I recently learned about the drain wire in a shielded cable. I'd always thought this was a ground wire, but it isn't and typically shouldn't be used as such. And as I understand it, it should only be connected (to earth ground) at one end of the cable. This "earth" grounded drain wire apparently completes the creation of a Faraday shield with the shielding in the cable - when connected only on one side. But, I don't completely know how to do this in a real world scenario.
I have an arduino that I'm connecting to several external devices (thru optocouplers - to avoid noise) to interface with a switch and an LED in the devices, using a foil shielded cable with a drain wire (to prevent interference). The arduino itself will be powered by POE thru a switch. The individual external devices all connect into a third device to get their power, that's not directly connected to the arduino circutry - so I think this means all the external devices are sharing the same common ground with that additional device. That device is powered by a cord plugged into a wall outlet. The POE switch has a ground connection on the back via a screw, which I've never connected to anything - I only just discovered it actually. Plus the POE switch is connected to a power outlet as well. In my limited knowledge I'm thinking that connection on either end of the cable might not make any difference since each side I think is connected to "earth" ground thru the power outlets. But my instinct is suggesting that connecting the drain wire on the side of the external devices, rather than at the arduino side.
Anyone have a better understanding on how to do this properly to ensure the shield is actually working as an interference protector to the wires in the cable?
They way I read your description, you are creating a VERY large antenna to pick up and/or radiate all kinds of noise.
If you are in North or South America, beware if your project goes from room to room, the mains wiring in each room will be up to 240 volts between the rooms, with only the common of the power protecting the system. Use a DVM to check the mains ground to your system ground. Should be zero to a milivolt or so.
The shield is usually ground and usually a "local ground" which may, or may not, also be earth ground.
In a "typical" application with 2 wires and a shield, one of the regular wires is ground. The shield can be connected in parallel and grounded at both ends but the shield is effective if it's only connected at one end.
In a balanced audio application (like a pro microphone) there are two differential signal wires and a ground-shield. And in the case of a microphone, the ground is connected to the microphone body to shield the internal components so both ends of the shield have to be grounded. And again, it's a local ground usually with a path to earth ground.
If you have DC signals and low-enough impedance/resistance you usually don't need a shield. And with DC, you can filter noise with a capacitor.
AC noise can get-in capacitively or inductively and the higher the impedance the easier it is for noise to get-in.
Typically in industrial panels the DC common and earth ground are tied together in each box, but only one panel connects the earth ground to the earth. This prevents ground loops and increased RF interference.
In your case the earth ground will be in your main fuse panel. I'd just connect the shield to your earth ground but don't tie the DC common to it. Only connect the ground at one location or the other to prevent loops.
Well this brings up a wonderment I had. I was initially thinking to use ethernet cable (or 4 wire phone cable) to connect between the external devices and the arduino. I knew that the twisted pair cable concept is something that is dealt with by electronics at either end to reverse each pairs errors, which exists in network hardware, but would not exist in my use. So, I got thinking that because some of these will have 40' runs, vs others that will likely be only 3', that I should look at shielded wire since the long runs could be near power cords, etc.
So, are you saying that I don't need to consider shielded cable for this use afterall? That'd be a plus because I need to use as thin a cable as possible to feed into the external devices existing connector thru a strain relief... and I'm looking a 2.5 - 4 mm outside diameter, 3 conductor, cable currently, with the expectation that I'm gonna need to be closer to 3 mm to fit. Not needing a shield opens up more cable options.
I was intially getting false readings (though I've not yet introduced optocouplers into the circuit) where arduino thought that a switch was being pressed for very short periods of time. When I added code to check the time of these, they were often close to 0ms duration. So I added code to detect if a press was less than 100ms duration, then ignore it. That resolved the issue... but I do realize it's a band aid not a fix. I was waiting for optocouplers to arrive to test further... they just arrived today, so I'm about to test that.
These are all just buttons? So straight digital signals of 1 or 0? What you are likely seeing in the arduino is "button bounce". It has nothing to do with the length of your cable but the way that mechanical switches (or buttons) work. Whenever you make a physical contact between the two parts of a switch it is never a static connection but will bounce from connected to unconnected and back several times before landing on the final state of connected or not. That is why with buttons you have to "debounce" the signal.
What are you using Opticouplers for? You would need to have a power source on the other side of the opticoupler in order to read a button push.
If you are connecting other "devices" to the arduino, an opticoupler could be the absolute worst thing for the connection. If the devices are I2C for example, to go that distance you need to have something to boost the current into the open drain system. There are boards out there that you could wire in that would let you communicate over I2C that far away. But you can't use an opticoupler for that.
Are these external devices AC or DC? When you say "Common ground" what exactly do you mean? I think you are conflating DC Ground to AC Earth ground and those are NOT the same thing. DC is a closed loop system. The shield in the cable might be tied to Earth ground but Earth ground cannot be tied to the DC ground in your device. AC and DC don't mix.
The false pushes happen without any button pushing at all - and I'm only testing at this point with about 1 foot of wires. The buttons are 5V when unpressed, and go towards 0 when pressed. I ahve no issues reading these when there's a real button press, it doesn't fluctuate at all, it's a clear push/release.
The point of the optocouplers is to isolate the power/ground at the external devices from the power/ground at the arduino. I was recommended to do this (by others on here) to resolve audio buzz that can be heard on those devices with headphones when the devices ground was connected to the arduino. When I researched, this seemed to be a very common thing to do.
I wasn't thinking too much into that, just trying to make sense of stuff I'm reading about. I'd tend to have thought that if an arduino, which i understand is DC, is powered by plugging it into a wall, or in my case, to a POE switch that's plugged into the wall... that in some way, the ground in the arduino is grounded to the wall... thus to earth. When I learned that a drain wire, to function properly, needs to be connected to earth ground... I wasn't thinking that I needed to extend the drain wire outside to connect to a stake in the ground. So, trying to understand how to make it work.
Are your external devices Open Drain collectors? Or are you tying your input pins to ground with pulldowns? Because you can't have an external device send a high signal and pull the arduino pin high and the same time. It's never going to be able to read correctly. Then you would need optioisolators. But again, you need to give more information about what devices you are connecting and how they are going to run if you want help figuring out problems.
In most cases, for shielded cables Curtiss-Wright/Parvus recommends shielding connected to CHASSIS or ground at only one end of the interface cable. In all cases, if the other end of the interface cable is to be terminated at all, it should be terminated with a capacitor between the cable shield and chassis ground
to avoid unintentional current loops. The value of the capacitor selected by the end user. Again, only one end of the cable shielding is connected to chassis and that end is at the device. Connecting both ends to chassis is likely to create undesirable current loops and contribute to overall EMI issues. Considerations for Aircraft
In some cases (as per customer requirements) it is permissible to ground whole cable shield harnesses at both ends, but only if the inner cable interfaces contain separate grounded shields connected to each at one end only. Such a case may occur when the system is being installed on an aircraft and is usually done for lightning strike considerations. Considerations include:
Lightning is more likely to strike aircraft than ground vehicles.
System failures on expensive aircraft are more likely to cause serious safety issues.
In aircraft, for purposes of lightning protection, the entire airframe is to be considered GND.