First I apologize for asking a seemingly non-Keil related question. I have 2 evaluation boards (STM32 Olimex and another, home-made). Their grounds are connected, supply is separate. Both are connected to a TCP/IP network. I am sending a signal (about 56 KHz) from one to the other via a wire, and the decoding of that works, but not always: automated tests that verify the software fail (while manual tests work...) after programming an board chip on one of the boards plus an FPGA. When I say fail I mean that the target board does not send out a reply via the network indicating it decoded the 56 KHz successfully. I have noticed that I can rectify the situation by either connecting a probe to the signal pin at the receiver and the ground it to the common ground, or connect a ST-Link V2 to the Olimex board. This begs for to be grounding/parasite capacitance problem, but how can I determine that let along solve it? Thanks in advance
A much shorter signal wire seems to help a lot.
So consider use of bus driver chips - a transmitter intended to be good at driving a long wire and a receiver intended to reject noise.
more likely
the OP thinks a wire is a wire
Also, in case if more than one wire is used for connection, make sure that the wires are not very close to each other. A flat ribbon cable is a 'total No No' _disaster_, wont work. Closed wires generate stray capacitance _if the wires dont have proper shielding_.
Remember that floppy stations and hard disks have been using ribbon cables for a good many years. But it helps a lot to interleave data and ground lines to reduce the coupling between the signals.
a flat cable is no worse than wires equally close.
In many cases the worst you can do is "free" wires, no impedance matching termination is possible
even shielding can be deterimental
Dhaval probably has had a bad experience with a flat ribbon cable and makes the common mistake of, totaly and application independent, rejecting something that was misapplied and failed even when the best solution where correctly applied.
just remember "a wire is not a wire"
Both are connected to a TCP/IP network does that not mean CAT-5 and nothing else
Erik
"does that not mean CAT-5 and nothing else"
Nope.
TCP/IP is not hardware-dependant.
Were you thinking of Ethernet - specifically, x-base-T...?
I were.
this brings up how does the OP connect the two. just two port pins connected can only handle a short trace
Does not sound like such a low frequency / short wire can contribute a lot to wire capacitance issues.