I'm working on a 2-part device (ECU and keypad) that currently use a ribbon cable to interconnect. I'm working on changing this to a true ethernet interface w/TCP/IP stack and everything.

The only catch is that I need to get power to the keypad device since it needs to power a PIC, LCD display, ethernet controller chip, etc.

The RJ45 ethernet standard only uses 4 of the 8 wires in a Cat5 bundle, but from what I gather many designs like to take those other 4 wires and ground them to provide EMI reduction on the line.

I'd like to be able to give users the freedom to disconnect the keypad and plug in a laptop and be able to access the ECU via a very simple web interface.

Where my problem comes in is that if I pass a +5v signal (or even +3.3v) over one of those wires than I risk sending that current straight into a laptop's ground.

Obviously I could put in a manual control such as a jumper on the PCB that would turn on or off the power on the pin, but that would be a royal pain in the butt to the user.

Is there a way to put in some type circuit or device that would cut off power to that pin/wire automatically if it sensed too much current was being drawn? What kind of effect would it have on a laptop if I dumped +5v into ground on what is probably a mostly 3.3v system?

Thanks,
Ray

 

how about not using a rj45 jack, and instead, provide the user with some interchangeable cables, one end is special, and goes to your ecu, the other end is rj45, for connecting to a laptop. the other cable would connect ecu to display, providing the power you need.

alternately, how about an rj45 jack and a display jack

or you could have the firmware test the type of connection when it senses a disconnect and reconnect ... power up the data lines first and try to get an ethernet link ... if it doesn't link up, then turn on the power lines ... you could have the firmware monitor current demands on the power lines, and say, if the power is more that 100mA, there might be a short and shut them down.

__________________
If you don't have a planet, what good are gold bars?

want to contact me directly? gmail gordonthree
check out my project website: http://projects.dimension-x.net
Favorite numbers:
09 F9 11 02 9D 74 E3 5B D8 41 56 C5 63 56 88 C0

 

There is a standard for providing power over Ethernet and it handles short circuit conditions due to non-compliant devices etc.
see: http://en.wikipedia.org/wiki/Power_over_Ethernet

 

Unused 10/100BT pairs are traditionally terminated into ~100 Ohms on the network card. Usually the two wires are shorted together, and the pair are attached to ground through the resistor. Obviously there will be variations. Putting power across one of the pairs is guaranteed to do something bad - most likely burn out the contacts on the connector.

A Gigabit ethernet card uses all the pairs, and in the unhealthy chance you plug into one of those, you'll see the input of a transformer. It's not exactly a good thing, but it'll probably survive...

The PoE (power over eth.) approach uses 2 pairs - each pair is shorted together and used as a power connection. This way it won't immediately nuke the transformer. The magical part of PoE is that powered devices put a 24K resistor across the pairs to indicate that it can accept power. Most normal ethernet devices should (and I haven't personally tried this) have a 0-200Ohm resistance between the pairs. The power supply tests for the signature resistance with a low voltage before applying real power.

Also, putting in a PTC that just fits the rating of the expected device is a good idea.

1) Apply a couple mA of test current, look for an expected resistance(i.e. two voltage comparators)
2) If you see the proper voltage, apply full power
3) If the link goes down, disconnect the power