I strongly hold the opinion that any competently designed high-grade amplifier properly us ed, by which I mean not overloaded, will sound exactly the same, no matter what careful subjective tests you give, as any other competently designed amplifier, and that this has been so for a long time. I'm truly amazed at the amount of writing in audio journals nowadays about this amplifier being very good except that it tends to give a rather too forward sort of sound to trombones, or something! People are absolutely convinced that there are these subtle subjective differences between one amplifier and another.
TAA: Do you think this is a difference in ears or in self-delusion?
PJB: Well, in my opinion it is usually a difference in self-delusion. All I can say is that in any instance with which I've personally been directly concerned, it has been possible to explain such subjective differences as were reported in much more mundane ways were supposed by the people reporting them to apply. One example. Several years ago an R.S.R.E. colleague said, ''There's something in these subtle things that are reported about amplifiers. I've got a quite definite instance of this at home at the moment." He had a Quad tube amplifier as his normal equipment and had borrowed a good Bang & Olufsen transistor amplifier. He arranged them with a change-over switch and asked if I would like to come and listen.
We agreed that I was not to know which position his switch was in. I carefully listened as he switched a number of times. I finally said which position I thought was the transistor one with the comment: "It sounds very slightly edgier or slightly more gritty than the other." He said, "Well, you're quite right, it is the transistor one." I then asked, "How careful have you been to get the gains of the two channels equal-this was a monophonic test- and he said, "Oh, I've been reasonably careful, certainly within a dB, I think. " I then started to alter the gain controls with the result that by the time we had altered gains slightly, I could not tell the slightest difference between which position the switch was in and neither could he-absolutely no trace of difference.
We were just guessing, 50/50. But both of us, before this had been done, had interpreted the difference not as being a difference in volume between the two, which was actually all it apparently was. We were trying to be perfectly sincere and genuine and we had come to the wrong conclusion. I wouldn't say that all things are explicable that way, and they certainly are not. Some transistor amplifiers are doubtless very bad. But I feel that some of the reported observations that a particular amplifier gives transparent highs but disappointing mid-frequency image definition, or some such thing, are just pathetic! Now all this leads to slewing rate, doesn't it? A little while ago I wrote a chapter for a new book whose editor is S.W. Amos. He retired a few years ago as head of B.B.C.'s technical writing department. My chapter, fourteen, is on amplifiers. (11). In talking about the normal type of loudspeaker amplifier that's in widespread use, I discussed the capacitor that's used to attenuate the forward gain to achieve overall stability of the feedback loop. I said that great care should be taken as to where this is put, because if the wrong things are done it may result in the first stage overloading with large high-frequency inputs or fast transients. I commented that the need to watch points such as this has been well known to enlightened designers of all kinds for several decades, but has been given prominence more recently by the coining of the phrase, "transient intermodulation distortion." Now, it seems the notion of TID is widely thought to be something new. To me this seems ridiculous. I mean, it's a new term, but I can assure you that in various amplifiers that I was connected with at R.S.R.E. for handling wide-band doppler radar signals and all sorts of things, we were acutely aware of the need to ensure that the slew rate was sufficient to handle the rate of change of signals going through them. We used to talk about " i = CDv/dr trouble", rather than slew-rate limitation. It seems rather amazing to me that this notion should suddenly come up in the audio field and be hailed as a sort of new and enlightened view.
I feel that Professor Otala has a tendency to regard quite simple matters in unnecessarily complex ways, and that, in doing so, he sometimes unintentionally misleads himself and mystifies his readers.
TAA: I believe that Jung felt (TAA, 1, 2, 3, 4/1977) that much of what Otala was saying about TID could be adequately explained in talking about slew rate.
PJB: Yes, of course it can. On the whole I find myself in broad agreement with Jung's way of looking at things, though I prefer to limit the terms of TID or SID to the situation where complete slew rate limitation occurs, that is where some internal stage runs out of current in the attempt to cope with i = CDv/dt.
It's true that the same circuit elements may give distortion at lower output levels and/or frequencies than this, but I feel it's then better regarded not as the early onset of SID, but simply as perfectly ordinary old fashioned non-linearity distortion, whose level is determined by the fact that one or more of the amplifying stages is having to turn on rather a lot of current to feed a capacitor.
