I need a 5V supply for the Arduino and a 12V supply for some lamps powered through a UNL2003. I know they need to share a common ground for the Darlingtons to work.
Does sharing grounds cause problems? I once shared a wallwart ground with a ground main (from a house plug), and the wallwart got toasted. It was a simple test, and I have no idea why it got toasted, but it makes me worried about sharing grounds on 2 wallwarts.
Thanks in advance.
As long as they are fully isolated wall warts it's not a problem. But if they are not (for example have a connection to mains earth) then it can cause problems. When in doubt, measure continuity between the mains and the secondary side for all possible connections.
Unless you use opto couplers or other isolated connections, all parts of a circuit must share a common line, typically named ground. Such a common connection is not dangerous, unless your wall warts have direct contact to the mains lines - what's dangerous by design.
If you don't want to use the on-board regulator, a step-down converter can be used to power the Arduino from a common 12V supply.
septillion:
When in doubt, measure continuity between the mains and the secondary side for all possible connections.
Maybe I don't understand what you mean, but I think that is how I toasted my prior 12V supply: by testing the resistance between the 12V ground and the mains ground.
DrDiettrich:
... use the on-board regulator...
That is what I am using, but I need a third power supply, for a distant circuit connected to this one in another room. Either I send over the 12V line along with the other connections (I have room in the Cat5, but I am worried about safety and noise), or I use another power supply in the other room. I'm not sure which is better. Any suggestions?
I'd use a separate power supply when the current exceeds a few mA, I.e. powering distant low-power sensors is okay. Higher currents often require a higher transmission voltage and a voltage regulator at the target side.
Remote circuits and power supplies deserve some considerations, for safe operation when the controller is powered off.
You're not going to damage a supply by measuring continuity or ohm (unless you do it weird wrong).
I would just sent 12V over and use a local regulator. A DC-DC converter if the load takes more then a few mA. Sticking to one mains supply keeps everything easy and happy when not everything is plugged in. (back feeding the Arduino, see: EEVblog #831 - Power A Micro With No Power Pin! - YouTube )
The remote circuit can draw up to 700mA. Does that mean I should go with a remote power source?
DrDiettrich:
Remote circuits and power supplies deserve some considerations, for safe operation when the controller is powered off.
What safety considerations are you referring to? The remote circuit will be in an office setting and will be left unattended for 2-5 days at a time.
septillion:
You're not going to damage a supply by measuring continuity or ohm (unless you do itweirdwrong).
I don't know what happened. I put the positive lead of the ohm meter on the ground of the wallwart, and the negative lead on the plug ground (3rd prong, the round one, in North America). After 10 seconds, the wallwart started smoking.
Feddar:
The remote circuit can draw up to 700mA. Does that mean I should go with a remote power source?
I would not sent that as 5V no. But like I said, sent 12V over the cable and use a DC-DC.
Feddar:
What safety considerations are you referring to?
The same as why I gave you the YouTube link 
Feddar:
I don't know what happened. I put the positive lead of the ohm meter on the ground of the wallwart, and the negative lead on the plug ground (3rd prong, the round one, in North America). After 10 seconds, the wallwart started smoking.
Impossible with a DMM alone, it just doesn't have enough power to do so. The 9V battery will boil befor you melt a power cord.
A option, you had it plugged in when you did the measurement (never a good idea when measuring continuity or ohm) and the supply isn't isolated or you used the Amp plug on the DMM. Just measuring it unplugged cannot kill the supply when using continuity or ohm mode.
Feddar:
I don't know what happened. I put the positive lead of the ohm meter on the ground of the wallwart, and the negative lead on the plug ground (3rd prong, the round one, in North America). After 10 seconds, the wallwart started smoking.
I have to say that you might believe that was what you were doing, but you almost certainly were not.
I would be surprised if you did not burn out the ohmmeter before smoking the wart, but most likely you actually accidentally short circuited its plug in the process.
If the "wart" has a two pin plug, then there would be no connection between any output terminal and ground (unless it was one of those notorious Chinese ones 
). If it has a ground pin, which generally only happens with laptop supplies, that will generally be connected to the negative output and will certainly be safe to connect together between two of them if they are plugged into the same dual outlet or "power board".
I am using 12V anyhow. So, I would be sending 700mA 12V over Cat5 along with my digital info. Would this be safe both for fire and data?
I watched the video. I presume you are saying that if the remote 12V supply is left on while the local circuit is off, then it could drive both circuits in unpredictable and potentially dangerous ways. But, I believe the UNL2003 has diodes that (I think) keep the 12V away from the rest if the circuit.
As far as the wall wart meltdown, the meter was not effected. It still works just fine. It was not plugged it, it was using a battery. The wallwart had 2 prongs. Maybe it was faulty, or I did something and dont know what it was.
For fire, yeah. I don't expect CAT5 to get warm with 700mA. Is that 700mA @ 12V? Or 700mA @ 5V (which is just ~300mA @ 12V )
But if you have a wire you can use 2 (aka a pair) for GND and 2 for 12V. Or if you do expect a to big of a lost, you can use a boost converter to up it to 24V, transport it and then buck convert it to 12V.
