Sun: There was a fair amount of working going into these rollers-- I think these original rollers were poron. (Poron is a urethene used for making cushions or bumpers.)
Yurchenco: Right. But when the idler wheel came along, we realized we could use hard rollers, and just went to over-molded ABS or polycarbon or whatever that material is. You wanted them rigid, because a change in diameter in a soft roller would affect the accuracy of the mouse: depending on how much pressure you put on the poron, because the ratio of the diameter of the roller to the diameter of the ball would change a little bit, and you'd lose accuracy. So it ended up that you had to go with a rigid roller for that reason alone.
Sun: Right. The bigger challenge was reducing the friction from the idle roller and how much energy it took out of the system. We also played some with stepping down the shaft diameters, but I don't see that in these models.
Pang: Wasn't there also a problem with the shafts bending if you dropped them from too great a height, and the stepped-down shafts were in part intended to strengthen them?
Yurchenco: Well, the idler wheel would never have that problem, because it was spring-loaded and would just move out of the way. This shaft ended up being quite well-supported in this design, because this thing is in this housing, the amount of travel this thing can take is pretty high. But I don't recall it ever really being an issue.
Sun: I recall there being some discussion, and we did some tests, but there weren't very many units who had that problem. Because if you think about it, you have to hit it pretty directly.
Yurchenco: In this design, there's a long unsupported length. And in this design here, the unsupported length is pretty darn short. And there's a tiny bit of clearance between these shafts....
Sun: I don't know who has those test results, but I remember getting them back and saying, "Oh, I'm glad it wasn't such a big deal."
Sun: If you make the ends of the shafts smaller where they are supported, it also reduces the friction.
Yurchenco: But we ended up not doing that because it was too expensive. We just went with a smaller diameter.
Sun: I remember working with Jim on the tolerance of the emitter-detector pairs. They just weren't made to be as precise as we needed them to be. Optron, who made these pairs-- we had a high reject rate at first. Eventually we got them to figure out how to make them to the precision levels we needed, but that was a challenge as well. The trade-off between precision in how much light and the alignment of the light with the molded lens, versus the mechanical alignment-- we were having conversations about whether we should do that, it's a very elaborate tolerance analysis to try and understand who should tightening up their tolerance-- whether it should be a tightening of the optical tolerance, or a tightening of the mechanical tolerance. But we were pushing the limits of the mechanical tolerance already, and they weren't going to get any better (laughs). So we went back to the optical system, and it's amazing it all worked out.
Yurchenco: It turns out you can have an incredible amount of sloppiness in the system and it still works. You look at mice today, and they're just slammed together. You go to a mouse line, and there people sitting there-- I've been in mouse lines all over Asia, and there was this level of precision we thought we needed, and maybe at the time we did, because of the optical components, but it's just gone out of the system now.