Electronic – arduino – Measuring power consumption in Arduino and raspberry-pi system

Do you have the DC supplies and Arduino grounds connected together? If not then any PWM from the Arduino will not work.
Most small plug type DC supplies are floating, so their grounds will not be connected to USB ground (of course the USB ground may be floating if e.g. on laptop battery)

As Russell mentioned a diagram of your connections would be useful.

Thanks for the diagrams:

Your problem is as stated above, you need to connect the grounds together. Otherwise, the Arduino PWM signal has no return path (remember a voltage is always relative to something) to Arduino ground since the grounds are not connected.
It's okay to connect all the grounds together (arduino and DC supplies)
The only time is isn't is in the very rare case where you might have a supply "ground" that is above earth ground relative to supply grounds (e.g. a 20 V supply with V+ at 50V and V- at 30V relative to earth/AC ground could not be connected to a 20V supply with V+ at 20V and V- at 0V)
If in doubt you can test for any voltage between the grounds with a multimeter (e.g. one probe to one ground, one to the other ground) but since most DC supplies are floating (not referenced to earth ground) then essentially the outputs are just like batteries.

Hope I haven't confused things here :-) Connect them together and let us know how it goes. Here's an altered version of your diagram (note extra black ground lines):

You could indeed! But first things first:

• determine each device's power consumption and sum them all up
• make sure your laptop PSU can supply at least this much

Things to watch out for:

• Even though they may run on the same voltage, the Rpi and Arduino might not get along. Noise and ripple from one device will be superimposed on its supply lines, so to save you the hours of debugging I would STRONGLY suggest separate supplies for anything with a clock.
• Inrush currents may be several times greater than average currents (big current spikes are common when charging up lots of parallel capacitors on each board at initial power-on) so if you don't have the inrush current spec available from the device's datasheet, just double (or triple) the average/rated current to be conservative.
• If you use linear regulators, you won't have any extra clock frequencies to worry about, but you will have lots of heat. Try to find a regulator such as the 7805 which has its tab connected to ground. That will make it much easier to sink heat to a metal enclosure, if you end up putting the completed project in one.
• Aim for "star" grounding such that all your individual device grounds come to a common point which is as close as possible to the lowest-impedance ground reference. This might be the negative return on the laptop supply.
• Your audio amp very likely used bipolar supply voltages. If you don't have a true negative supply available this could be a problem. It might be simpler to find an audio chipamp such as an LM386 that runs off a single supply than to generate a negative supply locally.