pepke@SCRI1.SCRI.FSU.EDU (Eric Pepke) (11/01/90)
The Spine-Chilling Saga of the Power Glove, Part 4 Since my last installment, I have done three things: gotten the switching power supply in the knuckle box to work, played with the fingers some more, and gotten a range counter to work. Reminder: The knuckle box has a 40 V switching power supply, controlled by a violet line that you kick and a gray line that returns about 2 V at high impedance when the voltage is right. Noticing that the 2 V that the feedback wire returns is amazingly close to the threshold I remember for TTL, I ran it through an emitter follower through an inverter into one input of an AND gate. I chose a grounding resistor of a few K, which works fine with LS, but standard TTL requires a bit lower resistance. Into the other side of the AND gate I put the ultrasonic frequency, which I have pretty much decided should be 50 KHz. (A lower frequency, such as 25 KHz, isn't enough to get the voltage up beyond 16 V or so.) I used the output to drive the violet line. The result? The voltage in the knuckle box stayed at 40 V on the nose. I did some more playing around with the resistive elements in the fingers. In the Power Glove, each resistance forms the top half of a voltage divider, with a 33K resistor to ground. I just hooked up the ohmmeter on the 1K setting to one of the fingers and played with it. The amount of precision that one can get is really quite good. The active range of the element extends from as far back as I can push my finger with my other hand (in my case about 80 degrees at the base) to as far as I can flex my finger. Bending very carefully and slowly, one can make very fine adjustments to the resistance. Clenching and unclenching the fingers rapidly does not appear to affect the endpoints very much. Each of the three ultrasonic recievers in the Power Glove goes through some circuitry to a TTL-compatible output. This is normally high when the reciever is quiet. When the reciever is going, the output goes down and up at some frequency, which I think is 50 KHz. It's not really important, because all you need is the first downward transition. I breadboarded a circuit to count the range. Once every second, it emitted a half-second burst of 50 KHz through one of the transmitters. When the burst started, a couple of decade counters started incrementing at 50 KHz. When the first downward transition was detected, they stopped. This, theoretically, should have given me about a half centimeter resolution. My reasoning for using 50 KHz was that it would give me precision of about the wavelength of the sound. I really have no idea whether this is a good theoretical limit or not. One could argue that the ultrasonic reciever needs a couple of cycles before it gets up to speed, so it might be. One could also argue that the number of cycles it needs to get up to speed should be constant, perhaps to a precision of a small fraction of a cycle. The recievers are VERY picky about their resonant frequency. If the transmitter isn't right on, the probability that a burst will kick the reciever is quite low. It was difficult to tune the 20 turn pot to the right frequency, and even when it was hit, there was enough drift to muck it up after a while. When I get more breadboards, I'll have to try it with a crystal clock rather than a 555. Nevertheless, when I managed to get it close, I got some good numbers. The best I could tune it, I got maybe nine hits out of ten. Although it was hard to check using my setup, it seemed to be quite accurate and precise. With a little software heuristics to eliminate big misses and some low-pass filtering, it should be perfectly adequate as a virtual reality controller. I am busy accumulating the parts for the Motorola MPU programmer. I think it will be fine for the job, presuming I can remember enough about asynchronous serial communication to program that part of the processor. It's an order of magnitude more fiddly than programming an 8251 or Z-80 SIO or something like that. One thing I do not like at all about the Power Glove is that the pad over the knuckles on the inside seems to be made out of a space-age synthetic equivalent of what medieval monks used for shirts. In order to endure it, I think I'll have to get a raquetball glove or something. Eric Pepke INTERNET: pepke@gw.scri.fsu.edu Supercomputer Computations Research Institute MFENET: pepke@fsu Florida State University SPAN: scri::pepke Tallahassee, FL 32306-4052 BITNET: pepke@fsu Disclaimer: My employers seldom even LISTEN to my opinions. Meta-disclaimer: Any society that needs disclaimers has too many lawyers.