Electromagnetic interference over microcontrollers

Thread Starter


Joined May 5, 2008
I have this schematic that I use to run some Gecko G201 modules (http://www.geckodrive.com/product.cfm?pid=9) and some triphase 380V contactors for a CNC machine.
The problem I face is that sometimes I get serious USB communication errors that force me to reconnect to the microchip USB driver losing important real time events. The microcontrollers hold on and do not reset themseves.
The communication errors come often when a contactor is starting a 2.5kW triphase motor but this is not an absolute rule. Sometimes I get errors for no particular reason.
I have tried to power the board with a battery, this helped me a lot reducing power line noise but even with the beterry powering the microcontroller (30cm wire length) I get communication errors.
Is there anybody who can help me optimize the schematic in order to overcome this problem?
Thank you!
ing. Rares Tohanean


Joined Feb 18, 2008
One of the best tools is an O-Scope. This will show what signals you have floating around on the circuit. Start with the power coming into the circuit to make sure noise is not coming in. There are lots of ways to cure that.

We had similar problems with some sensitive test equipment only to start looking and found out we had power spikes. They were caused by welding cables going through the ceiling over to where the workers were. But it did force us to take tougher measures to shield and ground our circuits.

One of the biggest cause of issue is not having a single-point grounding system. Next is proper shielding of circuitry and cables. Putting circuitry inside a metal box that is grounded will go a long ways.

Make sure your USB cable has a shield in it and that it's grounded as well. Those stand out like an antena picking up noise.

Also put an O-Scope on your USB line. You need to make sure you EMI stuff isn't jacking with any of the USB data. I've never had to use EMI filters so not sure how they will react with USB data lines. I've always had good luck with proper grounding, shielding, and good filter capacitors on power coming in and at power pins on ICs. I stress that because a circuit diagram shows filter capacitors but in location on actual circuit should be a close to power pins on chips as possible.