Received: from watserv1.uwaterloo.ca (watserv1.waterloo.edu) by karazm.math.UH.EDU with SMTP id AA20065 (5.65c/IDA-1.4.4 for ); Wed, 16 Oct 1991 19:52:33 -0500 Received: by watserv1.uwaterloo.ca id ; Wed, 16 Oct 91 20:48:34 -0400 Date: Wed, 16 Oct 91 20:48:34 -0400 From: Dave Stampe-Psy+Eng Message-Id: <9110170048.AA00603@watserv1.uwaterloo.ca> To: glove-list@karazm.math.uh.edu K. C. Lee (LEEK@QUCDN.QueensU.CA) replies: >I suspect the sampling rates of the glove can go much faster if the >glove CPU don't have to process the 6 sets of distances into (X,Y,Z,Rotation). >In that case the only limitation would be how fast can the receivers settle > so that the transmitter can send another ping. I don't believe the CPU is doing any special processing. I did some work on an implementation using an 80C196, and the math required a lot of 32-bit multiplications, divisions and roots, about 4 mS worth on that processor and out of the question for an 8-bit processor in any reasonable time. Also, if you compute the distortion that would be caused by just using the difference in delay between horizontal and vertical receiver pair response times, you get a response very like the one I've measured: about 50% better resolution at 100 cm distance as compared to 200 cm. I've also scoped the unit in operation and there isn't any extra time for calculations. The CPU waits for incoming pulses for 20 mS times 2 trans- mitters, for a total of 40 mS. This corresponds to the 25 Hz read rate seen if the finger timing circuit is defeated. BTW, 20 mS corresponds to a maximum echo path of 6.8 m (22'), which seems reasonable to me. Any less, and you could only use it in a padded cell! (B-{) Of course, the processor COULD attempt to do the math while it's waiting for the receivers to respond, but I don't think it is. -Dave Stampe