Grounding strategies in equipment

DragonPetter

When there are multiple electronics sub-system interconnected, I have seen some applications where one of the sub-systems will include a current limiting resistor as well as some EMC related components at its ground input to protect itself.

The case being that this device is on a vehicle, and when another device on the vehicle lost its ground, it began to use the ground connection of the first device and caused damage to the device (before it had a ground protection designed into it).

Now, if 2 boxes are talking to each other on some kind of digital interface, it is necessary for them to share a common ground reference. However, it is possible if one device malfunctions and has a bad ground, it could cause the other device to malfunction as well, right?

Is it better to protect against the potential ground fault issue, or to ensure both boxes always have a solid common ground?

the_emi_guy

Dragonpetter,
This is a very big topic that spans issues of signal integrity, electromagnetic compatibility, and electrical safety.

On this specific question, it is often necessary to transmit a digital interface between boxes that do not share a good common ground. In the old days we used optical isolators. There is also differential signalling (Ethernet, plain old telephone, DSL). These are not totally immune to gnd noise, they have specified common mode rejection. Fiber optic links are immune to gnd differences.

DragonPetter

Thanks for your response the_emi_guy. I have been considering optical isolators, and differential signaling sounds like a great way to help this too.

If I can go with what you said, my impression is that it would be best to isolate grounds to maintain safety/protection and combine this with the above two techniques to retain signal integrity that might be compromised by the ground isolation.

Studiot

You should note that the requirements for logic 1 and logic zero have a range of voltage values (dependent upon the power supply value).

So the fact that one device is referencing its pulses to a voltage above another is not necessarily a problem. Obviously there will be less overlap for the window of acceptance.

DragonPetter

Yes, I'm aware of the noise margins between logic high and low, but there is still possibility for noise and bad ground to compromise that margin.

the_emi_guy

Just be aware of human safety issues with isolating grounds. If a person can simultaneously touch both of these boxes (directly or indirectly), and one of the boxes may be subject to hazardous voltage (lightning, power cross), you will need safety bonding. In other words make sure you validate your grounding scheme against any safety specs you are required to meet (60950).

DragonPetter

Thanks again. The sub-systems I'm considering are directly connected with ground already, and the highest voltage between boxes is 5V, although they both get 5V from an AC-DC power supply.

I am thinking of just putting a 120 ohm resistor and maybe a t-filter (with respect to 5V) in series between the two sub-system's grounds. I don't think these components will affect signal integrity, as the signals are just toggle/flag states, rather than data transmission. Optical isolation and differential signals are still of value though.

Studiot

Capacitive isolation?

yungman

I agree this is a big topic without any specificity. The best is if you draw a block diagram.

I deal with big systems. Without specificity, in general, I believe having a central ground point and use heavy gauge wire connect to individual module. As for interconnecting signal, I still use good grounding if I don't go to differential signals. In ribbon cable, I use alternate ground for more critical signals, for not as critical signals, at least one ground for every two to three signals. The important thing is how the ground on the ribbon cable connector grounded on the pcb and pcb to the module. To prevent ground loop for picking up EM noise, use ferrite clips of the ribbon cable.

If you have a high current signal going from one box to the other, I would not let the return signal go through ground, have a separate return for the signal so current don't travel through the ground.

For high noise environment, you might have to consider opto coupler or differential signals. I would not put a resistor between the two grounds of the two modules.