I'm building a circuit on a breadboard with a microcontroller and a DAC. My breadboard ground/Vcc planes are set up like this: (The numbers indicate which lines are wired together, the supply voltage and ground are connected to the top rows.)

1----------- Digital ground
2----------- Vcc
||||||||||||
||||||||||||
1----------- Digital ground
2----------- Vcc
1----------3 Digital ground
2----------- Vcc
||||||||||||
||||||||||||
-----------3 Analog ground
2----------- Vcc

The Vcc buses are wired together, as are the digital ground buses. The analog ground is connected to the digital ground in only one place.

However, I'm getting lots of noise on the digital ground (amplitude between 100 and 200 mV). The noise on the analog ground is not as intense, but when a certain pin changes state on the microcontroller there is a glitch that is very noticeable on the analog ground.

How can I reduce noise on the ground lines, and prevent the digital glitches from leaking into the analog portion of my circuit? I know breadboards have lots of parasitic capacitance and shouldn't even be used for high-frequency digital circuits, but it's the best I can do for now. I'm using a regulated DC power supply at 5V and the voltage is very clean; the microcontroller is likely responsible for all the noise.

I can post scope readings and a schematic soon.

 

What is the power supply? Do you have a liberal sprinkling of bypass caps between the Vcc and Grounds?

Is the noise signature analog or digital?

Does it appear with no load, or heavy load?

 

Bypass caps are the most likely culprit, IMO.

 

I'm using a benchtop DC power supply. I added a couple more bypass capacitors and was able to isolate the analog ground from much of the high-frequency noise.

However, I'm still seeing spikes on the analog ground whenever one of the microcontroller I/O pins changes state, even though I have an 0.1uF bypass capacitor across its Vcc and ground pins.

On my oscilloscope, if I connect both ends of the probe to analog ground, it looks like the amplitude of the noise is now negligible (about 10 mV, identical to what I see when the circuit is powered off) but every few microseconds I still see a spike with an amplitude of about 50 mV. What else can I do to isolate them from the analog portion of my circuit?

 

I had a similar issue when breadboarding digital circuitry with analog instrumentation amplifers during a recent project. The solution that worked best for me was jumpering all the grounds together in a lot of places. Forget about trying to isolate analog and digital grounds on a breadboard. You need a ground plane to effectively reduce loop inductance.

For breadboarding, you need the lowest "resistance" grounding you can achieve to minimize the I x R drop in the wiring. Run jumpers on both ends of each Ground strip to the ends of the next Ground strip. Add jumpers in the middle of each Ground strip to the middle of the next Ground strip.

Do the same with each of your VCC strips.

Also, instead of one VCC wire and one ground wire from the power supply use four VCC wires and four ground wires from the power supply. Twist all 8 wires together. Fan out and connect one of the VCC wires to each of the VCC strips. Fan out and connect one of the Ground wires to each of the Ground strips.

creakndale