DPRG mailing list archives: [DPRG] Motor backlash / Back EMF

Mon Jan 7 07:01:48 CST 2008

On Mon, 7 Jan 2008, DeltaGraph at aol.com wrote:

> OK, Chuck,
> I think I am getting clued in.
>
> So really more than IR drops due to loads we are talking inductive drops or
> increases of voltage V=Ldi/dt depending on increasing or decreasing  current.
>
> So for an instant, when relay opens and current in motor circuit decreases
> very rapidly , Scott's ground lead looks like a battery supplying at least  a
> few volts to the ground wire -- got it.
>
> If per chance it supplied 5 volts, the voltage supply might drop to zero --
> assuming no capacitance at processor.
>
> This brings up a question. Could a reasonable capacitor say 10 uF tied to
> processor (or real close)  protect processor from the bounce?

That's exactly what decoupling capacitors are for. Also remember that cmos 
chips only take low currents on average - that low current is taken in 
very very short sniffs of up to a few amps, and they *need* 100nF or so of 
high frequency capacitor as close to their power pins as possible to keep 
their voltage good. Even an inch or two of track introduces enough 
inductance to cause the chip issues. Ceramics are excellent for this, and 
MKTs should be acceptable. Electrolytics simply don't have the high 
frequency response to work here, although they work well in parallel with 
something that does.

A star grounding system can help too - where each separate section's 
ground is connected to a single physical point, so all voltages everywhere 
are relative to that point, and high current sections have no effect on 
the ground leads of low voltage ones.

Remember, wires at right angles to each other have almost no coupling, 
whereas parallel wires have (relatively) good coupling. Keep this in mind 
when running any high current wires. Also, twisting wires with their 
corresponding grounds helps keep noise down as the currents (and hence 
fields) cancel.