Capacitors and resistors in low / high pass filters

[PRR]

> start "stacking" RC lowpass filters


Your drawing is *ultimately* 18dB/oct.

However the "corner" or "knee" is very broad. The slope won't get near 18dB/oct until around 2 octaves out. In most of the "small shave" area it is sloping 4 to 8dB/oct. It does not sound hugely different from one R-C network.

The soft-knee is especially bad when all R are the same value. It would be better to have them increase along the way, with complementary cap scaling. However if you go 10K 100K 1Meg, then you want 10Meg loading the end of the chain, and less than 1K feeding it. And the cap for 1Meg at 5KHz is 33pFd, a small value which is liable to attract sneakage from other signal paths. Past a few R-C stages you just run out of practical impedances.

It IS useful when you have a way-out interference you need to kill. The 11Hz tremolo rate for example. In simple circuits the trem signal may be 100 times larger than the audio signal. This may thump in later stages. A 3- or 4-pole high-pass around 30Hz can put 11Hz way-way down and still pass 82Hz with little loss.

For good sharp-knee filters (flat to a point then fall like a rock) you want L-C networks or Active Filters.

[Thecomedian]

what happens if you start "stacking" RC lowpass filters after each other? can you get a steeper cutoff (like 24db/octave or steeper depending on the number of the filters) or you will just get a circuit where the values can be simplified down to a single cap-resistor filter? (via series/parallel relationships)

Uploaded with ImageShack.us

you can observe that the behavior is not appreciably affecting the curvature, however it affects the dB slightly, then majorly.

It raises the thought in my bubbly mind that such a network could be fed from first node to third node into a Pot, which would provide you the ability to raise and lower dB of specific frequencies. get the right caps and resistors in there, and this would be the way to produce a treble/bass boost/cut that has store quality dB modulation or some such.

to mimic that aqua NAO system which has +/-18dB for numerous specific frequencies, band pass circuit is required. I have a dusty book which says that old radios band passed using a variable capacitor, which is a cap that has some plates which move out from or into a set of other plates as you move the dial. So multi-caps are definitely "a way" to band pass.

[Kesh]

then majorly.

that's because you are driving a 5 ohm impedance, which is only realistic for a power stage driving a speaker.

[Thecomedian]

Yes. It behaves the same if you remove the last resistor, but then you cant get a reading for the last node because it is "ground".

if that resistor is 500k as if a high ohm volume control the dB level is preserved.

[Kesh]

if you remove the last resistor

you mean if you short the last resistor

and if you do this the last cap does nothing as both ends of it are then connected to ground

so that is another not particularly realistic or useful sim

better you put 10k+ for typical line level input impedance, or much higher for typical transistor or op amp input impedances

[Thecomedian]

That inter and intRA circuit impedence matter is a headache in itself.

the "shorted resistor" is just the resistor removed, making the same circuit as the picture of OP. To read an approximation of the value of that "output" putting a resistor on it of low load. Bigger the load, less attenuation of signal.

I have to modify the vari-cap statement a bit. There's an inductor prior to the vari-cap unit, which is how band-pass occurs.

[chromesphere]

Hey everyone,
just wanted to update this thread to say that i finally understand how the resistor and capacitor values react to each other!  Understanding reactance helped.  Roughly graphing voltage output with changes in frequency also helped me see whats going on.

And as everyone else said, its a voltage divider!  If you increase the resistor value, you extend the corner frequency because the frequency has to be higher on the capacitor (lower resistance) to have any effect on the output!  Your just basically shifting the resistance zone where the capacitor will have a greater effect!  Actually that pretty much works the same for HP or LP.  Anyway, THAT was what i was stuck with!

Wow, i need to lie down after this one
Thanks everyone for your help.
Cheers!
Paul

[WhenBoredomPeaks]

> start "stacking" RC lowpass filters


Your drawing is *ultimately* 18dB/oct.

However the "corner" or "knee" is very broad. The slope won't get near 18dB/oct until around 2 octaves out. In most of the "small shave" area it is sloping 4 to 8dB/oct. It does not sound hugely different from one R-C network.

The soft-knee is especially bad when all R are the same value. It would be better to have them increase along the way, with complementary cap scaling. However if you go 10K 100K 1Meg, then you want 10Meg loading the end of the chain, and less than 1K feeding it. And the cap for 1Meg at 5KHz is 33pFd, a small value which is liable to attract sneakage from other signal paths. Past a few R-C stages you just run out of practical impedances.

It IS useful when you have a way-out interference you need to kill. The 11Hz tremolo rate for example. In simple circuits the trem signal may be 100 times larger than the audio signal. This may thump in later stages. A 3- or 4-pole high-pass around 30Hz can put 11Hz way-way down and still pass 82Hz with little loss.

For good sharp-knee filters (flat to a point then fall like a rock) you want L-C networks or Active Filters.

What is the name of that simulation program? I've seen it a lot of times here, i would like to play around with some values.

Is there any pedal (that we know here) which uses active filters or multiple passive filter stages to shape tone?
(i think i've seen one in the Rebote Delay 2.5, where the highs get filtered out of the delayed signal)

[nocentelli]

A number of Catalinbread overdrive pedals have two LP stages on the output, something like 10k series-> 2n to ground -> 10k series -> 2n to ground -> output (volume control) 

[Ronan]

Crunch Box uses two low-pass filters one after the other. Check the presence control, there is a slight twist.