Connecting Ground Planes From Two Devices

Thread Starter

PIC-User

Joined Sep 25, 2015
78
Hi, I have two separate devices talking to each other by using asynchronous serial communication. Each separate device has their own separate ground plane. Device A has a very noisy ground plane. When I connect the two ground planes in order to implement the serial communication, device A introduces a lot of noise in the ground plane of device B making it malfunction. I need to find an SMD component to filter out the noise from the ground in device A to the ground in device B. I tried a 10uH inductor, but that didn't do anything to help. Using a 10k resistor helps a little bit, but the noise is still there. Can you please give me a component name that I could use here? Thanks.
 

WBahn

Joined Mar 31, 2012
30,303
A lot depends on the spectral content of the noise and also the spectral sensitivity of the device being interfered with.

Depending on your constraints, a simple and effective method is to use optocouplers between the two decivices.

Ferrite beads are another option that can be effective, depending on the specifics.
 

crutschow

Joined Mar 14, 2008
34,855
How are you connecting the two ground planes?

The problem may be that the two grounds have different noise levels which is causing relative noise between the transmitter and receiver.
Try connecting the two together with low impedance braid (example below).
1714179130875.png
 
Last edited:

nsaspook

Joined Aug 27, 2009
13,590
Hi, I have two separate devices talking to each other by using asynchronous serial communication. Each separate device has their own separate ground plane. Device A has a very noisy ground plane. When I connect the two ground planes in order to implement the serial communication, device A introduces a lot of noise in the ground plane of device B making it malfunction. I need to find an SMD component to filter out the noise from the ground in device A to the ground in device B. I tried a 10uH inductor, but that didn't do anything to help. Using a 10k resistor helps a little bit, but the noise is still there. Can you please give me a component name that I could use here? Thanks.
Pictures, scope traces and schematics would be nice.
 

Thread Starter

PIC-User

Joined Sep 25, 2015
78
How are you connecting the two ground planes?

The problem may be that the two grounds have different noise levels which is causing relative noise between the transmitter and receiver.
Try connecting the two together with low impedance braid (example below).
View attachment 320880
Hi, I'm using a 3ft cable connected to one 25mm mono connector in each of the devices. The cable is a standard cable used for this type of connectors. You are absolutely right about the two devices having different noise levels. The easiest way for me to resolve this issue is to add some kind of SMD filtering in the noisy device.
 

Thread Starter

PIC-User

Joined Sep 25, 2015
78
A lot depends on the spectral content of the noise and also the spectral sensitivity of the device being interfered with.

Depending on your constraints, a simple and effective method is to use optocouplers between the two decivices.

Ferrite beads are another option that can be effective, depending on the specifics.
Yes, one of the devices is very noisy. It has an RFID reader at 13.56MHz. The other device is very sensitive to noice. I don't how it is possible to connect two ground planes using an optocoupler. I have used a lot of optocouplers to isolate two circuits, but no two different ground planes. Ferrite beads might be an option that I need to explore.
 

KeithWalker

Joined Jul 10, 2017
3,188
Hi, I'm using a 3ft cable connected to one 25mm mono connector in each of the devices. The cable is a standard cable used for this type of connectors. You are absolutely right about the two devices having different noise levels. The easiest way for me to resolve this issue is to add some kind of SMD filtering in the noisy device.
That is the obvious step to take. Get rid of the noise in the noisy device. Is it coming from the power supply or is it picking up radiated noise?
 

Thread Starter

PIC-User

Joined Sep 25, 2015
78
That is the obvious step to take. Get rid of the noise in the noisy device. Is it coming from the power supply or is it picking up radiated noise?
The noisy device has one of those RFID 13.56MHz modules with onboard etched antennas. I put a lot of caps to keep the power as clean as possible, but I never realized the ground plane was so noisy until I tried to send serial data to another device. By only connecting the two ground planes it drives the second device insane.
 

dendad

Joined Feb 20, 2016
4,503
Are they each running on a separate power supply or is it common?
If you have a common supply, then connecting the grounds together produces a loop. That can cause problems. Try dis-connecting the grounds and see if that helps.
But opto isolated comms is a good way to go.
I have used RS485 isolators from MornSun. They are pretty cheap and work well.
https://www.mornsun-power.com/html/products/20/transceiver-module.html
Or you could use a wireless link. See the ESP8266-01 boards.
 
Last edited:

LadySpark

Joined Feb 7, 2024
194
The noisy device has one of those RFID 13.56MHz modules with onboard etched antennas. I put a lot of caps to keep the power as clean as possible, but I never realized the ground plane was so noisy until I tried to send serial data to another device. By only connecting the two ground planes it drives the second device insane.
serial ground loops happen all the time in physical prototyping. Opto isolation is very common fix for isolating grounds. However, most of the time in these scenario the serial communication method is changed to a RS485 with opto isolation. Since the communication doesn't have a dependency on the ground of either unit.
 

