-6db/Octave...

[thehallofshields]

Okay. So I'm curious if it is possible to make passive Low-Pass-Filters with a steeper roll-off than -6db.

Would a 2-Pole (do they call these 2nd Order LPF?) give me like... -12db/octave?

I was really impressed with the Green Ringer Filter Mod, but I don't fully understand the 3-Pole design. Why does searching this on google only turn op-amp designs.

Any information on, and examples of a multiple-pole passive filters, especially ones that can implement Potentiometers, would really help.

[Seljer]

Yep. You can do that. You can keep slapping passive stages  and you get a steeper roll-off.
However, the math to calculate the corner frequency starts becoming complicated as all the stages start interacting with each other due to impedance issues. So if you want a 'proper' response it's easier to accomplish with an opamp. If you just want to cut off treble in a distortion circuit than go ahead and tweak by your ears

If you want to be able to adjust it you're basically limited to multi-gang potentiometers (or rotary switches to select between multiple capacitors/resistors).

The other way to get a steeper roll-off is to make an LC filter, but inductors for audio frequencies tend to be rather cumbersome so these filters tend to be limited to loudspeaker crossover circuits.

[Mark Hammer]

One of the quirks of more and less steep filters, especially of the lowpass variety, is that selecting a corner frequency depends, somewhat on the steepness.

If I have a 1-pole LPF with a corner frequency of 2khz, content at 4khz is only down by 6db, and content at 8khz is only down by 12db.  So, actually, a fair amount of treble is still passing, even though 2khz seems pretty low.  If I have a 2-pole filter, content is down by 12db at 4khz and 24db at 8khz.  Not exactly a "brick wall", but noticeably less audible treble.

The implication is that, sometimes, if using a 1-pole LPF, you need to set the corner frequency lower than you might think.  Conversely, implementing a multi-pole filter (2, 3, 4, etc.) gives you the freedom, as it were, to set the corner frequency much higher, by virtue of the selectivity of the filter.

[seedlings]

Here's a simulation of two 200Hz high pass filters and two 8000Hz low pass filters.  Each second filter is beneficial.  




Here's two more 200Hz filters:



CHAD

**edit... sorry, those are hard to read.  In the second picture there is a BLUE line between the Green and Red lines.

[midwayfair]

There's these, too, 8th order filter chips:
http://www.mouser.com/ProductDetail/Maxim-Integrated/MAX7401CPA+/?qs=sGAEpiMZZMu%252bahkv8RvG%2fSlVaJqajsj5pHzpq%252bFzIfI%3d

The steepest curve you can get in the least number of components.

[thehallofshields]

If I have a 2-pole filter, content is down by 12db at 4khz and 24db at 8khz.  Not exactly a "brick wall", but noticeably less audible treble.

Hah. Perfect. I actually felt stupid thinking that it might be as simple as -12db. I love it when things are nice and simple like that, even if its not exact.

[thehallofshields]

If you want to be able to adjust it you're basically limited to multi-gang potentiometers (or rotary switches to select between multiple capacitors/resistors).

Hmm... How much do you think it would take to bribe Mark Hammer into inventing the 2-Pole SWTC?

I keep looking at that very sweet Green Ringer 'Filter Mod' from 2005, and I don't understand why you couldn't add a trimpot between the caps-> ground. (SW2a)
http://www.diystompboxes.com/smfforum/index.php?topic=32326.0

[thehallofshields]

There's these, too, 8th order filter chips:
http://www.mouser.com/ProductDetail/Maxim-Integrated/MAX7401CPA+/?qs=sGAEpiMZZMu%252bahkv8RvG%2fSlVaJqajsj5pHzpq%252bFzIfI%3d

The steepest curve you can get in the least number of components.

Wow, I never would have guessed something like that existed. Looking at the graphs it looks like it can get as steep as -20db/octave?

Hey its got inductors in there too. I looked at multiple pole R-C-L filters and thought I'd only find those in large amps. Very Cool.

I don't really get why it oscillates, or how you would control the 'clock' pin. Can this thing be adjusted with a pot, or is it going to take capacitor swapping to play around with the corner frequency?

[ashcat_lt]

There aren't any inductors in that chip.  That picture in the datasheet just illustrates the circuit that the chip emulates via switched capacitor technology.  Don't ask me what that means, but...

You can vary the frequency either by sending it a clock signal or by adjusting its own internal clock signal via external capacitor.

[Mark Hammer]

Hmm... How much do you think it would take to bribe Mark Hammer into inventing the 2-Pole SWTC?

It wouldn't take anything to bribe me, but it would take a bit to make it actually work right.

Consider, if I have a small value fixed resistor, in series with a pot, moving the wiper from the "far" end towards the fixed resistor reduces the resistance used in tandem with the cap to ground, to gradually raise the 6db-down point, without changing the total resistance of initial fixed resistor, tone pot and volume pot.  That's what makes it so stupidly wonderful: it doesn't change the output level when you change the tone.

