USB signal interference

I’ve been testing the pathfinder BMS with various equipment, and found that the USB connection is vulnerable to electrical noise from a big cheap inverter. We noticed this problem in the past with the JBD BMSs as well. (This doesn’t affect Wifi or Bluetooth)

This interference causes the USB port to stop communication, and the BMS must be rebooted to get it going again. We are working an getting the firmware to automatically restart the USB connection, but it doesn’t work yet. Remember, all Pathfinder BMSs can get firmware upgrades via OTA or UF2, so everyone will be able get the firmware solution later.

Other than getting a better inverter, the best solution seems to be adding ferrite chokes to the USB cable. I found that 2 or 3 snap on ferrite chokes is enough to stabilize the connection. Wrapping the cable 3 times thru a large ferrite ring had the best results, almost completely eliminating the interference when measured on the PC side.

We will do some experiments with adding common mode chokes to the Pathfinder circuit board and possibly add them to the next revision.

Better quality inverters produce far less noise, so this isn’t an issue for most customers. For example in the past we found that a Renogy inverter produced the least noise out all the units tested in our shop.

Details and scope traces:

I measured the signal interference using an oscilloscope at the PC end of the USB cable. Not much to see here. Yellow and Blue are the signal lines, Data+ and Data-.

The Inverter produces pulses of RF interference, and between pulses, the USB signal looks fine.

Here, I’ve zoomed out the time scale and added the purple trace on the inverter’s DC terminals. The pulses are happening at 21kHz, which is the inverter’s switching frequency. On the left side you can see a USB communication packet, which occurred between pulses of interference.

The real problem happens when the pulse of interference causes ringing on the USB signal lines as seen here. USB signals are mostly differential, but both lines pulled low means EOP or end-of-packet. This time, the ringing interference pulled the blue trace low enough that the USB controller would read an EOP signal.

Here, I added 3 snap on chokes (left), or wrapped the cable thru a large ferrite ring choke (right). In both cases the interference is dampened enough to prevent spurious EOP signals, but the large ring has the best results by far.