This capacitor also reduces the amount of feedback in operation, so that the generated distortion appears all the more strongly at the amplifier output.
If you say you don't agree with some of Otala's ideas, you're just looked upon as a fool by many people. They say "But it's all well established by now, it's in the literature you know." TAA: Sometimes at TAA I have felt like David taking on Goliath.
PJB: It's a good thing that you have- I think it's high time there was some opposition. The notion that one must make sure an amplifier's internal stages can properly track the maximum rate of change in a signal seems to me to be fundamental and basic to feedback amplifier design-it's an old idea. But there's no absolute need for an amplifier to be able to track higher rates of change than those which actually occur in program waveforms. These rates of change are a lot lower than is often imagined, and I think Jung's criterion is really far too conservative. I think his factor is well over four times higher than necessary. But it's a sort of "right side error." TAA: I think he realizes he's being conservative, but I think he wanted to be very sure of his ground.
PJB: If you find that, for no subjectively perceptible quality impairment, an amplifier has to have a slew rate limit appreciably greater than the maximum rate of change in the program, this must be for one of two reasons. First, the true program rate of change may be augmented by the unwanted presence of FM multiplex carriers, tape bias waveforms, or radio interference. Second, the ordinary high-frequency non-linearity distortion of the amplifier may be too high; modifications aimed at increasing the slew rate limit may put this right, but it is not really the slew rate limit as such that is involved at all.
(Since the interview, I've expressed some of these ideas in greater detail, and I hope with clarity, in the first of a series of Wireless World articles on "Audio power amplifier design", in the January 1978 issue, pp. 53-57. But please see also the March 1978 issue, p. 45, for important corrections-I was not given the opportunity to check the proofs for this article! In the article I referred to the slew rates of pro gram waveforms, and the slew rate limits of amplifiers. I have been criticized for doing this by Richard H. Small, and I think, on reflection, that the criticism is justified.
I now agree with him, that, despite contrary usage in some other publications, one should reserve the term "slew rate" for the rate of change that occurs under conditions of actual slew rate limitation or overload. There is no need to use the term slew rate for a program input waveform, when what is meant is simply peak rate of change-better just to call it that.)
TAA: What sort of things do you think amateurs should be doing these days?
PJB: The danger, I think, is that it's only too easy for an amateur who hasn't really sufficient technical background to spend a lot of time empirically trying things which, had he got more understanding of what he was doing, he would see to be unnecessary. The important thing, really, is to try to get projects of such a nature that the amateur without much professional training can really understand what he's doing and get somewhere with, rather than just do lots of uncorrelated experiments. Obviously amateurs who like making things and who have aptitude with their hands can get great satisfaction from doing things like making microphones, even.
TAA: You mentioned earlier that you had fun making a phonograph pickup. But would you say today's phonograph pickup has gone to such a level of sophistication and dependence on sophisticated materials that it's not a very rewarding project?
PJB: Yes, I mean in the old days one could make things like a 78 moving-coil pickup with a coil which was about a centimeter long. By the time you start making things much smaller, and stereo moreover, it gets beyond the abilities of most amateurs, doesn't it?
TAA: How do you feel about integrated circuits?
PJB: I'm very much pro integrated circuits for all sorts of things. I certainly consider the IC op amp to be an enormous boon to audio in many ways. I've made use of them so far mainly in audio measuring equipment, particularly oscillators, rather than in program handling equipment. I spent the best part of two years recently designing an instrument of which my original prototype was made here.
It's rather an unusual oscillator because it covers the whole audio range in a single sweep, with very constant amplitude. It has one or two additional facilities that are quite useful for some audio work-you can make it sweep automatically, you can set it to any frequency you like such as 1 kHz, and either set a switch to the going-up; or the going down position, and then when you switch from dial to sweep it sweeps automatically upwards or downwards, either slowly or rather faster...
TAA: All the while maintaining...
PJB: very constant amplitude-it's within +/- 0. 05dB, actually, over the whole range. It can also put a warble on, which is a sort of noise sometimes used in acoustic measurement work. Frequency modulation with very little amplitude modulation-it's very clean in that respect-and it also has a facility for putting in frequency markers. You can use this for plotting loudspeaker response with output on an oscilloscope. If you put markers in, you press this button, when it goes through a hundred, there goes the marker, and the thousand, and if you can hear the 10kHz one, your hearing is better than mine! You need to hear 30kHz actually, to be able to detect the presence of that marker! Well, these may either be added to the sine wave, just for an easy visual display, or you can make them available on one of the pins of an auxiliary output socket, and you can then arrange, when plotting a loudspeaker response curve, to have the marker produce a vertical line at three positions which enables you to draw the frequency scale.
