Ground Noise

[jnnewton]

Hello, I have a project where i am trying to control 3 servo drives on my mini-mill from a PIC microcontroller. The system works fine, except i have a problem with noise on the ground plane of my board (and in the enclosure, on the ac ground, in the shielding, everywhere!) The noise can be seen on my oscilliscope when the servo drives are enabled. How can i filter out the noise on the ground plane? The AC side of things is not grounded to the dc side, and all grounding on both sides is solid and all cables are as short as possible and all shielded. I am seeing the noise as about 5v pk-pk and is a spike every 100 us that oscillates down to 0 in about 10-20 us. Can you filter ground noise and how can i choose the correct components?

 

[RadioRon]

The first question that comes to mind is: why do care about some noise on the ground? Is there something malfunctioning because of this noise? If not, then it isn't a problem.

The noise that you see on an oscilloscope is the potential difference between the tip of your scope probe and the ground connection of your probe. Where are you connecting your ground clip? If, like so many apprentice engineers that I've supervised over the years, you did not use your ground clip at all, then your measurement of noise is not valid.

If you are convinced that this noise is everywhere, then you might call it a "common mode" noise. Your electronic circuits respond to the voltage difference between a signal line and local ground. If both the signal line and the ground have precisely the same noise voltage on them, then the difference between signal line and ground is 0 volts, and the electronic circuit won't care and won't respond. So, in some cases, common mode noise doesn't matter. This is especially true if you use balanced circuits, but that is another story. The trouble often is that the noise is not truly common mode and it is possible to get into trouble especially with highly sensitive circuits. Logic is not what I would call highly sensitive circuitry, since you typically have large noise margins built-in to logic.

 

[jnnewton]

I have taken ground from shielding, from the frame, from the ground pin on the servo, from the plug at my power strip, tied that to my ground clip on the scope and then used the probe to measure the noise on ground and or power. It is not there when the drives aren't enabled and starts when i turn them on. The noise is also on the signal lines, and may be "common mode". The problem is, my drives are picking up invalid pulses and my microcontroller is recieving invalid inputs from this noise. I am using both channels of the scope and if one is on ground, the other on the signal line, i can see the two coincide (the false input and the noise on ground.)

 

[RadioRon]

OK, the noise that is causing false inputs would definitely be considered a problem. Its not surprising that the noise may be caused by the current flow to/from the servo motors as these probably take a fair bit of current. Try to analyze where the currents are flowing in a circuit from the power supply to each motor and back again. Then, ask yourself, "does any motor current flow through the ground around my controller logic". If so, then you may have a problem of the two subsystems sharing a common ground too much. In this case, the motor current flow will suffer a voltage drop as it flows through ground, and this potential is impressed across the logic circuits.

A common way to try and deal with such scenarios is called a "star" ground topology, wherein each load that pulls current from the power supply has its own "To" and "from" wiring for power going directly to the lowest impedance point available. All Power+ wiring joings together at only one point and all Power- wiring joins together at one point. So, for example, when a servo pulls current, that current is not shared by the ground wiring of other subsystems, like your controller logic.

Usually, the lowest impedance point is the mains connection to the wall, or in the case of a DC system, the battery terminals.

 

[jnnewton]

Ok, i think i have a "star topology" except for where the I/O from the servo ties back into my logic. Current to the servo motor flows from the plug through a breaker and a contactor and back to the plug. The power for my microcontroller is coming from a dc power supply direcly pulling current from the "plug". The problem that i can see is that the servo drive's I/O (DC) have to be tied to the icrocontroller, which means that basically i have connected two of the "loops" via a shielded cable. I disconnected the shield from the microcontroller side and it seemed to decrease the spikes, but i still need to tie the ground pin from the servo to my logic in order to pick up the signals. This is the problem. If i disconnect the two and manually enable the drive via jumper wire, the noise on the microcontroller (dc side) drops dramatically. So, i need to isolate the ground on the signal cable, but am unsure how to do this.

 

[RadioRon]

Yes, this network of grounding is a common problem in most systems. You are really stuck with providing a signal ground across between the two subsystems because if you don't have this direct signal ground, the signal will simply find its ground connection through the power wiring and this is the worst/noisiest of all. The best way to fight this is to insure that the power wiring for the motor is considerably lower impedance than that provided by the signal ground. That is, the wires are fatter and you are encouraging the majority of the motor current to flow through the lowest resistance path.

At this point, I am not sure what to suggest without seeing a schematic, circuit board layouts, and wiring diagram. Perhaps it would be more efficient to seek out a local expert to go over this grounding problem with you as it is difficult to do remotely.