Author Topic: Connecting multiple boards in a system  (Read 3452 times)

0 Members and 1 Guest are viewing this topic.

Offline MarkoAnteTopic starter

  • Regular Contributor
  • *
  • Posts: 73
Connecting multiple boards in a system
« on: December 07, 2015, 06:48:08 pm »
Hi,
I'm looking for a way to connect different modules of my induction system together. There will be around 6 boards, for expansion I'm looking to make it possibly to connect 10 boards. The connection will be between: LCD/Main CPU board, PFC controller board, DC/DC controller board, DC/AC controller board, general IO board, FPGA board.
The data: Data transfer raits will not be high as each module would work on its own and only needs basic config and monitoring data. In the last prototype we had, we used bit banging 16 bit parallel solution and we had problems with noise pick up form the various switching components. To solved that I was thinking about using differential drivers and receivers and moving form parallel to SPI communication. The boards would not be far apart (probably less then 1M all together- I don't know the exact length as they will be placed next to the power hardware each board controls). They would be daisy chained by 1 many pined flat cable for data. What do you think of this solution?
The power: I'm torn between using 24 V bus and then a buck converter on each board to make the 3.3V and using 1 buck converter and then supplying the 3.3V to each board true the flat cable or using 1 buck converter to make 5V and then using linear regulators to get 3.3V - for ripple and switching noise reduction(this solution would also let me, on mix signal boards, to use 2 linear regulators 1 for analog and 1 for digital and that would improve on analog performance). What would you recommend?
 
Best regards,
Marko
 

Online Ian.M

  • Super Contributor
  • ***
  • Posts: 12853
Re: Connecting multiple boards in a system
« Reply #1 on: December 07, 2015, 07:05:15 pm »
I've answered in some detail over at http://www.microchip.com/forums/m901510.aspx, but constructive criticism, second opinions and alternatives are always a good thing.
 

Offline MarkoAnteTopic starter

  • Regular Contributor
  • *
  • Posts: 73
Re: Connecting multiple boards in a system
« Reply #2 on: December 07, 2015, 07:25:39 pm »
Quote
constructive criticism, second opinions and alternatives are always a good thing
Exactly.  ;)
The thing is that in university they taught me just enough that I know that I know nothing. In the style - interference and noise are bad. The end.  So now I ask everywhere I can to get more info form people with real world advice and experience.
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 26892
  • Country: nl
    • NCT Developments
Re: Connecting multiple boards in a system
« Reply #3 on: December 07, 2015, 10:16:22 pm »
First rule: never ever use a central supply for low voltages! Put a 3.3V (or whatever voltage) on each board. An UART over RS485 based protocol (preferably with on master device) is a good choice to have reliable (half duplex) communication.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline MagicSmoker

  • Super Contributor
  • ***
  • Posts: 1408
  • Country: us
Re: Connecting multiple boards in a system
« Reply #4 on: December 08, 2015, 11:10:44 am »
You can run into stability problems from connecting multiple switchmode regulators in parallel; I like to instead use a single buck regulator to create a 7V rail, and then use linear regulators locally as needed.

As for board to board interconnects, if you have the space and aren't transmitting signals above ~20MHz or so, it is hard to beat plain old ribbon cable. You can get surprisingly good noise tolerance by assigning every other conductor on the ribbon cable to ground (ie - an entire row of pins on the connectors). If you need better common mode noise immunity then you can also slip on a rectangular ferrite core especially made for ribbon cables. For more space-constrained applications you can go with "Flat Flex Cable" (FFC).

I would only resort to differential signaling if the distance was more than 1m and/or the data rate was over 25MHz, but it is certainly a valid option.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf