Removing ground plane under a PCB

[floppy32]

In the PCB board, I have this 12V DC power relay mounted and an MCU on the same board with separated the ground planes with only trace connections between the ground planes. I have also read from some comments that many remove the relay's ground plane in such cases.

As a sequence of events:

0) Neither relay nor the MCU or any electronics is powered on the PCB board
1) The relay is activated by a push button and latches at 12V DC and 300mA
2) The MCU on the PCB board is now powered(from relay) right after the relay latches
3) When contacts are closed current passing the relay contacts gradually increases upto 4A (at that part PCB trace 2.5mm)
4) When turning off, the current passing through the relay contacts gradually decreases to 300mA

From the coil data the coil current will be: 12V/275Ω = 45mA. 300mA is the relay contact current sank by the external load relay is sourcing to through 2.5mm traces. So the reason I mentioned all these is that the MCU is powered after the relay contacts switches(since the relay provides power to it) at 12VDC 300mA. So the relay contact are not switched ON when the MCU is working. So I thought it would not upset things. And I also tried more than 20 times without issue. But in terms of relatability I'm not sure, if MCU(STM32F303) is each time being damaged or not due to EMI which I don't know how to measure an see if there is any or not.

Here is the schematic.

And I still separated the ground planes of the MCU and the relay related section as here. Call this a.

Can this be sufficient in my case, or should I still remove the ground plane of the section where the relay belongs?

And I have two alternatives where

I just removed the ground plane of the relay section here. Call it b.

and

I removed the ground plane of the relay section also tied the PS1 GND and relay ground(GISO) at a closer point, here . Call it c.

I don't want to make placement changes to the PCB but mostly asking which one of these PCB alternatives a, b and c I should better use? I very appreciate your opinions.

 

[KlausST]

Hi,

Please use the [insert image] button to insert your pictures.
I personally don´t find links to unknown sites safe.

Klaus

 

[floppy32]

Hi,

Please use the [insert image] button to insert your pictures.
I personally don´t find links to unknown sites safe.

Klaus

Hi KlausT,

I cannot find a way to edit my post. How is it done?

Otherwise I can assure you it is google drive link completely safe.

 

[KlausST]

Otherwise I can assure you it is google drive link completely safe.

I´ve never met anyone saying: I´m a thief, I´m a burglar, I´m a killer, my dog is dangerous, ... everyone says it´s safe, the good ones as well as the bad ones.

.. and there are forum rules .. mentioning remote file severs... and why not to use them.

***
Regarding editing the post: There only is a limited time to do so. If there is no EDIT button .. it´s impossible for members.

But you could have included the images in post#3 .. now it will be post#5 or even later ...

Klaus

added:
Your headline says "under a PCB" ... from the textual description it should rather be "under a relay". Or I misunderstood the problem.

 

[floppy32]

Yes you are absolutely correct, it supposed to be under a relay.

The thing is I really don't want to do big layout changes with component places so I came up with 3 PCB alternatives; and wondirn which one makes more sense.

Here I insert the images from the links here hope to hear your comment:

Schematic:

PCB A:

PCB B:

PCB C:

 

[BradtheRad]

I cannot find a way to edit my post. How is it done?

After posting, Edaboard's interface allows a limited number of minutes for editing.

Otherwise I can assure you it is google drive link completely safe.

Even though Google.drive is reputable, your post #1 links brought up error messages saying your images could not be displayed. Reason could be anything. It's why Edaboard discourages using external file servers.

Yet your images uploaded directly to Edaboard are visible automatically, no problem (post #5).

 

[KlausST]

My analysis:

* there is no safety isolation (high voltage) between relay coil and relay contacts
* indeed there not even is galvanic isolation ... all is referenced to the same GND

So what´s your concern about the GND plane under the relay at all?

****
GND is provided by all screws from GND plane to the case - so it seems. But the fourth screw seems to be connected also to the case.
With the lengthy GND line of A and B you risk to get a GND loop.

Klaus

 

[floppy32]

High voltage? It is 12V and switches at 300mA.

