Filtering switching spikes

[seyyah]

filtering switching

I have problems filtering the spikes that are created by fast switching components. I have two examples. First one is a 1MHz cmos type logic gate switching. As you can imagine the power is low in this scheme. It can easily be filtered by a low pass filter or a single proper capacitor. But the second one is a power supply switching scheme. Although it's frequency is low (~40KHz), i cannot filter it. What is the problem here? The energy of the spikes? What methods can i use for this purpose? Can transzorb or something like that helps?

 

[andre_luis]

If the power suply are not well shielded, it could produce induced noise at signal stage, depending up of Layout design. Check it.

 

[max0412]

Well if the noise is not due to measurement error, you can try a simple LC filter to integrate it down to an acceptable level. Or use higher quality output caps on the regulators output. I’ve been using polymer aluminums on the output of my SMP’s they do an amazing job on filtering out SMPS ripple due to esr. The additional cost is offset by the fact you really don’t need a post LC filter to get an acceptable ripple unless really stringent ripple requirements are necessary.

 

[FvM]

Hello,

from the term switching spikes I assume, you are complaining chiefly about higher frequency components and not the switcher fundamental. This should be analyzed carefully cause different interferences can require different means to overcome them.

Higher frequency interferences ore "spikes" can be most effectly eliminated with combinations of small inductances (can be ferrite beads with sufficient current capacity, ferrite or air chokes) and ceramic capacitors ranging from a few to several ten uF. Low ESR input and output capacitors as the suggested polymer aluminium type additional reduce ripple at fundamental frequency. Or use higher frequency switcher and all ceramic capacitor design if you're ready to face the challenge.

A "solid ground" is also required for effective filtering. But a SMPS placed on a common ground with the rest of the circuit could also inject interferences into the ground plane. Thus it can me meaningfull to have a separated power ground for either all or each individual SMPS tied to common ground at one point.

The important thing is no to have the switching currents, commutating from input capacitor and high-side switch to diode/low-side switch flowing through the common ground. A first LC filter should be placed at the interface of power ground related circuit then.

Regards,
Frank

 

[som_raparty]

hi fi you want to suppress ripple, the simplest way is to use low ESR electrolytic capacitors at the o/p and preferably two in paralles as it reducec the ESR still less
and use a tantalum and ceramic capacitor also as it will help in noise reduction