The thing about this oscillator is that unlike other, so far as I know, available oscillators it's neither a beat frequency oscillator nor a function generator.
TAA: Simplification is the most difficult achievement.
PJB: Yes, though I think I've never known an audio circuit which is in essence so simple and yet in practice has had so many small subtle problems which I've spent an awful lot of time sorting out. I think we've got it into a very clean state at last. In principle this device is really an R.C. oscillator, but its resistors are replaced by a diode circuit. Turning the dial simply alters the DC bias voltage on the diodes which in turn alters their resistance and hence the frequency. That's the principle.
The diodes are actually transistor junctions.
The background is rather amusing. Soon after I got going as a consultant I wanted to have a little oscillator that could produce warble tone and not having anything commercially available that I could afford, I made up a simple little RC oscillator with a double ganged pot and a circuit for producing this frequency modulation. This worked quite nicely, and I'd been doing quite a lot of work at the time for KEF Loudspeakers of Maidstone. A man there named Chris Moore is a keen circuit type and I showed him this little device.
He was rather fascinated with it. Nothing much happened for a year or so, and then he spotted in the "Circuit Ideas" page of Wireless World a little contribution from a Czechoslovakian reader, named Kraus. He described a rather crude RC oscillator which in itself was nothing very special but the resistors in his Wien bridge were formed by diodes. He varied the current through these and achieved a linear frequency variation.
Well Chris Moore put this idea into my circuit. It also occurred to him that if you varied the DC voltage applied to the diodes rather than the current through them, you would get a logarithmic frequency scale rather than a linear one. He tried it, and it worked quite well. He told me about it when we next met and lent me his original board. It covered from 50Hz to 10kHz or so with a more or less logarithmic scale. The amplitude bounced about a bit-it was a thermistor controlled oscillator and distortion was about 0.2 percent.
It seemed quite encouraging and I decided if everything was taken with rather more care it ought to lead to a very good type of oscillator.
It started about March 1975. By August I'd got this prototype built. A couple of people from industry were quite keen on it and one of these took it around and showed it to many possible users. He reported back enthusiastically, even to the extent of several firm orders.
However several said it was a. pity the distortion wasn't below 0.2 percent and that its upper frequency limit was 20kHz. A couple of loudspeaker firms said they'd like it to go higher, which might seem surprising at first sight since nobody, one would think, is very concerned with hearing frequencies above 20kHz. But apparently in testing tweeters it's of interest to know where the resonant frequency is, even though it is above audibility.
They asked if it could go to 40 or 50kHz.
I then began the hard work of getting the distortion down and extending the range. It's a bit like climbing a mountain, you keep feeling you are just about to the top but it sometimes takes a bit longer than you might think. By the time a good few more months had gone by, I began to feel that this had really become rather a bigger than anticipated project: it was high time to try to get back some of the development costs and began then to look about for an established instrument firm to take it on.
It has been looked at by several firms, including one from the U.S.A., all of which has consumed a lot more time, of course. However, several have now been hand made and one in use for factory testing by two British firms.
TAA: It sounds like something most engineers would trade their eye teeth for.
PJB: I think it ought to be, properly handled, a useful product. Its distortion is under .05 percent over the whole audio spectrum and its range extends up to 40kHz. Its got an ac curate 60051 attenuator, and it's battery operated. It has a novel type of battery-state indicator.
TAA: Does this use a lot of current?
PJB: No, because it just flashes the LED in short pulses. With new batteries it flashes about three times a second; by the time its got down to about one a second the message is that you'd better think about getting a new battery some time in the next day or two.
When it stops flashing, the oscillator will still be working but it's definitely time to put in a new battery. It's quite a simple two-transistor circuit.
TAA: You have a knack for those practical ideas, it seems. That's what has struck me about your earlier work. I have never gotten over my delight in reading (FAA 2/70, p. 5) that the cover material of a certain magazine was just the right thickness to form the gap in a choke for a crossover.