For the digital info, the 12V is not a problem. All the noise it can pick up from anywhere because of the long cable is a wayyyyyy bigger concern.
Feddar:
I presume you are saying that if the remote 12V supply is left on while the local circuit is off, then it could drive both circuits in unpredictable and potentially dangerous ways.
Yes, as in, power the Arduino in a way it's not designed which can kill the Arduino.
Feddar:
But, I believe the ULN2003 has diodes that (I think) keep the 12V away from the rest if the circuit.
You never told us what you are driving. If it's only ULN2003 it's indeed not a problem. But a one power supply solution is so much more easy, more elegant and cheaper 
And when driving the ULN, I would put a pull down resistor (of let's say 1k) on each inpout of the ULN on that side of the cable to keep them off good to make them more robust against noise/interference.
Feddar:
As far as the wall wart meltdown, the meter was not effected. It still works just fine. It was not plugged it, it was using a battery. The wallwart had 2 prongs. Maybe it was faulty, or I did something and dont know what it was.
My money is all-in on the latter
septillion:
For fire, yeah. I don't expect CAT5 to get warm with 700mA. Is that 700mA @ 12V?
Yes it is 12V 700mA. But drawing all that current would be rare. More like 12V 200-500mA most of the time.
septillion:
But if you have a wire you can use 2 (aka a pair) for GND and 2 for 12V.
Why 2 wires?
septillion:
All the noise it can pick up from anywhere because of the long cable is a wayyyyyy bigger concern.
I'm using Line Drivers and Receivers for this. Thanks.
septillion:
Yes, as in, power the Arduino in a way it's not designed which can kill the Arduino.
So, if I unplug the Cat5 (containing the power) on the local end at the end of the day, all should be well?
septillion:
You never told us what you are driving. If it's only ULN2003 it's indeed not a problem. But a one power supply solution is so much more easy, more elegant and cheaper
I see what you mean.
Again, thanks for your help.
Feddar:
Yes it is 12V 700mA. But drawing all that current would be rare. More like 12V 200-500mA most of the time.
If the Arduino side is less power hungry then I would just place the 12V at the ULN side and feed 12V to the Arduino via the CAT5 together with the control signals of the ULN
Feddar:
Why 2 wires?
More wires = more surface = more copper = less loss = less heat
Feddar:
I'm using Line Drivers and Receivers for this. Thanks.
If you want us to help more with the project it might be time to post a circuit
Feddar:
So, if I unplug the Cat5 (containing the power) on the local end at the end of the day, all should be well?
Relying on you unplugging things in the right order is a recipe for disaster.
septillion:
More wires = more surface = more copper = less loss = less heat
Cool
If you want us to help more with the project it might be time to post a circuit
I'm not sure if the drivers and receivers are set up correctly. They may need 2 lines for each SPI line.
That image is a litlllllllebit tiny :
septillion:
That image is a litlllllllebit tiny
No, actually it is huge!
Looked at the circuit and oef, where to start
Let's start with the overall circuit. It might be okay to ommit duplications but doing it a bit more clear would have been nice. Make clear what part you duplicate (box around it or something) and make clear how many times and to which pin.
Okay, the above is more of a "how to make a better circuit" but doesn't affect the circuit itself. So off to the drivers. And no, you did not connect them correct... They are differential drivers, whcih means they look at the difference between signals. So every connection needs two wires. The drivers have a Y and a Z output for each input. And in the same way the receiver has two inputs (A and B) for each output. So transmitter Y goes to receiver A and transmitter Z goes to receiver B. So to transmit data, clock and latch you need 3 x 2 = 6 wires. And it looks like you want to transmit data back as well so you need an extra 2 makes a total of 8...
So you have no wire left for power at all in a CAT5 wire. And if I look at the circuit in figure 13 of the receiver that's a problem because they need a shared ground...
And you also need to include a termination resistor on the receiving side. Rt in the circuit diagram. And if I look at the truth table (page 2 of transmitter) I would say at least G should be pulled high or !G should be pulled low.
You might be able to free up a pair by using the shared clock and latch (or data and clock) hack or even a 1 wire hack. See http://www.romanblack.com/shift1.htm But you have to watch timing if you do.
What I think is an easier solution is to just use an extra Arduino (Pro Mini or Nano for the win) and use a RS485 driver. In full duplex (so both ways without waiting) you only need 4 wires. The rest is just serial then Now you have 4 free wires for power.
Next you only show a single LED connected to an output (but I think you mean times 8) without resistor etc. A led @ 12V will only work very shot So what's really connected? Even more because you say you need 700mA. A single led (with resistor) does not add up to 700mA @ 12V...
So what is the purpose of the device?
I can zoom in pretty easily. Can you?
Update:
Sorry, I didn't see page 2. Thanks for the input. I'll respond when I get home.
septillion:
You're right! Isn't clickable so didn't look further.
That would involve a few more mouse clicks. 