I use plastic standoffs the holes are not connecting the grounds. I would specify it otherwise. So there is no ground loop.

--- Updated Sunday at 6:34 PM ---

My analysis:

* there is no safety isolation (high voltage) between relay coil and relay contacts
* indeed there not even is galvanic isolation ... all is referenced to the same GND

So what´s your concern about the GND plane under the relay at all?

****
GND is provided by all screws from GND plane to the case - so it seems. But the fourth screw seems to be connected also to the case.
With the lengthy GND line of A and B you risk to get a GND loop.

Klaus

High voltage? It is 12VDC and switches at 300mA.

I use plastic standoffs the holes are not connecting the grounds. I would specify it otherwise. So there is no ground loop.

 

[KlausST]

It is 12VDC and switches at 300mA.

That´s what I have analyzed. You seem not to agree with it.
On the other hand you don´t tell in which regard you disagree ... and did not answer my question about your concern.

****

I use plastic standoffs the holes are not connecting the grounds. I would specify it otherwise. So there is no ground loop.

Then why do you make the screw area copper plated and connect them to GND and GISO respectively?
You say the holes are not connected to grounds, the pictures tell otherwise.

Klaus

 

[floppy.32]

That´s what I have analyzed. You seem not to agree with it.
On the other hand you don´t tell in which regard you disagree ... and did not answer my question about your concern.

****

Then why do you make the screw area copper plated and connect them to GND and GISO respectively?
You say the holes are not connected to grounds, the pictures tell otherwise.

Klaus

Hi Klaus,

First, thank you for spending time. I'm not PCB expert and appreciate your opinions.

I couldnt login my account so use this way instead.

1-) The relay will not switch AC it will switch 12VDC. So both the relay coil and the contacts will work with same voltage 12VDC. So I didn't get why you mentioned, do I still need galvanic isolation(?)

Or maybe you already meant that I shouldn't worry? Just to be sure Im asking if I get it correct.

2-) Relay part hole is connected GISO ground and the other three holes are big plate are signal ground. So even they are connected at a point, there will be no ground loop through copper plates I will use isolated plastic standoffs.

(Print says GND but they are indeed GISO)

The reason I asked this question whether it makes sense to remove ground plane in this case. If you think PCB A which is current verison is fine then I wont remove the ground plane. Im sorry if I sound stupid, its just I dont have enough expertise.

 

[KlausST]

1-) The relay will not switch AC it will switch 12VDC. So both the relay coil and the contacts will work with same voltage 12VDC. So I didn't get why you mentioned, do I still need galvanic isolation(?)

I can´t answer this for you.

Usually a realy is for galvanic isolation .. and quite often a relay switches high voltage high current.
I realized that in your application it is not AC, not high voltage, no high current ... thus I wonder why you choose to use a relay. A BJT / MOSFET maybe is the better choice.

Additionally you "ground wise" treated the "relay section" as if it was a high voltage/ high current / galvanically isolated section.
Thus I asked why you did so.

You may have your reasons or not. We don´t know unless you tell us.
It is your application. And it was you who worried about the GND in the relay section. Talk to us. Tell us your ideas.

We don´t know what´s connected to the relay (contact side). We don´t know who/what controls the relay (coil side)

Or maybe you already meant that I shouldn't worry? Just to be sure Im asking if I get it correct.

To answer it we need your informations first.

2-) Relay part hole is connected GISO ground

What are you talking about here.
My worry about the ground loop? If so: I clearly talked about the screws .. for mounting the PCB to the case.
Not about the 3 contacts your picture shows.

Im sorry if I sound stupid, its just I dont have enough expertise.

There is nothing stupid. Just confusion.
You provided three SOULTIONS for an unknown problem. We need to understand the problem first .. then we can discuss about the solution.

***
Don´t misinterprete my text as "angry" reply. I´m just rather direct and not used to sugarcoat problems.
Nothing personal - just technical problem related.


Klaus

 

[floppy32]

Hi Klaus,

Thanks for the reply. I tried to write more about the use.

