Network choices?

• Orthogonal frequency-division multiplexing.

• Wireless.

Excuse me? From Wikipedia: "the tightly coupled transmission wires will reduce susceptibility to electromagnetic noise interference because the noise will equally affect each wire and appear as a common-mode noise"

Over a 5m cable?

LVDS itself should work ok in a noisy environment, but be careful of the common-mode maximum voltages that the transceivers can withstand. It does sound a bit of overkill for what you are trying to do.

I was hoping to hear something like:

there is a transceiver chip, X inputs, X outputs,

Two of these chips communicate w/o any intervention, passing the inputs through one pair of twisted wires from the first chip to the outputs of the second chip - in first time frame, then doing the same in the opposite direction in the second time frame.

They're called FPGAs, but the configuration can be a bit of a chore...

Not enough information, sir!

"3 signals and input 2"? Three outputs and two inputs, maybe?

(all 1 MHz)? Do you mean that each one, at top speed, can have a 1MHz square wave (i.e., a 500ns on-pulse followed by a 500ns off-pulse)? Or do you mean that each one, at top speed, can have 1 mega-baud data on it? Or something else?

What is "relatively noisy?" Relative to what?

What is "too big"? Obviously, you feel that 28 pins is too big, but what's your threshold?

What system are you putting this in to? Are you trying to paste something onto existing boards with minimal changes? Are you doing a brand new design? Are you trying to hang this on the outsides of existing boxes without loosening a screw? If you're doing a new design, or a significant spin, what sort of communications possibilities are already on the boards (microcontrollers, FPGA, etc.)?

Damn, so no-one finds that sort of thing interesting enough to produce it.

Yes... didn't I mention that in the first post?

Ah, you got me there. Dunno. It all depends on where the owner will put the final product.

It can be in his house, but also in his garage, or inside a factory where he cuts steel. (for example)

When I just see a number, 28 - it looks big. (a DIP pops to my mind) Having checked the footprint, its only 10*8mm - not that big after all.

But its expensive, Farnell has it at ~10EUR for sender. Thats about 40 EUR for two-way comm - too much.

On one side I'm creating a new design, and on the other I'm inserting new components.

You did mention FPGAs here and in one other post in this same thread. Yes, I have a FPGA on the main board.

The slave board doesn't - I could put a CPLD there. Not sure, though, if the logic will fit inside the CPLD.

Guess I will have to try and see.

Depending on your latency requirements you may be able to do something with RS485 an one pair of lines, or RS422 and two pairs. If you could figure out a solid framing scheme, then cramming the whole thing into a CPLD may be quite doable.

Some FPGAs come with LVDS drivers -- but as its been ages since I've had to go shopping for FPGAs, I couldn't tell you exactly where to look (beyond the Xilinx and Altera web sites).

You might look at some of the RF stuff. There are RF transceiver/modulators which do something along the lines of what you're asking (and you can easily omit the antenna and just run wire between the sides), but it's not an area I pay a lot of attention too.

Is the latency and jitter critical? Can you just run an Ethernet connection?

I do not understand your specifications.

If you have two receivers, each of which expects data from three different sources, six pairs (12 wires) are required, not ten wires.

If the sources are synchronized in some way, a single RS-485 system (two wires) should do the trick.