Sharing signals on partitioned PCB - Multi-power or not?

Go To Last Post
8 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi All,

I have read many excellent posts here and a whole heap of other docs but I am still going around in circles.

My question is regarding sharing digital signals between ICs that are supplied by different voltage regulators.

In the first case, I have a PCB that I have partitioned into analog and digital sections.

(Please see attached pdf with block diagram!)

The ADCs are controlled on SPI by an xmega with lots of nice UART ports.

I could connect all devices to the same voltage regulator ("Version 1", top), where I am assuming a star connection is the best.

Alternately, I could regulate each supply voltage separately, ("Version 2", bottom). Here,the different ground planes are separated to prevent disturbances from one voltage regulator getting into another ground plane.

Can anyone comment on the relative merits of the two potential designs?

Is there an acceptable method of routing digital signals between devices with different voltage regulators (preferably without optocouplers)?

In version 2, I have routed the SPI lines, on the back side of the board, through the 'star' connection of the ground planes. Is this sufficient to prevent disturbances, from one ground plane, entering the other ground plane via the SPI lines?

In version 2, is there likely to be a significant difference in the '0V' reference of the SPI lines?

Would be grateful for some words of wisdom!

Attachment(s): 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

As per 1 with just a single groundplane.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Except for the galvanically isolated interface of course. Don't forget a cap between the two grounds.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks guys,

I am certainly tending towards 1 for this design, particularly as this board doesn't have much digital stuff going on.

I would still like to know if it is possible to route digital signals between different ground planes, as in 2.

(Specifically, I want to add a 5V LCD module to this otherwise 3.3v board and need to connect UART to it from the processor- this problem is analogous to 'version 2').

Cheers

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Noddy,

Consider four layer design with ground and Vcc on inner lanes and signals on outside layers.

Costlier than a doublesided board but may provide with the isolation between sensitive areas.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Jayjay - good call regarding the cap between the two grounds. Only recently have I resolved a problem that had been plaguing a product for years due to the interwinding capacitance of a transformer in a dc-dc converter. Adding the capacitor went a long way to resolving the problem. High frequency currents can find the most unexpected path.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

noddy2 wrote:
I would still like to know if it is possible to route digital signals between different ground planes, as in 2.
Yes but strongly not recommended.
Crossing the River by Doug Smith.
Reason for Ground Splits by Dr. Howard Johnson.
Ground Fill by Dr. Howard Johnson.
Usually one ground plane for all works unless REALLY need low noise (data sheet(s) will recommend if needed).
Noise/ENOB/lsb are inter-related.
Sometimes a short "neck" in a single ground plane between low impedance circuit (digital) and high impedance circuit (analog).

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If you use a bit of clever software it probably all doesn't matter that much and SPI currents won't interfere with the analogue bits; I don't know what kinds of ADCs you're gonna use but I would request the ADC to take a sample, wait a while, then read out the result;e.g. no SPI comms during the conversion; you can extend this across all the ADCs. Unless you need all the speed you can get from the ADCs and you must do SPI comms to read the previous result while the ADC is working on the next one.