What do you expect the noise frequency on the MPU power line? Is it a MHz, kHz, or just some sparks?

Dear Stanley,

this also depends on what else is powered by the 5V regulator and, of course, on the used power supply decoupling measures!

Murata has published very nice plots, where you can see, what actual noise is generated on power supply line, when HCMOS chips are made to toggle their gates. Have a look at here:

http://www.murata.com/emc/knowh...5ey/4e.pdf

When having tested a modern microcontroller situation would look rather similar, maybe even a higher frequency range of interference must be expected.
A big portion of this noise is eliminated by power supply decoupling capacitors shared all over the board. But only, if concept of solid ground plane is used. Even more suppression of this noise can be achieved if concept of solid ground plane + solid power plane is used, demanding the use of a four layer board, of course.
If you now keep in mind, that a standard 5V regulator has a big stray capacitance between input and output (remember that there's this big series regulating NPN transistor on die), then there's not much suppression of noise by the regulator at frequencies above about 1MHz! Means, a standard 5V regulator does not create any barrier for high frequency noise on power supply line, neither from input to output, nor vice versa. That's the reason, why some sort of high frequency filtering is often used directly at input of 5V regulator, you know that LRC filter stuff, which I recommended to you.

Also, a voltage regulator only provides stabilisation of output voltage. But it does not introduce a barrier for sharp load currents! Means, if some circuitry at output will deliver a heavy load current, then the voltage regulator will also demand for this heavy load current from battery. Only if additional filtering at input of regulator is provided, then the regulator can fetch this heavy current from added storage capacitor, decreasing the need for delivering spike currents by battery. That's the benefit of filtering: Not only to keep voltage constant, but also to deliver spike current and by this to keep current loops small.
Of course, you can also put the storage capacitor at output side of regulator. But in order to keep constant voltage constantly you cannot use here an LRC filter with relevant R and L, because of voltage drops across these parts. At input side, on the other hand, L and R can be introduced to some amount, without affecting the performance of regulator.

With a proper LRC filter in front of regulator and with proper decoupling capacitors behind the voltage regulator you can decrease conducted interference to an extreme minimum. So, if you use the filter I proposed, I don't think that you will see relevant spikes at input of LRC filter. If you see some, than they could be caused by the motor and by unavoidable source impedance of battery, which increases with age, by the way...

If your 8052 board is NOT filtered adequately, you will see very narrow voltage spikes at output of 5V regulator, with a periodicy of more than about 1MHz. Also at input this high frequency noise can be observed, but a bit decreased. With proper LRC filter at input of regulator you will not see any longer relevant spikes at the battery side of this filter. But to achive this, the use of a soft ferrite is a must! A simple 10µH inductance, you know that stuff with these many windings sitting next to each other, will not give adequate performance, because of interwinding capacitance, which heaviliy degrades very high frequency performance.

Normally this LRC filter at input would not need a big storage capacitor, if the only purpose is to suppress high frequency noise. But if you want it to also filter out heavy dips and spikes of battery voltage caused by fast motor current spikes, then some hundreds of µF will be needed. But don't oversize this capacitor: If choosen too big, filter performance will suffer from to high equivalent series inductance. So, I guess 100...220µF is ok, especially if paralleled by 100nF X7R, connected to electrolytic capacitor by as short connections as possible!!

Hope this answered your question,
Kai