Okay, so far so good.  Now, let's imagine we have two of those things in series, such that we have:

Ra-->pot1-->Rb-->pot2-->volume-control

We have a cap from the wiper of each pot to ground.  As the wiper of pot1 moves towards Ra, the filter corner frequency moves upward.    But as we move the wiper of pot2 toward Rb, it still has the full resistance of pot1 in series with Rb.  YES, the resistance contributed by pot 2 is reduced, but we can't get rid of pot1, such that the second section has a corner frequency dictated by the value of its cap and pot1+Rb.

I suppose one could use a different cap value for the 2nd section, but consider that when rotated fully the "other" way, filter section one rolls off at the frequency given by Ra, pot1, and C1, and section 2 rolls off at the frequency given by Rb, pot2, and C2.  Start turning the pot shaft and the first resistance leading up to the wiper gets smaller, but every ohm it "gives up" gets added ONTO the resistance leading to C2, such that the second section never really has a chance to move upwards.

Believe me, I like 2-pole filters enough that I would have whipped it up, if I could have.

Two-pole filters WILL work and CAN be implemented, just not in a passive form that will eliminate interaction with the volume pot, like the 1-pole SWTC.  That said, it is not a hardship to have a tone control that requires a volume adjustment to complement tone changes, merely an inconvenience.  One can easily make a passive 2-pole lowpass with a dual-ganged pot that exhibits a bit of volume loss when rolled off lower, but not too much.

For example, let us say we have Ra(1k) --> 10k-pot1 --> Rb(1k) --> 10k-pot2 --> 100k-volume pot, with a 22nf cap to ground from the junction of pot1 and Rb, and another from the junction of pot2 and the volume pot input.  In this application the pot1 wiper goes directly to Rb, and the pot2 wiper goes directly to vol.  That is we are using only two lugs on each pot half, as a variable resistance.  When each of pot1 and pot2 are at min resistance, the corner frequency will be 2-poles at 7.2khz  (lotsa bite there).  As the pot1 and pot2 resistances are increased, the corner frequency gradually moves downward to 660hz.  Thats a pretty decent range of control.

But note, however, that when pots 1 and 2 are at max resistance, the whole thing, in combination with the volume pot, behaves like a 122k pot that can never be turned up more than 82% of the way.  As the corner frequency is raised (by paking the pot 1 and 2 resistances smaller), it ends up behaving like a 102k pot that can never be turned up more than 98% of the way.  Not a huge difference, but enough to require turning the volume up whenever you turn the tone down.

So, no 2-pole SWTC, I'm afraid, but not a big reason to NOT use a 2-pole lowpass.

[thehallofshields]

Believe me, I like 2-pole filters enough that I would have whipped it up, if I could have.

Thanks for the write-up Mark. I've got this diagrammed on paper and I'm walking through your post. I wish you had an instructional series. The SWTC is actually sufficient for the distortion design I'm working on, but I had to double to pot size (10k, b50k, 0.022.uf) to kill some unwanted harmonics. Of course a steeper filter would really help that design.

Much appreciated. Though I'm still curious what would happen if we tried stuffing a pot inbetween the cap->ground of the typical 2-pole/3-pole design.

[Seljer]

You'd have a filter with 2 poles, but the potentiometer would most likely make one pole 'move around' much more than the other.

And if you wire the pot in series with the capacitor you're making a shelving filter, so the slope eventually flattens out to the ratio of the resistors instead of going on down to -infinity as the frequency goes up.

[thehallofshields]

All right. That about answers everything. It looks pretty impractical to try to adjust the corner frequency of a multiple-pole filter.

Except... Has anyone tried blending between a multiple-pole filter and a bypass?

Maybe something like like a simple mpf102 bypass in parallel with that MAX7401 filter chip, blended by a 250k pot?

I know that's not going to adjust the corner frequency of the filter, but it might be good for killing harmonics going into a distortion or octave circuit.

[Seljer]

Blending like that would also result in a kind of shelving filter. More or less the corner frequency at the start of the slope would stay fixed and the frequency at the end of the slope where it flattens out would move forwards/backwards as the shelf moved down/up.

For the crudest passive implementation of a moveable 2nd order filter, you can just take a stereo log taper pot (which is not that exotic of a component to find) and use it as two variable resistors. Wire it as two regular low pass filters in series. But you need to put a buffer after it because anything thats got a low input impedance will noticeably load it down.



But you really have to evaluate what you want to accomplish, again, to get rid of fizz in a distortion circuit anything will do and you don't need 'perfect' adjustment abilities. If you're after a more interesting 'vocal' sounding filter with a bit of resonance you basically have to use an active filter where you can change the damping.