As per the incessant nagging of one Alberthrocks, I have added a little link activity module to my six-calculator CALCnet2.2 hub, as well as drawn up a schematic and taken a picture of the unit in action. You'll notice that the schematic below has two identical two-transistor halves, one for the clock line, one for the data line. The first transistor in each pair functions as a logical NOT gate, because the calculator hardware specifies logical high on 0V (electrical ground). The first transistor therefore inverts this logic so that a logical low means no current through the resistor connecting the two transistors in each pair. The second transistor in each pair switches the associated LED on or off; just about the simplest transistor circuit you can make.

So one LED blinks with the clock, and one with data? Is that data in and out to/from any calc?

Indeed. One LED lights when any calculator pulls the clock line low, and the other LED lights when any calculator pulls the data line low. Because CALCnet2.2 follows a bus model, there's no way to differentiate which calculator is responsible for the link activity and thus have N different pairs of LEDs for an N-calculator network.

I'm sure this has been addressed somewhere else, but what's the max N? And is it a soft limit or a hard limit?

Unlike I2C, for example (which has a 7-bit address space and therefore has a hard limit), a CALCnet2.2 network is limited only by practicality. It uses the 6-byte/12-hex-digit/48-bit address space defined by the ID of each calculator on the network, so the only limitation is the superlinear increase in collisions on an uncoordinated network as the number of calculators grows. This can easily be overcome by linking multiple networks over a gCn bridge, should such a thing ever be created, as the higher-speed, higher-processing-power bridge would introduce one calculator's worth of collisions and overhead in exchange for M calculators added to the existing N calculator network. As you can see, I'm testing a six-calculator network right now, with no significant problems found so far.

Ah, very nice work indeed. I am excited to build and play with this. Is there a parts list yet available?

You could also use a single PNP transistor instead of the two NPN transistors.

You could indeed, but I couldn't find where my 2N3906's had disappeared to.

As diagrammed, it is:

(2) 330-ohm resistors, 1/4 watt
(2) ~2.1V LEDs of your choice
(4) 10K-ohm resistors, 1/4 watt
(2) 1K-ohm resistors, 1/4 watt
(4) 2N3904 NPN bipolar transistors
(-) +5V supply (I'm stealing it from an Arduino here)

I mean for the whole thing.

The rest of it is just three I/O cables cut apart and spliced with some random 3-conductor audio cabling that I had lying around. It requires no power or components at all. I also used 0.1"-spaced headers to neatly breadboard it, but that's not necessary.

Is the Arduino not part of Calcnet?

Hell no, that's just to help me debug things by allowing me to probe the network via a computer-side logic analyzer. CALCnet2.2 is in its simplest form simply connecting the I/O ports of all the calculators in parallel. No external components, no external power.

Nice.

Yeah, I felt that was necessary to make it more widely-accessible for people who have no EE background and don't want to buy hardware in order to use CALCnet. I'm disappointed that it looks like I might need something other than the standard SilverLink to create gCn. :/

Awesome stuff here!
This probably isn't a link hub yet... but still something awesome!
I would probably use those tiny bright LEDs (those 1mm ish rectangular ones), combine it with either a reflector or a cut-through logo.

Next would be USB sockets.... if you ever implement that for Calcnet2. (As I've said, contract BrandonW for that work. At least it's better than asking some random person in India to do it! No offense to anybody who is Indian...)

I hear ya, but I'm much more concerned with final debugging (like your reported eventual-freeze issue?) and whitepaper-writing and hopefully Doors CS integration than prettyfying the hardware that very few people have the parts/expertise to build at this point. What do you mean it probably isn't a link hub? Do you not see the six link cables coming out of it? Or do you mean a gCn host?

I seriously, seriously doubt that I'll do it. Every calculator has an I/O port, right?

Ahh, ok. I saw the six wires, but 1) didn't see the connection areas in the schematic (I need to brush up on electronic schematic reading); 2) those don't look like 2.5mm sockets to me.

...and not everyone has an I/O cable. About 90% of my school has a TI-84+ [SE], and from those 90%, only 0.5% has an I/O cable. (You should know who) I don't think they will be buying any I/O cables soon... (Gotta love TI's inconsistency with everything, eh?)

But yeah, I seriously think that USB for Calcnet2 is a good idea. As I've said in my previous post, BrandonW can handle the fun [when he get's time].

Ah, because I didn't see a point drawing 6x3 = 18 wires into the diagram where it's just six clocks bound together, six datas bound together, and six grounds bound together. I cut apart three I/O cables into six endpoints; note how the connection point of each of the six wires to that breadboard breaks out into a red, white, and copper wire.

Ahh ok. I'll have to do some thinking for that then.

USB SUPPORT! USB SUPPORT! BASIC SUPPORT! BASIC SUPPORT!

Do you think it's stable enough for me to begin building?

Nice subtle subliminal messaging. I don't see any reason for you not to start building; it's not like the hardware design it going to change any time soon. The only thing that might happen is software improvements and bugfixes, but I don't see that changing the hardware design.

In other news, hours of patient soldering yielded me two more fully-functional TI-83+ calculators out of my piles and piles of calculators with "dead" LCDs, except now I have seven fully-functional TI-83+/ TI-84+/ TI-84+SE calculators, and only six endpoints in my network. Who wants to mail me their spare I/O cables?