USB connection with ground differences

SunSet-

We have TVS devices that we recommend to protect the USB2.0 480Mbit signal, but I'm a bit confused as to what you are asking about. Are you saying that you are worried current might flow through the D+/D- lines because of an uneven ground reference between the host and the device? As long as you have a connected ground line, that effect should be very minimal. What are you hoping to do with a TVS that would prevent this? Please give more details if I am not thinking about this properly, preferably with a drawn diagram.

As for a TVS to protection USB2.0, we generally recommend the TPD2E2U06.

Thanks,
Alec

Hello Alec,

The problem I may see, is current flowing on the grounds and not necessarily on D+ and D-.

The concerning is about the different ground potential of the device and the host.
While the cable between the two, accordingly to the USB spec, could be up to 5m (with good cables), the device might be connected to a power cable that might be quite long and distant from the point where the host/PC will be powered. This may generate a ground difference that can create current flowing on the two connected ground even if D+ and D- are not connected.
Typically to solve that you can isolate the power, but isolating the USB data bus at 480Mbit/s is not trivial.

So I was thinking if the TVS can solve such scenario. An isolated power and data line can allow keeping the communication alive while the TVS would short h

For instance connecting the TVS between GND and earth (as well the other lines) should create a current path to earth in case there is a voltage difference among GND. I'm not sure if you have seen such solution to protect the Host/Device from ground differences.
I was just thinking about it as possible solution but I'm not confident myself that can work in all scenarios.

Thanks,

SunSet

Hello Alec,

actually the ground difference could be quite a lot since it's not about the 5 m length but the 100m or more power cable.

Indeed the system gets the powered from point 2 while the PC from point 1 ( as shown in the system below).

maybe this was not clear from previous description.

TVS while may protect the system they will not allow the USB connection to work (if they will be activated).

Thus, to keep the connection working also with a big ground difference, it might be better to have a separated USB controller and isolate the TX and RX lines. The power would also need to be isolated.

Thanks

SunSet

Hi SunSet,

I think I understand the problem fully now.  What type of voltage gradient do you expect between Power 1 and Power 2 in an actual system?  Are these power rails going to be off of AC power, or a DC source?  If we're only talking a few hundred mV, then the system should work as-is.  To me, the best method would be a low resistance path between the 'System' and a ground near Power 1 via a ground strap or something similar.  That way the grounds should be well equalized and you wouldn't need to use isolation.  If the voltage delta is 1+ volts, isolation may be the best bet. Is it possible to just lift the ground connection on one of these two systems?  This would allow one of the two systems to reference itself to the other, and the current gradient across USB should be very small.

Using a TVS diode wouldn't be any better than a hard ground connection, since ultimately you want the ground gradient from 'System' to 'PC' to be a little as possible.  Otherwise, you'll have violations of the common mode range on the USB transceivers and it won't work anyways.