What's the deal with RC filters? When they stop being effective?

[Rob Strand]

And I keep it. It's even more true in this context of DIY pedals where many of us don't buy from big companies. Not all web sites provides 1uF values, and in the best case they are Z5U type. This type it's no the best MLCC, actually it's among the worst. Of course regarding capacitance drift under voltage. Good 1uF ones are not that easy to find in small package.

I find it hard to the ESL numbers from caps. Plenty of datasheets don't even show you the ESR.

Good and bad can only be judged as being significant from how it is being used.  No parts are ideal.  In most cases some aspects of non-idealness aren't a big issue.  We aren't going to use 1uF air dielectric capacitors or evacuated capacitors for a PSU filter cap.

The voltage dependency of large value SMD caps can be an issue.  The solution is to use larger packages - simple as that.   Non-idealness  like temperature coefficient isn't important for a bypass cap.

The ESR can be read-off the dip on the impedance curve.   Not all brands gives this data.   If you use brands which don't specify ESR then it's your gamble.   In reality even cheap caps with no specs will have a limit on how high the ESR is.  It could be the same as, or twice the value of, brand name caps -  you just don't know but it won't be 1000 times higher.  You might even find the ESR of cheap caps varies more from batch to batch because the production processes and materials are less controlled.

If the cap provides an impedance plot the dip in the impedance plot is at the resonant frequency.   The minimum impedance is the ESR.  You can estimate the inductance from the resonant frequency ESL ~ 1/( C * (2*pi*fr)^2 ).

The ESL is largely a result of the size of the part - basically the laws of physics.
This article gives a formula for ball-park estimates,
https://article.murata.com/en-global/article/impedance-esr-frequency-characteristics-in-capacitors
You can also approximate the body of the cap as a wire,
https://www.eeweb.com/tools/wire-self-inductance-calculator/

So, this is snake oil(?):

It looks too good to be true to be honest... The graphic tells that once frequencies reach the 0 point, attenuation is no longer improving but stays at the peak value. Which is something huge.

It's not snake oil.   It's only the basic theory using a simplified circuit.   The non-ideal parameters like ESR change the details on what happens with the circuit.   Inductors have loses, capacitance.  Capacitance have loses and inductance.   If you want to see accurate results then you need to put those finer details into the circuit model.

In your example, the ESR and the series resistance form a resistive divider.   At resonance Vout/Vin = ESR / (R + ESR) = 37m / (5.6 + 37m) = 1/152 = -43dB, which is what you see on the plot.  The response plot is predictable provided you add non-ideal elements to the circuit model.

In the same vein,  PCB traces have resistance and inductance.  If your have wires going to the cap you will find the inductance of the wires greatly exceeds that of the ESL of the capacitor (a trace the same length as the cap nearly doubles the ESL).   That will change the impedance plot enormously and also shift the resonant frequency will be lower.  The impedance at high frequencies will be somewhat higher than the capacitor's impedance plot.

[R.G.]

Found this:
https://www.electronics-notes.com/articles/basic_concepts/capacitance/esr-dissipation-factor-loss-tangent-q.php
ESR, dissipation factor, and loss tangent are all different ways of expressing the effect of internal series resistance. It's a PITA to convert back to the ESR I actually want to know, but capacitor makers seem to avoid expressing ESR as ESR, and use dissipation factor and loss tangents instead.

[Rob Strand]

You can come unstuck with tabulated data.  In many cases a maximum is specified then on top of that it is specified at 1kHz.

You can see how using the tabulated data in its raw form can produce results way off typical,




(I should point out the improved model isn't good enough to match the ESR plot.  It just creates a rising ESR at low frequencies
You actually need a multi-cap multi-resistor taper network to get a good match.

For completeness,