We are using Ethernet connector in Automotive High voltage components,
750V - 850V and 1000A.
I would like to know if the ethernet connection requires shielding or not, given the close proximity to the HV cables?
Could you please provide your feedback.
This article from Fluke should help.
4 Likes
Hello ManojMJR,
This problem is a real tiger.
Please allow me to add a few comments for our readers.
As I understand your application, you have a primary high-voltage battery, an inverter for propulsion, an inverter for cabin heating, and an inverter for the traditional 12 VDC accessories. You are adding control and monitoring functionality using a copper wire communication network.
There are so many things to consider when it comes to interference including:
• The nature of the signal(s) in the high voltage cable. For example, that 850 VDC level may contain significant AC components emanating from the switching inverter(s).
• The length of the wire run where the Ethernet cable is in parallel with the high voltage cable.
• The variable nature of the inverter. As a worst-case scenario, the energy in the HV cable may periodically conspire to corrupt data flowing in the copper communications cable. This result is an intermittent problem that is very difficult to replicate.
Unless I’m mistaken, there is no way to answer this question without physically building the system. You would then need to conduct a series of tests in varying conditions. Shielding is desirable. It may or may not help in this application.
To help illustrate this point, let me share a heat pump problem I encountered in my home. The pump featured multiple zones all powered by a single compressor. It worked very well. But occasionally it would start to squeal. It started with a whisper one unit. It would grow louder and louder and then move to the adjacent unit in the next room. By this time, it was objectively loud.
It took months with multiple service calls along with consultation with the OEM support team. To solve the problem the OEM sent a custom control PCB. Their solution was to program a dead band – a no go zone – for the controller. To eliminate the refrigerant oscillation, they prevented the unit from operating in certain conditions. Instead of being linear, the unit would jump over the zone of instability.
That’s the difference between things working on the bench and working in practice.
As I said, it’s a tiger.
Best Wishes,
APDahlen
P.S. Have you considered fiber optic control? This adds another set of problems but could eliminate the problem with interference. Another option is to use forms of communication with improved error correction and detection protocols. One such example is Reed–Solomon. This modification would slow down the data rate but improve integrity.