Use of galvanic isolation in EV project

[Trotters_Independent2]

Hi all

My team are entering into a competition which involves designing and building of a low speed electric vehicle. We are designing a multitude of different PCBs to implement things like telemetry, sensing of various voltages/currents/temperature etc. I have a separate post about the same project on this forum (see board to board connections in multi PCB project), where I am asking about how best to connect the PCBs together. I thought it best to ask that question separately as these are 2 different topics.

The vehicle will be powered by a bank of batteries that outputs up to 63V DC and up to 150A and two DC motors, controlled by an off-the-shelf controller. One of our PCBs will be interfacing with the motor controller and providing controlling signals - this includes an analog voltage acting as a speed demand and a few other digital IO. See attachment "01_Overall_Circuit."

I am hoping to implement galvanic isolation to eliminate possibility of any ground loops/ground-bounce and potentially remove noise coupled over to the PCBs from all the traction power stuff.

My first thought is to implement it something like in attachment "02_Proposed_Isolation_Use". The analog and digital signals between them would be galvanically isolated, that way, if the PCB1's "ground" is at a different potential to the motor controller's "ground", then this would circumvent any problems that could arise from that - I'm thinking opto-isolators.

My first question is: Does this make sense? Or is there a flaw in my reasoning? Or perhaps there would be easier/better ways of dealing with such a problem.

My second question is, would it be worth also galvanically isolating the 12V DC-DC converter, see attachment "03_Iso_DC_DC_Converter"? Could this have any benefits? I feel like it could be good from noise reduction point of view. But I am worried about safety, since it would essentially be floating, what would happen if the battery +ve accidentally shorted over to one of the PCBs? Since we are grounding the battery -ve to the conductive chassis of the vehicle, this could present a hazard if someone were to touch the chassis and the PCB at the same time. There must be a feasible way around that, but I don't know what it might be - does anyone have any insight? Is it worth pursuing?

[Andy Chee]

Just use sensible earthing e.g. star earthing, and you should be fine.

[moffy]

If you only have digital signals then use version 2 if you need analogue then version 3 and use differential output and input for the analogue signals. The controller in version 3 needs its ground connected to the motor driver board. If you can use PWM to communicate the analogue signals then version 2 works easily, but the filtering of the PWM will introduce a delay which can effect the control loop so take that into account.

[tszaboo]

Just use sensible earthing e.g. star earthing, and you should be fine.

Or just use the relevant automotive standards that the design team is required by law to follow

[Andy Chee]

Or just use the relevant automotive standards that the design team is required by law to follow

I had the impression that this vehicle was experimental and only to be driven on a closed circuit, as such, automotive standards and laws are optional.

But yes, if the vehicle is to be driven on public roads, then there are design standards that need to be complied with.

[nctnico]

Just use sensible earthing e.g. star earthing, and you should be fine.

Or just use the relevant automotive standards that the design team is required by law to follow

Either way, proper, singular grounding is super important. Having circuits floating typically makes things worse instead of better. Likely automotive standards will address this issue.

[Phil1977]

Can you tell more about the signal interface to the motor controller?

If you have an analogue or PWM interface I would isolate the signal path with an optocoupler in the PWM signal.

If it´s CAN or anything differential then it should be fine with common ground.

As mentioned, to keep whole parts of the system floating often is problematic by itself. It´s better to design the subsystems as galvanically isolated, and then set a well defined ground connection. If this ground connection is on 0V of both subsystems, you can use a non-isolated supply there.

[Trotters_Independent2]

Hi, yes it is not going to be driven on public roads and it's not a commercial product either. There is no specific standard that we need to follow that I am aware of.

[Trotters_Independent2]

Can you tell more about the signal interface to the motor controller?

If you have an analogue or PWM interface I would isolate the signal path with an optocoupler in the PWM signal.

If it´s CAN or anything differential then it should be fine with common ground.

As mentioned, to keep whole parts of the system floating often is problematic by itself. It´s better to design the subsystems as galvanically isolated, and then set a well defined ground connection. If this ground connection is on 0V of both subsystems, you can use a non-isolated supply there.

Thanks for your reply.

The motor controller takes a 3.3V voltage speed demand.

The are also a number of basic digital IO that we must apply 3.3V or short to GND (or each other) that serve as things like an "ignition", a safety cut-off and some other basic setup bits that determine running characteristics, eg: forward/reverse & some other things.

[max_torque]

I take it you and your team understand that a 63v nominal DC system actually falls under the "low voltage directive" in the UK (AC 50V, 60V max)

Ie you need to actually carry out all the same design mitigations and safety cases as for a system that runs at 1kV!  (the "Low voltage directive" covers the voltage range of 60Vdc to 1000Vdc, and broadly is aimed at putting in place suitable protections to prevent electric shock from direct contact (the "high voltage directive" covers from 1kV upwards and is significantly more onerous because at this potential voltages start to become capable of leaping out and biting you before you physically touch them!)


But even meeting the LVD is a significant enterprise, and one not to be taken lightly, esp in the UK where the HSE can and do allso apply the machinery directive, meaning any device "put into use" irrespective of commerical status) falls under this umbrella.


My suggesting is limit your DC link to 48v and avoid the enourmous headache (and potential (sic) jail time you might land yourself with...)

[max_torque]

BTW for your original question, with a 63v DC link, in the UK your device, even in prototype form, is required to have it's DC link fully galavanicaly isolated from the chassis / earth, to meet the leakage current and stored energy limits, and to actively monitor that isolation unless you can absolutely (ISO26262 level of rigour!!) prove there is no probable failure mode that can result in a failure of that galvanic isolation.

[Trotters_Independent2]

We were beginning research on this, but weren't sure where to start! I was hopeful, but aware that after we go above 50V or thereabouts, we might have to worry about things like this, so I made it so that the design can easily be changed at this stage to keep things below 50V!

Many thanks for lending me your expertise on this.