Mysterious Unwanted Signals

As you work with USB power instruments you will run into the problem of unidentified noise getting into your electronics. It can manifest itself as intermittent signals with no apparent source. But, if you connect an audio amplifier to your analog output, you will soon figure out the source of the noise: the computer.

Your ears are an excellent tool for tracking down problems. You will notice that whenever the mouse hovers over an icon on your screen the noise changes amplitude and/or frequency. Interestingly, an oscilloscope will not give a clear indication of the problem.

The noise gets into the instrumentation through the USB's 5 Volt power line. The solution is to power your instrument independently from the computer. I recommend staying away from switching power supplies.

Low-Noise Linear Power Supply

The schematic below is for a linear supply. Batteries are the ultimate low-noise energy source, but a line frequency transformer-based supply can be almost as good, and will never need replacing.

J1 is not needed if your instrument runs independently from your computer.

To make this power supply, I cut a USB cable in half and stripped the individual wires.

I plugged the USB-A connector (the one that goes into the computer into a phone-charger. NOT INTO MY COMPUTER. I didn't want to break it!

Using a multimeter, I determined which wires were ground and +5 Volts. Turns out it was the black and red wires. (Just like a car.)

Given that information I could wire up the supply. Before using it on my equipment I checked that yes, the regulator was putting out +5 Volts, again using the cheap phone charger.

It works.

Without the power supply, the output of the DDS Function generator is contaminated with obnoxious noises from the computer. With the power, the sound is satisfyingly clean. Victory!

I chose the LP2950-5.0 Voltage regulator because it will continue to provide 5 Volts, even if the battery is close to 5 Volts. However it is limited to 100 mA. For bigger loads, such as the dE0-Nano board, a higher power regulator will be needed. I need to look into what would work best. A 7805-family regulator might be a good choice.