Should I connect all the grounds together in my optocoupler circuit?

[awaiting]

Hi everyone,

I wanted ask about the optocoupler circuit used to isolate a microcontroller and a motor driver board.Is it alright if I connect the input and output grounds together to make a common ground and connect it to the ground of the controller.
I was of the opinion that an opto coupler does not allow any high currents/voltages to go back to the controller in case of a malfunction of the driver board. But I am not sure about if should connect all the grounds together.
Please let me know your views about it.

Thanks,

 

[meBigGuy]

One major purpose of a optocoupler is to provide ground isolation. Otherwise you can just use an ordinary transistor. If you don't need the ground isolation, then you can connect the grounds. But, you need to be careful that you don't create ground loops or safety issues.

Perhaps you can some more information regarding the controller, driver and power supplies.

 

[awaiting]

Thanks for the reply!
I am using TMS320F28335 as the controller, the driver board consists of HCPL 4502 as the optocoupler and IR2104 as the MOSFET driver. I supplying 60VDC as the power supply and deriving the power supply for both the IC's using a DC-DC converter.
I am facing some problems in the communication between the controller and the PC. As soon as I switch on the supply, the code editor (CCS) freezes and gives an error message. I am unsure if the optocoupler grounding could be a source of the problem or some other noise. I also tried using a Ferrite core USB cable but it could not resolve the issue.

 

[davenn]

hi there

how about uploading a schematic of your project so we all can more easily see what you are doing
there may be an obvious ( to one of us) a design wiring error that can be corrected

Dave

 

[berkeman]

One major purpose of a optocoupler is to provide ground isolation.

Agreed.

But when I do a Google Images search, I find both coupled ground and isolated ground circuit examples. The connected ground images/schematics could be in error, but I need to think more about this and how transient currents from the load would be controlled. Anybody dealt with a coupled ground, opto-isolated motor drive circuit before?

 

[meBigGuy]

But when I do a Google Images search, I find both coupled ground and isolated ground circuit examples. The connected ground images/schematics could be in error, but I need to think more about this and how transient currents from the load would be controlled. Anybody dealt with a coupled ground, opto-isolated motor drive circuit before?

I don't think the connected ground examples are in error. I really meant that ONE purpose is isolating grounds. But, that's not the only purpose.

As soon as I switch on the supply, the code editor (CCS) freezes and gives an error message.

Do you literally mean as soon as you switch the supply on (and the motor is not being driven at that time). Does it freeze if the motor is not connected?
How is the supply grounded, including the mains?
How is the PC grounded, including the mains?
How is the motor connected and grounded?

It could be that as soon as you power up the controller it freezes the PC. You need to debug the system without the motor connected. My guess is that the controller is freezing the PC. The reason for that could be any of many, like FW errors, noise from the motor, etc.

The thing you need to avoid is ground loops, where currents from the motor cause voltage drops that are seen by other components.

The physical distribution of grounds and current flow needs to be designed properly. For example segments carrying motor current need to be isolated from other circuits, with a common tie point right at the battery.

 

[awaiting]

I uploaded a jpg of the routing. On top there are the power supplies, below it (from left to right) are the optocoupler, mosfet driver, one half bridge and the current sensors and the bottom part is the controller. It is a bit messy as of now as I wanted to test it on a single layer, then go for a two layered PCB.

Do you literally mean as soon as you switch the supply on (and the motor is not being driven at that time). Does it freeze if the motor is not connected?
How is the supply grounded, including the mains?
How is the PC grounded, including the mains?
How is the motor connected and grounded?

yes, it was the same in both cases. The PC and the supply were connected to the same point using an extension cord. I also tried connecting them to directly to the different points but there was no change.
I used a PMDC and a PM synchronous motor with floating neutral. For PMDC, one end was connected to board ground.

I think it is mainly ground looping is the main problem as you pointed out.
I separated the grounds on my board but the output of the optocoupler was very noisy. Although I made sure that there was no connection between the two grounds, the board ground seems to be affecting the input of the optocoupler. The input becomes nosiy after I switch on the supply without any connection to a motor.

Is there any way I could solve this issue and avoid making a new board.

Thanks again for everything!

 

[davenn]

I think it is mainly ground looping is the main problem as you pointed out.
I separated the grounds on my board but the output of the optocoupler was very noisy. Although I made sure that there was no connection between the two grounds, the board ground seems to be affecting the input of the optocoupler. The input becomes nosiy after I switch on the supply without any connection to a motor.

that's Very difficult to read, apart from being small and dark, nothing is labelled

Again I ask, Please upload a schematic of your design

cheers
Dave

 

[Windadct]

Grounding is as much art as science and typically it is a matter of best practices. The controller is a digital system, low power & sensitive signals, and the Power side ( with the MOSFET) is a completely different animal - and it creates a lot of noise. This is a good case as well where the schematic can show something as clearly being wrong, or "apparently" nothing is wrong, but the physical design is not good.
The basic rule I like to follow is all of the systems of the same type(Control, digital signals, analog signals, high power) have their own "local" ground, if fed from the same supply they each have ONE point of contact back to the supply ground - eliminating any loops in the grounding scheme. In a proper schematic the independent and single point grounding scheme can be represented, if you know from the beginning that you need a clean grounding scheme. Pretty much any power electronic system should have a distinct plan to manage the grounding.