For a big machine (rope shovel) we put a custom electronics board (gps) in an enclosure. It is connected to the cabinet with Power over ethernet (POE). The POE switch is connected to an isolated power supply (115V AC to 24V DC, no ground at ac input). So, the ground from cabinet (machine) is not connected to our electronics board.
THE PROBLEM:
The unit works fine when no antennas are plugged in (3 antennas: 2 gps and 1 correction UHF signal). But, the moment, I touch with one of the antenna cable TNC connectors(basically the shield of the coax cable), an antenna connector on the box (basically the GND of the Pcb), the whole unit shuts down (instantly).
So, something is happening there. All antenna cables are correct. No shorts detected. On other machines, this unit has worked well for a long time.
Our gps antennas are situated on top of the machine. The signal comes down over a 100 feet (30m) coax cable. Now, the shield of the coax cable is connected to the frame of the antenna, so, when the antenna is installed, it is touching the ground of the machine.
Our debugging time on the machine is very limited, so I want to think this over.
My boss tells me to just connect the ground from the 24V, which is indeed floating, to the ground of the cabinet. I don't know if that would solve the problem, it seems like a hack, and, there would still be a ground loop. Besides, I am not sure if an electrician would approve to connect anything to ground.
I feel like we should isolate these antenna brackets from the ground. To simply keep things isolated. This is however more a hands off approach, as our mechanical designer has to create an isolation part out of plastic.
Now I am confused again, if ground of the device were isolated. Then, touching a ground should not have an influence on the first place? No?
Any thoughts? Thanks!
My next test will be to hold the antenna in my hand, instead of it being connected to the machine frame, to see if this would solve the error. To be continued.
tldr;
-gps antenna is connected to machine ground, gps unit has a floating ground.
The moment the two grounds touch, the device shuts down.
UPDATE (SOLVED):
-I went back to the machine. Yes, disconnecting the antenna bracket from chassi ground made it all work. So this was caused by the isolated power supply.
-a POE injector (with ground prong) was used and, this way, even when the antenna bracket was touching the chassis ground, all worked.
As a test: when I was connecting the "-" side from the (isolated) power supply to ground in the cabinet, the power supply failed right away (even if nothing was connected).
Now I am looking into what exactly is happening in a grounded power supply.
Best Answer
You change the circuit. You add a resistance and inductance to ground, and you add this anywhere you touch it. This can cause signal integrity problems if you do not touch a low impedance supply line.
If it's a switch mode power supply, you get it's common mode noise as a bonus over trough that connection.
And the entire circuits total voltage difference over the parasitic capacitance equalizes trough that connection at first touch.
Do you have specific need for the 24V to be floating? If not, then just ground it at the supply. Typically any exposed metal that can be touched has to be grounded. Unless you have a good reason not to.