Robin Marshall: A Modicum of Genius Page 5

Atkinson: They tried to notch out its amplitude peak.

Marshall: And they made it sound worse. Most people I think were in agreement that the best thing you could do with an SL6 was to take that kill circuit out. That made it sound so much better.

Atkinson: I said earlier that you make your own drive-units, including the 26mm dome for the ES-14 and the 32mm for the old '20. I've always been told that tweeter manufacture is fraught with problems because of the very close tolerances required on something so small. You went into that . . .

Marshall: Foolishly.

Atkinson: Did it take you a long time to get a good tweeter into production?

Marshall: No, not really, because it's just an engineering problem. There's nothing difficult really about it. The only difficulty in making a tweeter is that all the masses are so low you've got to be very careful with adhesive bonding to make sure you're not changing the masses of things and introducing compliances. It's only an engineering problem. There's nothing difficult in licking that, providing you have the experience and the resources to do it. I think most people are scared off unnecessarily. Most manufacturers say "Well, we can make a bass driver, but a tweeter, hmm, no." They've never tried it.

Atkinson: Is there anything special about the ES-14 woofer?

Marshall: There's a lot special. It's a back-to-front design in that it uses a 17mm-long magnet gap and a 5mm-long coil as opposed to the normal system which is a long coil, perhaps 12 to 14mm long, working in a gap 6mm thick. This gives us linearity. It also gives us tremendous thermal stability because the coil is always totally enclosed by a huge amount of steel. Within the limits of sensitivity, you've got no real coil heating to worry about.

The only drawback is that it's horribly expensive. A 17mm-thick magnet plate is not exactly a cheap way of making a bass driver. The magnet system in our bass driver costs double the price of a complete bass driver in most loudspeakers. We pay $20 for the pieces in the magnet system. The average OEM 8" bass driver might be costing its manufacturer $10 or $12. Total, finished. Just stuff it in the box and there you are. So it's a masochistic way to make loudspeakers.

Atkinson: A loudspeaker is more than the sum of its drive-units, however. You have to pay very careful attention to what the box is doing.

Marshall: Yes, a lot of care goes into the box. It's strange, though, the ES-14 box is relatively conventional. I did so much messing around with different materials, different structures of boxes, so many exotic ways of making a box, but eventually came back to the conclusion that the best way of making a box was the good old conventional way—wood, nothing fancy about it. The only thing I do which is remotely different, I suppose, is to have a metal tie rod across the box.

Atkinson: One cabinet, two drive-units, you still need a crossover. Traditional British thinking has been that you aim for as much out-of-band rejection as possible. You use quite high slopes—18, 24dB/octave—and end up with very complicated crossovers like that in the LS3/5A. For some time, however, you've advocated that the less complex the crossover, the better. Why is that?

Marshall: I don't believe at all in the "let's get rid of the drive-unit as quickly as we possibly can" approach. Because if you're crossing over from a large unit into a small one, you've got a radiation pattern change straightaway. If you get this business where they're both rolling off extremely steeply so you've got a very fast change between two quite different polar patterns, that's very noticeable. The coloration generated by the off-axis performance of the speaker is going to be very noticeable. On-axis, it may measure as flat as a pancake; in theory, if you could listen to the speaker in an anechoic chamber, it might sound great. But once you put it into a real live room it's going to sound awful.