Hello everyone, I am looking for grounding suggestions for a robot that we are building.
Let me explain the power architecture:
We have a battery pack which is 48v (150 AH)
This 48v goes to:

1. 7 Motor drivers cards.
2. To the input of a 5KW inverter. (Loads include off the shelf items)
3. To an isolated DC/DC buck which outputs 12v/30A.(Loads include driving 2 contactor coils, 3 MCUs, a raspberry pi, solenoids and 2 Lidar sensors. The 12v is further stepped down to 5 and 3.3 using non isolated switching regulators for the MCUs in their respective units).
4. To another isolated DC/DC buck which outputs 12v/15A which is used to drive a high power LED string.

The 5KW inverter output has 3 terminals, line, neutral and PE. The PE pin shorts to the inverter body(Checked with a DMM, also FYI neutral is not shorted to the PE pin).

Now the robot chassis is metallic and is in direct contact with the inverter.(Meaning that the inverter PE pin is connected to the chassis)
The Lidar sensors whose casing is also metallic are mounted on the chassis. Note that the lidar sensor module's negative shorts to its metallic casing.

Now the problem, since the robot does not have an "actual physical earth", i am considering the chassis as the main reference ground point.
With this in mind how do we go about grounding to power rails for safe and EMI friendly isolations.

1. Do i short the 48v ground to the chassis or leave it floating?
2. Do i insulate the sensor modules so that they don't touch the chassis, that way 12v ground is isolated?
3. Or do i short all three 48v ground, 12v ground and inverter's PE pin to the chassis, but at different points(if this is the case then won't 48 to 12v isolation fail)?
4. Any other possible solution ?

 

Why do you need 5000V in a robot?

 

I don't where did you see that? Inverter is 5KW rated. Apologies for the typo

 

If the DC to DC converter is a buck regulator then it is not isolated. If it is an isolated DC/DC converter then it's not a buck.

What loads are connected to the inverter? Does the inverter supply loads that are outside the robot?

 

If the DC to DC converter is a buck regulator then it is not isolated. If it is an isolated DC/DC converter then it's not a buck.

The DC to DC converter is an off the shelf converter from meanwell. As per the datasheet it is a isolated DC/DC (flyback topology)

What loads are connected to the inverter? Does the inverter supply loads that are outside the robot?

The inverter generally powers a pump, a paint sprayer and a drywall sanding tool

 

The DC to DC converter is an off the shelf converter from meanwell. As per the datasheet it is a isolated DC/DC (flyback topology)


The inverter generally powers a pump, a paint sprayer and a drywall sanding tool

The 230V part is fairly easy. Use a Residual Current Circuit breaker for all 230V parts. Connect the protective earth to the chassis, connect the neutral to the protective earth BEFORE the RCCB. Having the neutral earthed avoids the need for double-pole fusing whenever the wire gauge reduces.
I would suggest that you keep the main battery floating, and double-pole fuse it, but check that there are no recommendations for floating/earthed in the manual. The battery current will be in the order of 100A, with lots of 100Hz ripple, so keep that out of the chassis. Keep both battery leads as short as possible and as close together as possible. And you will need to double-pole fuse when you reduce the wire gauge to connect to the Meanwell PSUs.
I have used Victron 5kW inverters with B- connected to mains earth and floating and it seems to make little difference.
Connect the PSU 0V outputs to the chassis.
If the chassis is articulated in any way, and 230V cables run through the articulated parts, you should have earth bonding cables.

Other contributors may have other ideas. . . .

 

Thanks @Ian0
I have a few doubts in your approach:
For the AC part:
Are you sure about shorting the neutral to the earth pin? What we have now is that from the inverter output , Line and neutral to into a 2 pole MCB, the MCB's other end connects to a RCCB and from the other end of the RCCB the loads connect to it.
If neutral and PE pin are shorted and during any fault, for a brief moment the chassis would be 'live' before the RCCB trips?

For the DC, if 48v ground is isolated, Are you suggesting that we keep 12v ground from the DC/DC as the main ground reference?
Also thinking in terms of ease of debugging (oscilloscope probes , logic analyzers) , we would need isolated probes to probe 48v?
Also it was mu understanding that the 48v ground can handle the most 'abuse' or return currents during any faults. For example if a live wire accidentally touches the chassis , then the whole chassis would become live and the DC MCB may not even trip!

 

Thanks @Ian0
I have a few doubts in your approach:
For the AC part:
Are you sure about shorting the neutral to the earth pin? What we have now is that from the inverter output , Line and neutral to into a 2 pole MCB, the MCB's other end connects to a RCCB and from the other end of the RCCB the loads connect to it.
If neutral and PE pin are shorted and during any fault, for a brief moment the chassis would be 'live' before the RCCB trips?

Neutral and PE MUST be connected together for circuit protection to work. IT earthing systems are not allowed because they are single-fault-tolerant, and you have three appliances connected to the inverter.

For the DC, if 48v ground is isolated, Are you suggesting that we keep 12v ground from the DC/DC as the main ground reference?
Also thinking in terms of ease of debugging (oscilloscope probes , logic analyzers) , we would need isolated probes to probe 48v?
Also it was mu understanding that the 48v ground can handle the most 'abuse' or return currents during any faults. For example if a live wire accidentally touches the chassis , then the whole chassis would become live and the DC MCB may not even trip!

"May become live"? Live in respect to which ground?
You would not need isolated probes to test 48V as it will tend to float with the battery at ±24V with respect to the chassis. There is no reason that B- could not be connected to chassis for debugging. I think it is better if there is no way battery current could flow through the chassis in the even of a chassis connection becoming loose. Connecting B- to the chassis is much more of a debatable point.

