A.J. van den Hul: Going Dutch Page 2
Atkinson: Do you think the digital process will be improved significantly?
van den Hul: It cannot be very much improved. It can only come up to the level it was intended to have at the beginning. The promotional activity was extremely good, but the technology lagged behind, and the technology has to catch up with what is was supposed to be. We have had 14-bit single sampling, 14-bit double sampling, 14-bit with oversampling, we've had 16-bit single, 16-bit double, 16-bit 4-times oversampling. We have had analog filters, we have had digital filters, we have had a combination of the two. How many more variations will we have in three years' time? Does this show a stable system? No sir!
I think there's a different way of thinking between analog people and digital people. Measurements, more than the listening, seem to be what convinces people when they work digital. It's a different type of approach: when the bits are working nice, then the recording should be working nice. But it doesn't tell you how sound should give you the thrill, the goosebumps.
Atkinson: And that is something that live acoustic music does very well.
van den Hul: One of the things that astonished me mostly is that when you listen to real music, it sounds mainly more harsh, more aggressive, more overtoned, more distorted even, than a lot of hi-fi systems, where the sound tends to be very smooth, very polished, with all the sharp edges taken away so that it does not hurt at all. That may be what hi-fi lovers call "hi-fi," but I think that it is not a realistic approach. If you listen to real music, real sound, it's just the opposite; you should accept this because when you listen to music as it is, you should accept the truth. You should not change the world to be how you like it. (Although on the other hand, as a consumer, you have the right to make any changes you want to adapt things to your own way of listening.)
Atkinson: As someone once said, it is then no longer "hi-fi", but "my-fi." But if you listen to an orchestra playing, there isn't a great deal of high-frequency energy compared with a typical hi-fi system, yet everything is dramatically clean and clear, violins having an astringent edge to their sweetness which you almost never get from reproduced music. What, therefore, do you think are the most fragile aspects of live sound which are most easily destroyed during the transfer to record?
van den Hul: I think it's the transient response. The transient response, the dynamics, is somewhat different from what CD companies call dynamics. Let me explain this. CD companies equate dynamics with the signal/noise ratio which, theoretically for a 16-bit system, will be 16 times 6dB, ie, 96dB. Practically, it will be less than this, 90dB or 85dB, it doesn't matter so much. But it takes a certain time to build up this 85dB because each bit of 16 takes about 20µs to switch. Therefore if you build up, say, 100dB—I exaggerate, I make CD better than it really is—it will build up at 5dB/µs. If you take a good analog cartridge, it is faster than 5µs, but let us take its risetime as 5µs. A good analog record can easily reach 60dB, giving a slew rate of 12dB/µs. The build-up of a signal is much faster in analog than it can be in digital. CD may be better when it comes to S/N ratio, therefore, but the dynamics from an analog record, I will not say are always better, but can certainly be better, by a factor of 2.4.
Another difference between real music and reproduced music, I think, lies mainly in the phase relation between the harmonics and the fundamentals in the total tonal structure. This is missing, or tends to be missing, in a lot of records because while the equalization curve for the cutter gives a perfectly flat frequency response, so far I have only heard of only one company in the whole world who has cared about the phase relation. And that is the Island company's Sound Clinic in London, who use equipment designed by Tim de Paravicini. His correction curve is much more neutral sonically, especially in the midrange and the high end, than I've ever found on any conventional good pressing. What this means is that we have already left the analog island and swum to the digital island without having done all our homework. There is still a lot of work to be done in analog. This is extremely important: please make not only a correct frequency response but a correct phase response as well.
Atkinson: In live acoustic music, you know the phase response of the direct signal is correct. All the harmonics of the instrumental tone arrive at your ears at the same time as their associated fundamentals, and yet, if you look in such journals as the JAES, there is much work published saying that departures from this behavior such as linear distortion—timing errors—are not audible unless extreme. How does this tie in with what you are saying?
van den Hul: Let me put it this way. If you walk around in a town and a jazz band is playing, even at three or four streets' distance, you know exactly it's live! And you know exactly when it comes over a public address system. You know exactly when it's live and when it's not live. Everyone who says that the phase relation is not important ignores this fact. It always amazes me when I am able to know "This is live" and when I am able to know "This is reproduced." Partly it is because of the frequency range, but mainly, I think, it is because of the phase relation. And correct phase relation also implies very fast risetimes. If speaker manufacturers want to pay attention to this, they should reduce cone mass any how, and improve stiffness. If they do just the opposite, then they move away. Bad floppy speakers with a lot of mass are not able to reproduce sound very well, and these are mainly the speakers used in public address systems. I feel that one cone should do the whole frequency range, so we should go down first to small cone speakers with big excursions, low mass, and extremely high stiffness. This is the way, I think, to continue in the future.
Atkinson: What do you find so fascinating about cartridges?
van den Hul: Like a loudspeaker, a cartridge is a kind of transducer: you have an input and an output. A lot of attention has been paid to one part of the input—the mechanical part. To avoid dispersion of sound over the cantilever, and get a very fast pulse response, means using short cantilevers made from stiff materials like sapphire and diamond. What happens then is that the mechanical relation between the cantilever and the armature is somewhat troublesome, because often the two are joined with adhesive rather than tied together. If you care about pulse response from the beginning to the end, it involves the interface between one material and another, so it is better to mount the coil directly on the cantilever instead of using an adhesive. Take an aluminum tube, put the aluminum tube in an armature, and wind the coil over the armature without any adhesive—but if you use adhesive, please do not use a very soft adhesive as do some manufacturers in order to damp vibration. Let there be one tight construction. When you solve how to do this, you may have a good mechanical system.
However, then you come to the magnetic field modulator. This is another problem. I think many manufacturers are mechanical engineers, know how to wind a coil, know how to make a cantilever, but they do not know how to modulate a magnetic field.
There must be a magnet, so they take a magnet. There must be something that moves, so they take an armature. The armature should be between the two poles of a magnet, so they put the armature between two poles. There is now a constant flux over the armature, but they don't care so much about that. That thing is vital, however; you should not have a constant flux over the armature. The pole-pieces should be shaped to give a linear function between the angle variation of the cantilever and the flux variation in the gap.
One of the disadvantages of an MM cartridge compared with an MC is that an MM cartridge, which has always a constant flux and a little modulation of that flux, also has stray fields within the cartridge body which obscure the small variations due to the cantilever movement.
Finally, the coil winding is an electrical device. That gives three essential aspects of a cartridge: mechanical, magnetic, and electrical. If the magnetic design is wrong, then you never get the good electrical signal; it's also wrong. So you need to pay a lot of attention to the magnetic circuit before you can achieve a good electrical signal.
Atkinson: Are you saying, then, that in general this is something to which cartridge manufacturers haven't paid a great deal of attention?