Well, I am planning on being an EE but that is a long way away. I have no knowledge of such things. although i can get allot of parts and info at school.

So i should be able to get almost all of the physical resources.

I'll see where i stand tomorrow when i get to school.

I know an EE and asked him. He looked at me and asked whether it was going into the EMP device I built (which was very fun, by the way).

You asked him if he had any PICs? You know two EEs, then.

I asked him if it was possible to make an infrared transmitter work with the calc.

I have wondered about this. I /do/ have a small antenna with a headphone jack that could function as the transmitter, I would just need a micro-controller to encode/decode the signals.

Of course it is. As I (I think) mentioned, many years before I had any of my EE degrees I built Sami's IR link and found it to be quite good, at least for what it is.

I know, this is a necropost I'm horrible like that.

Theoretically, a half-duplex radio/IR device should be enough, since the calcs themselves are only half-duplex. However, the transition for who is communicating would have to be directly controlled by the calc's own lines, right?

Why can't the calcs do full-duplex? Isn't there a way to directly control the lines from asm? That should allow for full-duplex, since there's 2 data lines and 1 ground, I think. IIRC the second data line is used to determine who is in control.

The calculators are not half-duplex; who said they were? The TI protocol happens to follow a half-duplex protocol, but there's no reason not to use one line for outgoing and one line for incoming, except for the timing issues. With the TI protocol, the second line is used for clocking. With CALCnet, it's used for a combination of priority ("everyone be quiet, I want to talk") and clocking.

Ah. That makes more sense. So, the calcs, using TI's protocol, can communicate at varying rates? That's interesting.

It seems like the calcs could establish a speed initially, then go full-duplex afterwards. Of course, for calcnet you need the network control a little more than a 2-calc link.

TI's protocol is somewhat interesting to begin with; the bit being sent is determined by which of the two lines is pulled low. Then the other line is pulled low by the receiver to acknowledge, they release their line in turn, and repeat. There is no innate concept of timing built into it except a 2-second limit for sending a bit and an ~250μs error signal.

Sloppy timing and large error windows makes the hardware cheaper, you can use an RC oscillator instead of a crystal, and save a nickle per unit

Dave