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

Back to Subject List

Old thread has been locked -- no new posts accepted in this thread
???
02/23/07 15:43
Modified:
  02/23/07 15:56

Read: times


 
Msg Score: +1
 +1 Good Answer/Helpful
#133554 - Some hints...
Responding to: ???'s previous message
Hendryawan said:
Sorry I'm swear, its true. I already tell that I have been design four PCBs properly. After install into panel one of them was problem. Why only one? If all PCBs hung maybe my design is not correct.

I believe you. Of course, chips are not always 100.00000%-ly identical. Nevertheless, your conlusion, that all prototypes must fail to prove that your design is wrong, is dangerously wrong.

Why?
Because you must always provide some headroom when designing a project. I tell you, the fact that even only one prototype failed is a prove that your design is wrong. You just must take into consideration, that chips are slightly different. Your project must work even with these eventually more sensitive chips.

I have had a brief look at your homepage (link given in your user profile) and found a schematic of a different project. Let's have some disussion:




This plot shows a single layer board (!) containing some bridges. This layout looks very nice and you need only 4 bridges. But the mistake is, that this layout is only suited to handle low frequency analog signals, but not the fast digital signals of a microcontroller circuit! Your GND traces are much too narrow and long, which will result in extreme inductance. Also, your GND routing forms a closed loop arround the entire board, working like a transformer winding, sensing each magnetic field and adding noise on GND. This loop alone can be responsible for your noise sensitivity!! Open the loop somewhere!

But much better is the use of a solid ground plane, which is nearly a must when working with todays micros. Have a look at this little demonstration layout (take note, vias have still to be added, which are omitted here for clarity):


Only the signals (top layer):



The signals and ground fill (top layer):



The solid ground plane (bottom layer):



Some enlarged detail of top layer:


It's only a double sided board, but noise immunity is incredibly better than with a single layer board!!! Take note, how much better the Vcc decoupling of micro is!!!!! The solid ground plane offers the least inductance ever possible!!

Some more hints: Add decoupling caps directly to each chip. Add decoupling caps directly to each connector. Add decoupling caps directly to the regulator. Add an electrolytic of at least 47ยต/25V directly at the input of regulator. (No wonder, that the relay makes trouble.) Add some filter resistance or perhaps some soft ferrite choke where the supply voltage enters the board. This makes your circuit not only immune against HF noise coming from outer world, but also stops the noise leaving your board!


Good luck,
Kai

List of 51 messages in thread
TopicAuthorDate
How its happen?            01/01/70 00:00      
   correction            01/01/70 00:00      
   my 2 cents            01/01/70 00:00      
   What is exactly RLY fuction?            01/01/70 00:00      
      Effect of relay            01/01/70 00:00      
         Isn't it called EMI?            01/01/70 00:00      
            Sure - radiated and conducted            01/01/70 00:00      
               Ok, thought so :)            01/01/70 00:00      
                  Maybe it's a test?            01/01/70 00:00      
                     Not very kind to operator :)            01/01/70 00:00      
         A diode helps.            01/01/70 00:00      
            But immunity is best!            01/01/70 00:00      
         Arcing machine            01/01/70 00:00      
            RF oscillations            01/01/70 00:00      
               School experiment            01/01/70 00:00      
   Ok, we posted in the same minute :)            01/01/70 00:00      
      This time it was Andy and AGAIN same time :)            01/01/70 00:00      
      I'll build in some random delay next time :-)            01/01/70 00:00      
   the solution will be here            01/01/70 00:00      
   I'm puzzled            01/01/70 00:00      
      It damps out the relay voltage spikes?            01/01/70 00:00      
         not much, if any - 33pF?            01/01/70 00:00      
            Should be a printing error.            01/01/70 00:00      
   let's focus            01/01/70 00:00      
      Of course it is possible.            01/01/70 00:00      
      why focus            01/01/70 00:00      
         because......            01/01/70 00:00      
            Part more susceptable to EMI?            01/01/70 00:00      
            'experiment' all you want            01/01/70 00:00      
            prevent problems            01/01/70 00:00      
      where did we not focus?            01/01/70 00:00      
         conclusion            01/01/70 00:00      
            You are a funny guy...            01/01/70 00:00      
               chips is different            01/01/70 00:00      
                  already answered            01/01/70 00:00      
                  That's the definition of "Unreliable"!!            01/01/70 00:00      
                  pure unadulterated male cow manure            01/01/70 00:00      
                     The Relay            01/01/70 00:00      
                        The purpose of a Watchdog            01/01/70 00:00      
                           Woof            01/01/70 00:00      
                              Grrr            01/01/70 00:00      
                                 Watchdog Hit Report            01/01/70 00:00      
                  Some hints...            01/01/70 00:00      
                     The decoupling Cap            01/01/70 00:00      
                        With shortest leads to Vcc and GND pins            01/01/70 00:00      
                     What software,Kai?            01/01/70 00:00      
                        Target            01/01/70 00:00      
                           yes            01/01/70 00:00      
                     Thankyou Kai            01/01/70 00:00      
   isn't it obvious?            01/01/70 00:00      
      No you missed that it is the coil...            01/01/70 00:00      

Back to Subject List