(Lets forget about the ground loop, I might not explain well but there will be no ground loop in this. Yesy that pic is not mounting holes but I put them(GISO grounds) to show that they are connected to one mounting hole(right bottom one). Other ground to other three mounting holes. And all screw holes will use plastic standoffs.) The two grounds(GISO and the MCU ground) are only connected at a single point, because I added a keepout zone for the relay section. Why? I will explain below.

Let me try to write about why I asked if I need to remove the relay or not eben though all is 12VDC.

My main concern was: during relay switching would that cause sudden peaks and maybe affect other ICs in schantic such as the MCU. So I thought removing ground plane might mitigate if there is such bouncing. And it is just precaution. Maybe there is no bouncing and I don't know how to measure it as well to see if such thing exists.

What´s connected to the relay (contact side): It is a SW controlled driver and gradually ramps to 4A from 300mA to 4A. So after the relay switches contacts at 12V 300mA the driver starts drawing more current as a ramp. And likewise the SW gradually decreases current to 300mA. So the contacts switches ON only at 12VDC 300mA state. After relay is latched the carry current gradually ramps up by SW.

We don´t know who/what controls the relay (coil side): Coil is controlled by a push button. The relay is wired to be sued as latching action. So a push button energizes and latches the relay via 12VDC.

 

[D.A.(Tony)Stewart]

I supposed Plan B is better because it uses fewer vias, but there is hardly much difference. I consider than the thin traces for power add 10 nH/cm is not a lot but if there was any accidental short, could the power supply exceed a via current rating or a trace fuse open. Probably not, but std practice is to make all Vcc traces as fat as possible to lower switched inductance. If you haven't seen or done a visual inspection Japanese electronics such as printers, you will gain a lot of wisdom on how to design electronics, PCB, interconnects and cables. Many of the boards are single sided.

I would be surprised if the Buck regulator did not have a low ESR cap in the front end, but I could be wrong if they used a surge suppressor ICL component. Dry contacts have that near 0 dt transfer time that makes equations see the resistance limits. for surge currents on closure or flyback voltage on release, but as you explained the uC is inactive. So the only threat is floating interface pins for ESD protection and common mode ground noise. But from what I have seen presented, I see no major issues. The layout could be cleaned up, but I don't see anything obvious other than the skinny Vcc traces surviving a loose screw short.

It's good that you asked though, so spend some time in consumer goods teardowns online or at home and look at how power is delivered and signals are routed. WIth bigger or motorized loads the "PDN Designs" do get a lot more complicated with Bode plots and resonance interactions with all the caps and parasitics.

 

[floppy.32]

I supposed Plan B is better because it uses fewer vias, but there is hardly much difference. I consider than the thin traces for power add 10 nH/cm is not a lot but if there was any accidental short, could the power supply exceed a via current rating or a trace fuse open. Probably not, but std practice is to make all Vcc traces as fat as possible to lower switched inductance. If you haven't seen or done a visual inspection Japanese electronics such as printers, you will gain a lot of wisdom on how to design electronics, PCB, interconnects and cables. Many of the boards are single sided.

I would be surprised if the Buck regulator did not have a low ESR cap in the front end, but I could be wrong if they used a surge suppressor ICL component. Dry contacts have that near 0 dt transfer time that makes equations see the resistance limits. for surge currents on closure or flyback voltage on release, but as you explained the uC is inactive. So the only threat is floating interface pins for ESD protection and common mode ground noise. But from what I have seen presented, I see no major issues. The layout could be cleaned up, but I don't see anything obvious other than the skinny Vcc traces surviving a loose screw short.

It's good that you asked though, so spend some time in consumer goods teardowns online or at home and look at how power is delivered and signals are routed. WIth bigger or motorized loads the "PDN Designs" do get a lot more complicated with Bode plots and resonance interactions with all the caps and parasitics.

Hi Tony,

Thank you for the answer. I can make the traces thicker. When you say Vcc trace(to make fatter) what particular trace/s it is?
I would be glad if you can explicitly mention "the trace between this and that node" on the PCB layouts above.