Choosing R-C Combo for High/Low Pass Filter Question

[rocket8810]

Hmmmm.......R.G. and Paul I think get what you guys are say. So if I understand correctly and in simple terms, we want the filter to go beyond where we really want it in order to ensure we accomplish our goal, meaning the 10% frequency.Now my question is, when I was reading about frequencies to filter out one was 16Hz and below. So going as low as 3.2Hz wouldn't that cause trouble?

[R.G.]

Yep, that's a problem.

Worse, passing 32Hz and attenuating 16Hz by a lot is HARD because they're so close together. A single R-C filter simply won't do that very well, because by definition, a single-pole filter will suppress no more than half-power in an octave. A three or four-pole filter might do OK, depending on how much you need the 16Hz suppressed. A four-pole filter would give you something like 24db suppression between 32 and 16Hz, which is about 1/20th the power. Is suppressing 16Hz by 95% enough?

But now you're messing with ripples in response in the pass-band, damping issues, and a host of other minutae.

There do exist filters that will eliminate 16Hz almost entirely by putting a deep notch in the stop band response, but then it has to be tuned especially for that one frequency, and it won't do nearly as much for 18Hz and 14Hz.

Filters is one place where it's easy to over-specify yourself into not being able to meet the design requirements at all. I highly recommend "The Active Filter Cookbook" again.

[rocket8810]

As always R.G. thanks. I'm waiting from amazon for my copy of "The Active Filter Cookbook" to arrive, and I actually can't wait to dive into it some more. It seems like I fell down the rabbithole for design. You mention something about octaves and filters. Now, does that mean if I design a high pass filter for a normal bass C, which I think is something around 130Hz, it will allow all octaves of C to pass, which would include the low C around 64Hz and the C at 32Hz? The reason for me wanting to get that low is sometimes I do have my bass tuned to drop C or open C which is incredibly low, thick, and heavy sounding. I do love my doom metal want to ensure I can nail it with this design. 

[R.G.]

You mention something about octaves and filters. Now, does that mean if I design a high pass filter for a normal bass C, which I think is something around 130Hz, it will allow all octaves of C to pass, which would include the low C around 64Hz and the C at 32Hz?

I used "octave" just to mean "two frequencies related by exactly a ratio of 2".

If you design a single-R, single-C high pass to have a "cutoff" frequency of 130Hz, then it produces half the power at 130Hz that it does way higher, over 10X in frequency. 260Hz and 520Hz will be nearer full power and 1040 and 2080 will be reduced by almost nothing. 65Hz will be 12db down, and 32-ish will be 18db down. In fact, 16 and 8 will "pass", but will be attenuated by 24 and 30db, those being at 1/20th power and 1/40th power, about. 

So they all "pass". But things much below the cutoff frequency are heavily attenuated - which is what filters are for, to attenuate what you don't want. The note that things are attenuated by 6db per octave just means that every time you cut the frequency below the cutoff point in half, the resulting signal power is cut in half, too.

As a reference, a frequency ratio of 10 to 1 is called a "decade". The attenuation of a single pole filter is 20db per decade, or a loss of 9/10 of the power every time the frequency is cut by 10 (for the example of a high-pass filter). The loss of 6db per octave is numerically the same as 20db per decade. It's just the way techies talk about frequencies and power, not musical octaves.

The reason for me wanting to get that low is sometimes I do have my bass tuned to drop C or open C which is incredibly low, thick, and heavy sounding. I do love my doom metal want to ensure I can nail it with this design.

So if you need to pass 32Hz relatively un-scathed, you have to design your highpass to let it through with not much loss. Put your highpass rolloff at 16Hz or lower, or you'll lose the 32Hz you care about. If that's not useful, you're going to need to design a multi-pole active filter to cut 16 and below more, while leaving 32 almost un-touched.

[rocket8810]

Ahh, now that makes more sense. Thanks R.G. for clearing that up. Now I can get back to work on this project. Btw, I see what you meant before, this is all my fault and I accept full responsibility for this. and pretty damn happy about it too.

[PRR]

> going as low as 3.2Hz wouldn't that cause trouble?

Depends.

If you are building a small part of a larger, pre-tested, system, a pedal in a complete guitar+amp chain, odds are that you can calculate 0.3456uFd and throw-in a 10uFd. The guitar amp (and in this case the guitar) already define some "appropriate" bass cut. Your pedal's goal may be to do no further harm to the signal. And a 10uFd e-cap is a lot cheaper than most 0.3456uFd caps. (10uFd e-cap is the 10-penny nail of audio circuits.... use it wherever you don't have a good reason to do something else.)

OTOH.... if you want to scream like a woman, or fuzz like a porcupine, you *may* want to lose some bass to find your tone. This is more guess-and-try. (Though hint: most women's voices have nothing below 250Hz, and maybe little below 600Hz when working higher registers.)

OTOOH... some situations DO have large subsonics. Or you need to reduce flap in woofers that can't handle the lowest notes of the band. Then many-pole filters are justified.

[rocket8810]

Oh man, I thought I finally started to get it, but now I'm not so sure. Paul are you saying that since the filter is a small part of a bigger system, ie the rest of the pedal circuit, and eventually the amp, its not as much of a concern as long as the larger system is set up correctly? Also, what do you mean about using a 10uF cap instead of a 0.3456uF? Where did those numbers come from? Wouldn't the resistor size matter too?

[R.G.]

