Analog and Digital Grounds

[Luca]

You can watch those two videos, it's too much complicated and log to explain in words here.

https://www.youtube.com/watch?v=ySuUZEjARPY

https://www.youtube.com/watch?v=ySuUZEjARPY

[Mahsheed]

Current is an "electrical way" to view the story, so the return current path for two tracks (first on L1 and second on layer 3) that are both referred to same GND plane (L2) will be in this plane as well.

Will it generate noise (or effect signal integrity) by any means? (please mention both the cases of DC signals and high speed signals)

Energy is an "EMI/EMC way" to view the same story, so the energy are not in the tracks and not in the GND plane (or Power plane), the energy "travels" in the dielectric material between the trace and plane. Voltage and current are electrical way to "define" your signal, but neither are the energy..

I believe EMI/EMC effects appear because energy/power of one signal track interferes with other signal track if it is in the vicinity of that signal track energy/power, right? then why do we consider this effect at only higher frequencies? since it is the energy that will interfere. (changing magnetic field inducing changing electric field and voltage will be a different view in this energy case). Please clarify my confusion. Thanks in advance.

[Luca]

Current is an "electrical way" to view the story, so the return current path for two tracks (first on L1 and second on layer 3) that are both referred to same GND plane (L2) will be in this plane as well.
Energy is an "EMI/EMC way" to view the same story, so the energy are not in the tracks and not in the GND plane (or Power plane), the energy "travels" in the dielectric material between the trace and plane. Voltage and current are electrical way to "define" your signal, but neither are the energy..
If you don't have a GND plane... simply the current will return in a different way, the smallest impedance path... and if the current can't return... just simply you don't have that current.
Don't forget, in order to have a "current" the circuit MUST BE closed. Completely different story for the energy, think about PCB antenna for example.
I hope this post help you and clarify this two aspects, they seems "easy" but they aren't and very often create confusion.

[Mahsheed]

- A lot of the energy travels between plane and track. Maybe have a look at this my video, it can help https://www.youtube.com/watch?v=ZyoqhJB_E9Q

Yes, I have seen this video before, but I am stuck at a different concept. This video shows how power flows in tracks, but I want to clarify about the return currents. As it is a well known fact that for every current passing in the track, there will be return current in ground plane exactly beneath the track (if possible).

Now if there is no ground plane, where would the return current flow flow? similarly, if there are tracks on 1st and 3rd layers at same coordinates and 2nd plane is ground, then how will the return currents of both the tracks flow in ground plane?

[robertferanec]

If the ground plane is on the 2nd layer and signal tracks are moving exactly at the same positions in 1st and 3rd layer. Then what effect would the return current have on the signal integrity, noise etc

- A lot of the energy travels between plane and track. Maybe have a look at this my video, it can help https://www.youtube.com/watch?v=ZyoqhJB_E9Q

[Mahsheed]

"where is the return current path for each signal that you route".

If the ground plane is on the 2nd layer and signal tracks are moving exactly at the same positions in 1st and 3rd layer. Then what effect would the return current have on the signal integrity, noise etc

Also another key aspect it's the rise and fall time of signals that travel on your PCB, this is valid for both digital and analogue signals.

how can we use this information for pcb designing? and what physical interpretation we can get from the rise and fall time of signals?

thanks in advance

[Luca]

I can also suggest to think and choose "the best you can" for PCB stackup and "where is the return current path for each signal that you route".
Those two aspect are "the main key" for each PCB layout process, and PCB layout process start AFTER the best schematic that you can do and again AFTER a long time for component placement...
Also another key aspect it's the rise and fall time of signals that travel on your PCB, this is valid for both digital and analogue signals.
Without any other infos it's not possible to say much more.

[Mahsheed]

Got it. Thank you so much

[robertferanec]

Depends on speed of the SD card interface, but you can google for something like "mmc pcb layout" to get an idea about requirements.

http://cache.freescale.com/files/32b...ote/AN4215.pdf

[Mahsheed]

- do you mean SD card interface?

Yes, I mean SD card interference.

[robertferanec]

Do you mean that using AD converters to first convert Analog Data into Digital data and then feed the data at Micro controller's digital pins instead of analog pins. right?

- yes

But all I am concerned is if the return currents of GSM and other digital tracks would badly effect my analog data or not?

- yes, noise on power can affect sensitive analogue data if not designed properly. Now, you may be asking - what is the proper way to design it? It depends on your project, there is no one general answer.

simple memory card interference

- do you mean SD card interface?

[Mahsheed]

Soft reminder for answer.

[Mahsheed]

I would maybe consider separate precise AD converter located on different board or isolated from the digital part.

Do you mean that using AD converters to first convert Analog Data into Digital data and then feed the data at Micro controller's digital pins instead of analog pins. right?

For GSM I often use external antenna + GSM power can make a lot of noise on the board, so be sure it has enough filtering.

Yes, I am using second order low pass filter for both GSM as well as microcontroller and using separate power tracks. But all I am concerned is if the return currents of GSM and other digital tracks would badly effect my analog data or not?

Lastly, I would like to know if there is something special to take care in routing the simple memory card interference? (Data sheet does not say anything about this)

Thanks in advance.

[robertferanec]

I would say, that depends on the sensitivity you are looking for.

For standard approximate measurements I would probably use one solid GND plane, however if you are looking for precise measurements, then I would maybe consider separate precise AD converter located on different board or isolated from the digital part. However I am not expert in analogue, I just have seen this in some implementations and I have used something similar in past.

PS: Possibly you may use a accelerometer + gyroscope chip which only has digital interface.

PSS: For GSM I often use external antenna + GSM power can make a lot of noise on the board, so be sure it has enough filtering.

[Mahsheed]

Hi, I am using analog as well as digital data to be processed by microcontroller, but I am using single ground plane for the return path of both analog and digital tracks. Since there is single microcontroller, so I can not seperate ground planes, and also size restrictions does not allow me to separate the analog and digital tracks spaces on PCB. My analog side contains accelerometer, Gyroscope and some resister divieder's data fed at the analog pins while my digital side contains serial communications. Also my board contains some RF tracks of antennas os GSM and GPS.


Now my question is whether single ground plane would effect/ allow interference between analog and digital data?

Thanks in advance.