PJB: Yes, I'd thought to myself, well, how do you convey to an independent amateur what thickness he ought to use? I can assure you I measured quite a lot of covers with a micrometer. I also rang up the editor to ask if they were about to change cover stocks! We were talking earlier about distortion measuring equipment. I have another oscillator I've done recently. It has very low distortion and is a fairly cheap and simple instrument. It's not continuously tuned. . .
TAA: Is it stepped?
PJB: Yes, it's intended for amplifier distortion measurements, and it goes, as you see, 30, 100, 500Hz, then to 1, 3, 5, 10, and 20kHz.
TAA: Covering the normal places that you'd want to test.
PJB: Well, as I judged-I really designed it rather to go with this distortion measuring instrument I also made some time ago, it has the same frequencies as that, and the distortion is under 0.001 percent over most of the audio range. It's under 0.002 percent even at 30Hz, at other settings it's under .001. And that's a very cheap and simple little device, really--it uses just two op amps and 9-volt batteries and thermistor control.
TAA: It is surprising what's possible, isn't it. Well, as you say, "with careful attention to details..."
PJB: That's right. I think the strongest point I would wish to underline that I've been able to learn by association with effective people in the field is to think about fundamentals. I feel too many people, when confronted with a problem say, "Let's look around in a cookbook for as many oscillator circuits as possible and select one." That's never my approach. I am more likely to say, "On what factors does distortion depend and what can we do to minimize it?" Of course, the result of that attitude is that one finds oneself sometimes re inventing things someone else has already in vented. But I've found it to be a good philosophy by and large. And I'm sure this inclination to look sideways at what other people have already done is the cause of not inventing things. Very often you close your mind to the basic problem and just look for an existing solution. It is better to look for your own.
REFERENCES
1. Williams, F. C., "Introduction to Circuit Techniques for Radiolocation", J.I.E.E., Vol. 93, Part IIIA, No. 1, pp. 289-308 (1946).
2. Voigt, P. G. A. H., "Getting the Best From Records-Part III", Wireless World, Vol. 46, No. 6, pp. 210-213 (April 1940).
3. Scroggie, M. G., "The Genius of A. D. Blumlein", Wireless World, Vol. 66, No. 9, pp. 451-456 (Sept. 1960).
4. Benzimra, B.J., "A.D. Blumlein-An Elec tronic Genius", Electronics & Power-Journal of the (British) I.E.E., Vol. 13, pp. 218-224 (June 1967).
5. Clark, H.A.M., and Vanderlyn, P.B., "Double-Ratio A. C. Bridges With Inductively Coupled Ratio Arms", Proc. I.E.E., Vol. 96, Part III No. 41, pp. 189-202 (May 1949). (Much of this was taken verbatim from a 1941 Blumlein memorandum.)
6. Blumlein, A.D. and E.M.I. Ltd., "Improvements in and Relating to Sound Transmission, Sound-Recording and Sound Reproducing Systems", British Patent Spec. 394, 325, 22 pages.
7. Walker, P.J., "Wide Range Electrostatic Loudspeakers-Part 3", Wireless World, Vol. 61, No. 8, pp. 381-384 (Aug. 1955; parts 1 & 2 May and June 1955).
8. Baxandall, P.J., "Audible Amplifier Distortion is not a Mystery", Wireless World, Vol. 83, No. 1503, pp. 63-66. (Nov. 1977).
9. Hope, A., "Can You Hear Any Difference?", Hi-Fi News & Record Review, Vol. 23, No. 6, pp. 73-77 (June 1978).
10. Moir, J., "Valves Versus Transistors", Wireless World, Vol. 84, No. 1511, pp. 55-58 (July 1978).
11. Baxandall, P.J., "Radio, TV & Audio Technical Reference Book-Chapter 14". Edited by S. W. Amos. (Newnes-Butterworths, 1977).
Also see:
Conversations with Peter Baxandall; Part 1--A visit with the designer of those fed back tone controls.
A Family of Power Amplifier Power Supplies -- A versatile circuit for regulated DC in quantity, by James E. Boak
A High Accuracy Inverse RIAA Network -- New light and hardware for the phono curve's basic shape, by Stanley P. Lipshitz and Walt Jung
Audio Aids--On Phono arm geometry, turn-on/off noise, and zener size, by readers Goodman, Shults and Grandfeldt