I'm reading different things about grounding and trying to get a straight answer. The Nano has 2 pins for GND, presumably because the 328 has 2 pins for GND. I don't know if these pins are electrically isolated from each other in the 328 circuitry or not but is there any benefit in putting power on 1 and signal on the other? If the GND pins are internally connected and are no different from each other, how can I keep power noise off my signal ground, on a nano which is mounted on a proto board?
GND is GND. There is no isolation whatsoever between them.
Also, if your project is a small one without small amounts of wire length, connecting power GND to one of the nano's GNDs and signal GND to the other nano GND is fine. But, if you are running a big project with lots of wire, you would want to use what is called a "star ground" configuration and only use one GND pin.
If you want to keep noise off the power rails, add some big 10uF bypass capacitors between power and GND. They do a lot to reduce noise. You also can reduce noise by keeping wires and traces small and using twisted pair wiring for power and ground. If about noise even more so, you could also use conductive paint over your wires to shield them in a similar way to how coaxial cables shield.
Nano has a ground plane.
SMD 328P has 3 GND pins on the chip.
"how can I keep power noise off my signal ground, on a nano which is mounted on a proto board?"
Pay attention to how wires are routed on the protoboard. Keep signal lines away power lines. Avoid long stretches of parallel wires when some of the wires have high speed clock signals.
I see, so there's no built in protection and I have to use best practices. I laid out my components so that the power runs around the perimeter of the board in a 'C' shape. Here's a couple pictures. Soldering is incomplete but you can see what I've done so far.
Power_Broker:
GND is GND. There is no isolation whatsoever between them.Also, if your project is a small one without small amounts of wire length, connecting power GND to one of the nano's GNDs and signal GND to the other nano GND is fine. But, if you are running a big project with lots of wire, you would want to use what is called a "star ground" configuration and only use one GND pin.
If you want to keep noise off the power rails, add some big 10uF bypass capacitors between power and GND. They do a lot to reduce noise. You also can reduce noise by keeping wires and traces small and using twisted pair wiring for power and ground. If about noise even more so, you could also use conductive paint over your wires to shield them in a similar way to how coaxial cables shield.
I've heard of the star thing. I don't want to make it impossible to service my board by hiding it under a cobweb of ground wires but I may be able to do something like that. Do you think the size of my board qualifies or is it too small to worry about?
I have everything from 1 to 1000 MF caps so I could bridge the output terminals of my dc converter with it, if that would help.
Twisted pair may not be possible on the board since most wires are single wires going off in their own direction, not really pairs.
Conductive paint? That's cool.
So based on the pictures above, is there any strategy that would work well for that layout?
Perhaps give a description of all those components/modules you have on the board. What voltages are present where and what kind of signals. It looks like the nano is surrounded by noise with minimal ground tracks.
Sure:
Starting from the top right you have the supply voltage IN, which I designed so that it could be pretty much whatever, anything from about 9-16V should be ok. That goes through a DC-DC converter across the top and gets boosted to 25V on the top left. Then it goes into a high power MOSFET rated at 9 amps just below it on the left edge and that's driven by Arduino PWM to create a variable internal power supply, which travels down the left edge, underneath the OLED display and over to the current sensing unit at the bottom left. That tells the nano how much current is passing. From there, the power goes out of the board and into a battery. Alternatively, if the battery is set to discharge, the current from the battery will dump further to the right across the large gold resistor via a power MOSFET at a lower current. All the shit in the middle is voltage dividers, filters, voltage regulators and switches for the arduino inputs and outputs... all signal stuff. This, IMO, was the best way to keep the power as far away as possible from the more sensitive signal sensing areas.