Also, if you're using a cheap China module for 5V, you may find it's noisy for the same reasons.
You can PWM filter to the motors, so the switching noise is confined to the transistors. You may also find it's more effective to filter the noise at its source. So, in this way you can probe common mode (often, ground-ground, but actually the generalization of that idea) noise and see what's, well, common with the scope (least noise) perhaps. Well, speaking of wires, what's that ground clip hanging off of? Oh. Or filters low frequency ripple, but you're left with persistent sharp squigglies that show up almost everywhere, even when probing with the scope probe shorted to its ground clip (and in turn touching circuit "ground"). So, you may find simply adding an LC filter anywhere, has little effect. Or in other words: when is ground not ground? At AC, when it's made of wires, for sure. So, depending on how you've got everything wired up, given this knowledge, you may suddenly realize that, what you are measuring, isn't actually going to be addressed by the filter at all, but is between - what you thought was - ground. Wires still have stray inductance, in fact about 10nH/cm in typical sizes and applications. By the way, what is an inductor? Just a piece of wire, right? Well, it's wrapped around a bit, but, how much does that really matter? None at all, as it happens. For equal capacitors C with ESR, choose C >= 1 / (2 pi Zo Fc), ESR = Zo (within say a factor of 2 or so), and L = (ESR)^2 / C. You can easily afford to just throw enough either side of an inductor to get a well damped network. They have appreciable ESR, and are cheap capacitance. For PS filters, the most common ingredients are electrolytic capacitors. Use a shunting R+C, or a series R||L, on one or both sides. If source and load are weird impedances (very high or low), then a damping network should be provided. You need to provide adequate termination, to get the expected frequency response. How near? And what is the impedance at and around Fc, is it exactly Zo, or something else? When properly terminated (source and load are resistors R = Zo), the resistance is, well, about that. So we really only have to worry about frequencies near Fc. At high frequencies, the output side capacitor has low impedance shunting the output. At low frequencies, the series inductor has low impedance, so if the source impedance is low, so is the output impedance. For this filter, the impedance peaks around Fc (the cutoff frequency - a special case of Fo), and drops either side. For example, the ratio of output voltage (supply ripple) to load current. So what is the impedance? Well, it's the ratio of voltage to current, at a given frequency. or does that have nothing to do with it? Thanks, BuzzĬlose! When working with LC networks, you'll find these pop up a lot: Fo = 1 / (2 pi sqrt(L C)) Zo = sqrt(L/C) In fact, this is all the calculator is really doing, starting from tabulated values that give the desired response (Butterworth, etc.) then scaling them proportionally for the desired cutoff and impedance. 5V/3A = 1.6 ohms? Microcontroller might consume 150ma and 5V. What kind of impedance would the output of the buck converter want to see? What kind of input impedance does the microcontroller module present to the converter? Is it as simple as. These filter calculators want to know the input and output impedance. I thought that instead of guessing values I'd actually use an online calculator to figure out the best components for a Pi filter at a given cutoff frequency. Through trial and error, I've found I can clean it up well will a simple LC filter between the output of the buck converter and the microcontroller. Ok, so when I am driving the motors I see PWM noise on the 5 volts coming from the buck converter module. (a) powering a few motors through PWM driven mosfets, and (b) into a buck converter module that outputs 5 volts for my microcontroller which is sending the pwm to a couple of BJTs which are in turn controlling the mosfets.
I've got a project where I have a 12V power source.