MrSoftware

Joined Oct 29, 2013
2,212
The RFID circuit is spewing radiation and every wire is an antenna. If optocouplers aren't an option, look at the noise-sensitive circuit and figure out what you can do to make it less sensitive. It might be as easy as determining which device(s) are causing the issue and add bypass capacitors to the power pins of those devices. The value of the capacitor(s) will depend on the frequency you need to filter. I went through something similar in the past, we had a 16-bit ADC on board, as well as a number of switching power supplies (2.8V, 3.3V, 5V). The noise was causing all sorts of jitter on the ADC. Stopping the noise just didn't seem possible, but adding bypass caps to the ADC power pins and similar devices cleaned it right up.
 

WBahn

Joined Mar 31, 2012
30,303
Yes, one of the devices is very noisy. It has an RFID reader at 13.56MHz. The other device is very sensitive to noice. I don't how it is possible to connect two ground planes using an optocoupler. I have used a lot of optocouplers to isolate two circuits, but no two different ground planes. Ferrite beads might be an option that I need to explore.
Why do you need the two circuits to have the same ground? You indicated that you are just trying to get the two devices to communicate with each other.

A schematic, or at least a decent block diagram, would be most helpful.
 

crutschow

Joined Mar 14, 2008
34,855
You are absolutely right about the two devices having different noise levels. The easiest way for me to resolve this issue is to add some kind of SMD filtering in the noisy device.
The problem is the different noise levels in the two device planes are seen as signal noise between the two circuits, so trying to isolate the noise between the planes won't help.
My suggestion was to connect the two grounds with a low impedance connection, so the two ground planes would be as one, reducing the relative noise between the two.
Wide electrical braid has a low impedance for this purpose (having both low resistance and inductance), the shorter and wider the better of course.
Soldering the braid to the ground planes would be the best for that purpose.

The alternate, as suggested, is to use opto couplers for the signals, so the relative ground noise is not seen by the signal.
 

Thread Starter

PIC-User

Joined Sep 25, 2015
78
serial ground loops happen all the time in physical prototyping. Opto isolation is very common fix for isolating grounds. However, most of the time in these scenario the serial communication method is changed to a RS485 with opto isolation. Since the communication doesn't have a dependency on the ground of either unit.
To use opto isolation I will have to do it in the second device with a cleaner ground. I can only modify the noisy device. Still trying to figure it out.
 

Thread Starter

PIC-User

Joined Sep 25, 2015
78
Why do you need the two circuits to have the same ground? You indicated that you are just trying to get the two devices to communicate with each other.

A schematic, or at least a decent block diagram, would be most helpful.
I need to send serial asynchronous data. I disconnected the ground connection to see if it works, but it didn't work.
 

crutschow

Joined Mar 14, 2008
34,855
To use opto isolation I will have to do it in the second device with a cleaner ground. I can only modify the noisy device.
It will work just as well at the noisy device, since the ground noise (which appears equally on both the ground and the signal) will not be seen by the opto coupler.

Is this signal connection bidirectional?
 
Last edited:

Thread Starter

PIC-User

Joined Sep 25, 2015
78
It will work just as well at the noisy device, since the ground noise (which appears equally on both the ground and the signal) will not be seen by the opto coupler.
I don't see how to put the optocoupler in device 1. I can't modify device 2. The serial data is bi-directional and that complicates things. The cable is 3ft long with 25mm mono connectors on both ends.

I tried putting some ferrite beads that I had in my lab, but that didn't help. The ones I found and all the ones I saw on Mouser have a resonance frequency at 100MHz. The best result that I have found so far is putting a 10k resistor in series between the two grounds.

1714193454217.png
 

crutschow

Joined Mar 14, 2008
34,855
Exactly what serial interface are you using?
There are bi-directional isolated couplers that may work for you, such as this for the I²C protocol..
 

WBahn

Joined Mar 31, 2012
30,303
I need to send serial asynchronous data. I disconnected the ground connection to see if it works, but it didn't work.
Of course it didn't work -- you were trying to send an electrical signal on a single wire with no reference.

1714194966950.png
Here's a crude sketch of how to do it.

The one on the left is the one you can change, the one of the right is the one you can't. You bring over your two signal lines (TX and RX, assuming your communications are full duplex) and the ground from the right side.

You connect these to the appropriate side of the optocouplers, using an isolated ground plane if needed.

If your comms are half duplex, then that makes things a bit trickier, but there are optocouplers specifically designed for this.

What protocol are you using and what is the baud rate?
 
Top