 

@Ian0
Regarding the Neutral and PE connection, Have a look at page 8 of the document attached. This closely depicts my robots power architecture. Would like to hear your take on it.

 

@Ian0
Regarding the Neutral and PE connection, Have a look at page 8 of the document attached. This closely depicts my robots power architecture. Would like to hear your take on it.

Its purpose
is to bond the inverter's metal case to the rest of the robot's metallic structure. In the event of
an internal inverter fault where a "hot" 230V AC conductor touches the case, this ground wire
provides a low-impedance path for the fault current to flow. This massive current flow should
immediately trip the inverter's internal fuse or circuit breaker, de-energizing the output and
preventing the entire robot chassis from becoming live at 230V

That's just plain wrong. With neutral floating, there is no return path for the current. No current flows, you have an undetectable fault. When you say "preventing the entire robot chassis from becoming live at 230V", where are you measuring the 230V?

An artificial N-G bond can create parallel paths for
neutral current that can blind the GFCI to a real fault. Safety for the AC system is
achieved through the combination of the equipment ground connection and the
proper functioning of the inverter's built-in protection mechanisms.

Also wrong. All RCDs operate with the neutral connected to PE immediately before the RCD. It does not create parallel paths if done correctly.

 

First, that "green wire ground" is to prevent shocks from touching the supply housing if there has been a failure or damage. The items not connected with the AC mains power do not need a safety ground connection.
Providing a ground-fault circuit interrupter at the main AC connection should avoid any shock injury possibility. That scheme works very well, I have tripped a GFCI many times using a power tool in "non-ideal-conditions" and never felt any shock at all.
I am not sure just what folks mean when they speak of a "residual current". Is that a current that continues to flow after the power is switched off?? The term is rather ambiguous as I see it.

 

I am not sure just what folks mean when they speak of a "residual current"

Residual Current Circuit Breaker (or Residual Current Device) and Ground Fault Interruptor are one and the same.
The Residual Current is the difference between the live current and the neutral current. (i.e. the current that has gone somewhere that is shouldn't have)
When I joined this forum I didn't know what a Ground Fault Interruptor was.
And by the way, the green wire here has yellow stripes if it is the Protective Earth, which we seem to want to refer to as the CPC these days, but I'll still be calling it the protective earth.

 

When I joined this forum I didn't know what a Ground Fault Interruptor was.

AKA in some areas as an "Earth Leakage Trip" or breaker.

 

AKA in some areas as an "Earth Leakage Trip" or breaker.

The terminology changed when the technology changed. When they were called Earth Leakage trips they were operated by a coil on the protective earth conductor between the point where they were all commoned in a fuse box and the incoming supply earth. Obviously, such devices weren't very effective as they failed to detect faults that went to earth by any other means than the protective earth conductor, so they were replaced by the residual current device which was rather more effective.
https://flameport.com/electric_museum/old_elcb/index.cs4

 

Of the different names, Ground Fault Circuit interrupter is by far the most explicit. AND, given that they do trip when a fault current flows to ground, such as thru my hands, that is rather a benefit. Unlikely to be misunderstood.
One such device gave me a few hours of tracing effort exercise a couple weeks ago. It was hidden in an office and protected a quarter of the wiring in an older house.
What I favor is a GFCI device protecting that outlet only. Then when it trips the cause will be clear. A hidden GFCI protecting a large area, just to save some money, is asking for inconvenience later on.

 

No easy-to-understand acronyms here! RCCB got changed to RCD (Residual Current Device) just to make it more vague.
There are now RCBOs (Residual Current Breaker with Over-Current) which combines the RCD and MCB (Miniature Circuit breaker - just a thermal-magnetic over-current breaker - see what I mean about the vague acronyms). And then we renamed "fuse box" (easy to comprehend) as "Consumer Unit" (could mean anything). The term only stuck for domestic premises - elsewhere it's called a distribution board (not quite so vague). And the "protective earth" (obvious) got renamed CPC (which I had to look up, and stands for Circuit Protective Conductor, not to be confused with the electronics supplier of the same initials)

 

Residual Current Circuit Breaker (or Residual Current Device) and Ground Fault Interruptor are one and the same.
The Residual Current is the difference between the live current and the neutral current. (i.e. the current that has gone somewhere that is shouldn't have)

They were Earth Leakage trips when I left the UK.
They seem to be much more popular used in the main panel than N.A, which appears to favour the installation using individual sockets. In vunerable areas.

• • RCDs/RCCBs: These are standalone devices that provide earth fault protection.
    
    
  • RCBOs: These combine the functionality of an RCD and a Miniature Circuit Breaker (MCB), providing both earth fault and overload protection in a single unit.
    
    
  • GFCIs: While often used interchangeably with RCDs, Ground Fault Circuit Interrupters (GFCIs) are primarily used in the US and may have slightly different operating characteristics and applications.
    

 

Don't forget AFCIs, which make it impossible to use things like shop vacs and other power tools.

 

Don't forget AFCIs, which make it impossible to use things like shop vacs and other power tools.

Are they really that bad? When I saw that they were being recommended, I wondered how one would run an arc-welder.
They don't work on ring-mains because every socket is supplied by two different paths, it's not very good at detecting a loose connection on one side.

 

I was really puzzled about the basis of that whole effort. How many fires or shocks actually involve arcing wiring or connections?? In my whole career I have come across two instances in which an arcing connection was involved that did not result in an immediately failed connection or an immediately activated circuit current limiter. (FUSE).
Certainly those behind the arc-fault circuit breaker mandate must have had instances in mind, knowing what they were would be educational. It would be interesting to get reports on actual arc-fault incidents.