EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: banedon on September 08, 2013, 01:14:55 pm
-
Hi guys
Here's a very basic question that you can all laugh at me for asking about :).
I know a voltage rail is normally (in a DC circuit anyway) + voltage on one rail and 0 voltage/ground on another (I may be a beginner, but I do know *this* at least! :D).
Mooching around the net I've noticed a few forum posts mention ground bounce and the importance of this being reduced as much as possible - especially on microcontroller/processor boards.
My understanding of ground bounce (after reading the wiki entry :D) is when ICs do any kind of switching they cause noise (electrcial pulses/ripples?) on the ground plane as some of these signals are switched to earth. Such noise can the cause issues with other ICs. This is reduced by putting is decoupling/smoothng capacitors (100nF for example) every few ICs between VCC/VDD and ground.
Have I got it right so far?
If so, can you give advice on how else this effect can be mitigated? Thicker ground wires? Some people mention using a star formation if using wire wrap...
-
Ah, another one of these often misunderstood concepts.
What is ground bounce ? in short : the turning-off of an output transistor due to a resistive path in the emitter or source for common emitter or common source drive stages.
Take the following:
A simple npn transistor. Base gets a control signal via a suitable resistor. Emitter goes to ground , but via a small and long trace , possibly some vias in there ..
Collector goes to a load pulling peak currents. This can be a reactive load , or it can be a fast digital circuit.. Doesnt matter. Peak currents is what we need.
The emitter connection is 'bad' so to speak. , bad as in not low impedant . I specifically say impedant and not ohms as ground bounce is created during the switching itself. During the closure of the switching we are dealing with a rising current . Any inductance in that emitter to ground pathway is going te behave as impedance for this varying current.
So, what happens ?
We turn on the transistor, the current swells and this creates a voltage across the emiiter-to-ground connection. Current* impedance(ohms) is voltage.
If you are driving the transistor from a 1 volt source , and the voltage across emitter to ground reachesmore than 0.4 volt ... The transistor will turn off because you no longer have the required ase emitter voltage (1 volt -0.4 between emitter-ground) to keep the transistor on. The transistor begins to turn off, thus reducing the current. This in effect drops the emitter ground connection and the transistor turns back on ! The current rises again , transistor goes back of .
Transistors dont switch at infinite speeds. So it is the 'swelling' and 'subsiding' of the current that induces a voltage in the emitter pathway. If the voltage is large enough it will interact and you get 'ground bounce'
Sources of ground bounce:
Bondwires.. Driving too much current through a part that isnt designed for it can cause this.
Bad connections (too small traces for the current, too many vias , wrongly designed thermal reliefs)
Driving the transistor in saturation too fast for the load you are trying to engage. Slope control does wonders to avoid ground bounce.
So, in your case with wirewrapping: you do need a star connection to the supply. If you have 10 transistors and you go from emitter 1 to 2, then from 2 to 3, then from 3 to 4 and so on. And all transistor switch on at the same time : the wire from emitter 10 to ground gets the sum of all these currents .. If that is a long wire you are going to see so e weird shit happen there....
If you give each transistor its own wire from emitter to ground , problem solved
-
Hi free_electron
Thanks for your explanation.
I'm not sure I completely understand what you've put there, but here's how iit comes across to me (in simplified terms):
The over-loading of the ground wire/connections causes impediance/resistance which then causes the attached device(s) to act erratically because of the stauration. By giving a better line to ground - such as thicker wires, better connections and/or a connection for each device then this saturation is avoided.
-
The Wikipedia entry on ground bounce lists a few articles and application notes. You will find even more with Google.
The appnote from Fairchild is pretty thorough: http://www.fairchildsemi.com/an/AN/AN-640.pdf (http://www.fairchildsemi.com/an/AN/AN-640.pdf)