I'm looking into a way to build a repeater for the link cable since even the shielded link cables I use go max 10m with full power. gCn wouldn't work since I don't have an arduino board or a computer with USB ports. So I am looking for a circuit that takes the signal from the calculator and using external power transforms it in to high voltage/current signaling. Or a simple repeater that gives me another 10m would also be fine since I could daisy chain them to get the desired length.

I know this isn't an answer to your question, but why not use more calculators? As long as each calculator is with 10m of another one, you should be fine because each calculator provides the same amount of power to the link cable. What are you planning to use this for, anyway?

This is not CALCnet, so you can't have more than 2 calculators.

I am having deja vu that I answered this question already, but you could do a variety of clever things. You need to amplify it in both directions without creating a feedback loop, which is rather complex with discrete components. Another option is two microcontrollers, connected by something by more resilient (Ethernet and switches?) in between. What sort of distances are we talking about?

Yep, I had the same feeling. http://www.cemetech.net/forum/viewtopic.php?t=8123&start=0
It's usually best to not make duplicate topics. If a question in an old topic that belongs to you hasn't been answered and you'd like it to be, then necroposting would be fine in such a scenario. Just an FYI.
The biggest issue with amplifying the signal over a longer cable would be deamplifying it at the other end.

In this situation i am aiming for 30m

Why is this hard with discrete components? Wouldn't a simple transistor or op-amp circuit work beautifully for this?

The calculator's I/O lines are bidirectional and implemented as open-collector - when not driven they float high (via a weak pull-up resistor) but either calculator can hold the line low. If the line is currently low that means that one or both of the calculators are holding it low - there's no way to tell for certain which calculators are doing the driving from the cable between the two.

If your cable is from A to B and point A goes low you would assume that you therefore need to drive end B low - however, as your cable is bidirectional and needs to carry data from B to A you would then see that B was low and would try to drive A low. This is the feedback loop KermM was alluding to, and is quite difficult to resolve.

You could avoid this by making each end of the cable speak the TI-OS protocol and shuttling data between them with an RS-485 or similar serial connection, but this limits the device to only working with the TI-OS protocol.

I understand that, but with a simple diode circuit such as the one in the Parallel link and two extra data lines you have 4 simple unidirectional lines rather than 2 bidirectional. And since many Op-amps come in packages with two amps per chip you can still have one chip per end.