High voltage pulse interfering with USB

Thread Starter


Joined Jan 3, 2011
I work in testing at a contract manufacturer and we're having an issue with a test setup on a device that generates a high voltage pulse.

The test involves a box that includes a gas discharge tube and a resistor to output a 1000V, 1μs pulse to a high voltage oscilloscope probe:


The test setup also includes a MCU-controlled relay card connected to a PC via USB. When the pulse is triggered it randomly kicks the USB device off of the PC. We found that the pulse is inducing a noise pulse in the USB connections that can be as high as 50V pk-pk (measured on the VBUS line). It seems to be a cumulative effect because the noise will decrease a little with the removal of most of the connections to the DUT, individually. There are programmer connections and a trigger output to the scope.

The high voltage output from the DUT is isolated from the low voltage nets (including ground), and the circuit is battery powered. The connection from the test box HV test point to the HV scope probe is via banana jacks, and we found that the noise decreases significantly (but not completely) when no cables are plugged into the jacks. We tried putting the test box and HV scope probe in a grounded metal box but it had no effect. Also tried putting the relay card in a grounded metal box and it had no effect. I tried putting a TVS diode on the VBUS line at the relay card but it had no effect. I put a ferrite bead & capacitor filter in the middle of the USB connection and it reduced the noise a little but not enough. We tried a USB isolator and it made the problem worse. Also, the noise is induced in the USB cable even when BOTH sides are disconnected.

I'm tearing my hair out trying to figure this out. Any ideas on what we can do about this?