Oh man, I thought I finally started to get it, but now I'm not so sure. Paul are you saying that since the filter is a small part of a bigger system, ie the rest of the pedal circuit, and eventually the amp, its not as much of a concern as long as the larger system is set up correctly?

I think what Paul may have meant is that in all systems, some things matter and some things don't. And that things may matter more or less in different electronics contexts. It's your job to know the larger context and therefore know what matters on this one that may not have mattered in other situations. Like the song, The Gambler - you gotta know when to hold 'em, know when to fold 'em.

If the filter you're working on is the only thing in the system keeping some bad thing from happening by its response, then it's absolutely critical. If there are other parts of the system that are helping prevent the bad thing, then that one filter may only contribute a little, or it may not matter at all. That's one way to get a multipole filter response by the way - make each section of your system have part of the response. Or you may make one filter at one point in the system do all the filtering work, and open everything else up. Your call, and your responsibility to know.

Sorry - you just graduated to the next layer of concerns up from "how do I do a filter". 

Also, what do you mean about using a 10uF cap instead of a 0.3456uF? Where did those numbers come from? Wouldn't the resistor size matter too?

I think he means that in a situation where you have calculated a high-precision cap for a filtering function, you have to ask yourself in the context of the larger system "Do I need that precision here?" If you do, go make it so. If you can be a little sloppy to one side or the other on a value, you can free yourself to use a very, very standard part that's much cheaper than the precision part you calculated, and you may save some money on the parts, and at the same time save a lot of time trying to find a high-precision part. The numbers were pulled out of the air, I think, to make that point. And yes, resistor size matters.

Even more importantly, (more sophisicated design concept here...) resistors are available in many more sizes and tighter tolerances than capacitors, so pick standard capacitor values and customize filter responses and other precision things with precision resistors where you can. It's cheaper and easier.

[rocket8810]

Omg R.G. I can't believe you've allowed me to graduate from filters, I feel so honored. Now, only if you taught a class I would be front row. Now graduating for your course, that would be an accomplishment. 

[tubegeek]

Omg R.G. I can't believe you've allowed me to graduate from filters, I feel so honored.

Today rocket8810 is a man.

I guess I need to buy you a pen & pencil set now.

rocket8810 - it may amuse you to know that I have Chess 1458 in the collection. Maybe the best thing in it, in fact. (This photo is not my copy though.)

[rocket8810]

Haha, tubegeek. I will take whatever I can get. I'm just glad I'm getting to move past being a noob, and that the community is so supportive.

[PRR]

How much bass?

Depends.

If given a spec like "16Hz", is that a minimum? -3dB? -1dB? Or is it exact? Will there be a test?

If building a part of a whole system, what does the part do?

If it "passes everything the system passes without injury", then it probably needs to be "better" than the system.

That's why "clean effects" stompboxes, reverbs and phasers and mixers, can generally be bass-cut at very low frequency. In studio or broadcast use, sub-Hz, so nobody complains that 20Hz is off a part-dB. In stage-amp use, where transportable speakers don't go below 50Hz, a 20Hz (or 16Hz, pi are round) is plenty good enuff.

And sometimes over-kill is the best policy. Your porch sags. You compute the stresses and calculate a 0.654" prop under the center. You look around and see a chunk of 4x4 about the right length, laying idle. Use the whole 4x4 instead of shaving-out a "just right" stick.

OTOH if the part must filter-off some frequency range which is causing trouble, then clearly it has its own spec.

Simple -3dB specs are usually "minimums". In hi-fi you must be better than -3dB at 20Hz, more likely -1dB at 20Hz, and fanatics like -0.1dB at 20Hz.

OTOH in phono systems where rumble is a problem, you may realize the real spec is "not much harm at 20Hz, BIG drop at 1.8Hz". Taking "big" as much-more than 20dB, and looking at single-pole response, you have a problem. You can't quite get 6dB/octave or 20dB/decade. 20Hz to 1.8hz is about a decade. A single-pole _AT_ 20Hz is -3dB there and not-quite -20dB at 2Hz.

So a complicated frequency spec has PassBand and StopBand numbers. -1dB 20Hz-20KHz, -30dB at 1.8Hz.

It may be even more complicated. Does this "rumble" stop below 1.8Hz, or extend to infinitely low frequency? Turntable rumble doesn't, but fan rumble does increase in the sub-Hz zone. For the turntable we might try one narrow dip at 1.8Hz; the fan will require more than 9dB/oct all the way down (yet still be -1dB at 20Hz).

The way we get steep stopband slope usually leads to "ripple in the passband". If I'm filtering a subway announcement system so that Baritone Barry's deep voice does not overload the cheap 4" speakers, +/-3dB wobble in the 250Hz-1KHz band may not matter much. Hi-Fi and studio users will expect much less, under +/-1dB in the passband.

[tubegeek]

The best single-capsule place is "The Active-filter Cookbook" by Don Lancaster. You can usually get these at Amazon.

Thanks for the recommendation - mine arrived today. When I ordered it I was expecting a fat textbook but once I opened it up I saw that it has quite a lovely assortment of info, so the disappointment was momentary at best, and turned into satisfaction when I saw how easily it slipped into my shoulder bag.

Did yours get  to you yet, rocket8810?

[rocket8810]

Oh yea Tubegeek, mine came yesterday so I got a little head start on you.  . Great book so far, tones of info, and I can tell why R.G. destroyed one copy. lol