Email: Password: Remember Me | Create Account (Free)

Back to Subject List

Old thread has been locked -- no new posts accepted in this thread
???
03/03/05 18:44
Modified:
  03/03/05 18:59

Read: times


 
#89033 - Yes it is
Responding to: ???'s previous message
Ashish said:
Please refer to figure 1 (first approach).
Is that the same concept you are mentioning here ?

Yes. And it will work even as alternative to the methode described in figure 2, if a further analog ground plane is used where the analog cables enter the chassis. This additional local analog ground plane is called 'chassis plane' in my post you refer to.

Some additional comments are necessary:
What the author of article calls 'Micky Mouse' should be avoided. Means the analog and digital circuitry should be very well separated from each other. If this is not possible, then a shield which is directly soldered onto the digital ground plane, surrounding all the digital circuitry might be helpful.
Also, the slots in the ground plane to split the ground plane into an analog and digital section should be as wide as possible. There should be a true separation, otherwise stray capacitance will join the two splitted ground planes together again for high frequency noise. So, a slot which runs over the whole board and is only 1...2mm wide is ridiculous. The slot should be wider than 5mm at least and should not run for a longer distance.
Here again a shield soldered directly onto the local analog ground plane, totally surrounding all the local analog circuitry can help a lot. Of course, costs might rise and rise with this approach...

Of course, all the individual local analog grounds must not be tied together elsewhere. Any further connection between them must only be done by a suited differential amplifier, like this AD830 for instance.

Also, analog power supply must be decoupled very well by using LRC filters (soft ferrit chokes and X7R SMD ceramic high caps preferred).

Especially valid is what Joseph stated in his original post: You must never cross the slot by any analog or digital trace. This will ruin all efforts. If you have to cross the slot, then you must low pass filter the signal doing this, preffered by a pi filter, where there's a capacitance from the signal to ground at each plane and a resistor and/or inductance crossing the slit and joining both capacitors together. Here leaded components are helpful which allow to cross the slot like a bridge does over a river, which will provide the least amount of coupling capacitance between both ground planes.
There are still some other tricks if the slot must be crossed, like using low capacitance optocouplers or symmetrical signal routing, both in combination with sophisticated common mode filtering, but that's a different story...

Kai

List of 29 messages in thread
TopicAuthorDate
WEOnT; Splitting Planes            01/01/70 00:00      
   Re: Splitting Planes            01/01/70 00:00      
      dont connect planes at more than once.            01/01/70 00:00      
         the artwork person            01/01/70 00:00      
            get rid of him            01/01/70 00:00      
               Some of us designers are guilty too            01/01/70 00:00      
                  treat others ....            01/01/70 00:00      
                     How is the rate calculated?            01/01/70 00:00      
                        Rates            01/01/70 00:00      
                           optimisation            01/01/70 00:00      
                              Rational Incentive            01/01/70 00:00      
                              Options            01/01/70 00:00      
                        rate            01/01/70 00:00      
                           But            01/01/70 00:00      
            Artwork Person            01/01/70 00:00      
            True episode just remembered            01/01/70 00:00      
               Hillarious!!!            01/01/70 00:00      
         connecting planes            01/01/70 00:00      
            Concept of local analog grounds            01/01/70 00:00      
               Re: Concept of local analog grounds            01/01/70 00:00      
                  Yes it is            01/01/70 00:00      
            How did you manage it?            01/01/70 00:00      
               re: ground management            01/01/70 00:00      
                  Astronomers ?            01/01/70 00:00      
                     re: Astronomers ?            01/01/70 00:00      
   Ground noise <100µV?            01/01/70 00:00      
      Freely?            01/01/70 00:00      
         resistance vs inductance            01/01/70 00:00      
         Ground return current will spread            01/01/70 00:00      

Back to Subject List