Custom Hi-Pass Filter Stompbox

[AudioEcstasy]

I was approached by someone to build them a hi-pass filter stompbox. They want to be able to select the cutoff frequency, and they would like to have about 4 options.

They would also like it to be passive (require no power), and no switch. I.e. always on.

I've been having some problems in the design. What would you guys recommend I use to set up the switching? I was thinking of like a five-way toggle or something similar.

I can't seem to get a jump on this design and all advice is welcome!

[ashcat_lt]

Any passive filter is going to be extremely sensitive to the impedances around it.  That's fine if it's only ever going to be used in one situation, and you can get realistic specs for both source and load.  Then it's just a set of caps, maybe a resistor, and a switch.  I would try to put all the caps in series and shunt them as necessary to get the different cutoff points.

[R.G.]

What he said.

Plus: RC filters inherently have Q's under 1 and sometimes much under 1. It is quite difficult to make an R-C filter with more than one-pole (i.e. more than 6db/octave rolloff) and keep any kind of reasonable performance. Ditto for R-L filters.

LC filters can sidestep this issue because they are inherently non-dissipative (that is, they can ring like a bell), but they suffer from being hard to design and hard to manufacture. As well as hard to tune. And custom inductors are expensive. Sometimes REAL expensive.

These difficulties are why the engineering world left passive R-C or R-L filters and L-C filters for audio long ago, and opted for R-C active filters. One comes out way ahead on the triple constraints of good, cheap, and fast with RC active filters compared to other filters.

And this is what you and your friend are facing. He can pick any two of good, cheap, and fast. Good passive filters are in general not cheap. Cheap filters in general are not good. Fast filters are usually not good or cheap if they're not active filters (and usually digital to boot on that one).

One could argue that passive filters are so very sensitive to the impedances around them that the buffering needed costs as much as an active filter, and also makes it not passive.

As an alternate suggestion: go design (or copy) a state variable filter with a few opamps and a dual-gang pot. You get truly wide range, high performance, independent control of frequency, Q, and gain, and simultaneous high pass, low pass, band pass, and notch filters. The only catch is it can't be passive.

[Keppy]

I designed a custom continuously-variable LOWpass filter using OTAs. As many poles as you want controlled by a single-gang pot. The datasheet has filters you can copy. If you want it passive, I see only two practical options:

1) Get his rig, plug in all his stuff, and tune your filter by ear. Even then, the cutoff frequency may change if his signal chain is not buffered at the beginning. If there are true-bypass effects before the filter, they will have a much lower output impedance when on than the guitar does.

2) If he runs any kind of buffer in his chain ahead of the filter you might be okay. If there are buffers both before and after the filter in his pedal board, then your passive box is no longer passive, functionally speaking, and I see no real problem (well, as long as you're happy with a single-pole filter).

As far as the switching, rotary switches seem to make the most sense if you're using more than three options.

Question: Given that a guitar is already a high-impedance signal, have you tried just rigging up a switch to select different coupling caps? You might not even need an entire RC circuit, and that would have the least effect on the signal's overall AC impedance. The effect would differ from guitar to guitar though, so definitely get the customer's own axe to dial it in.

[R.G.]

What he said.

[wavley]

Question: Given that a guitar is already a high-impedance signal, have you tried just rigging up a switch to select different coupling caps? You might not even need an entire RC circuit, and that would have the least effect on the signal's overall AC impedance. The effect would differ from guitar to guitar though, so definitely get the customer's own axe to dial it in.

The Fender Jaguar's choke switch is just this, a .003 cap in series with the pickups that can be switched on and off. I ditched mine in favor of a bridge polarity flip when I switched to series/parallel wiring and kinda miss it.  I've been thinking about putting it back because I don't ever use the out of phase sound.

[AudioEcstasy]

Thanks for all of the wisdom everyone! Unfortunately, I'm unable to test his rig as I'm in LA and the client is in Nashville.

I've never really attempted anything like this, and I'm enjoying the challenge.

[lion]

This may not be exactly what you are after – but I thought I could share it anyway.

In circles dedicated to replicate british 60's guitar instrumentals (The Shadows) there's a HP filter widely used. It was at one point offered as a commercial product by 2 parts, but production stopped years ago due to the limited interest/sales. Since then the circuits used in one of the pedals was publiced for DIY (rumours has it the circuit was identical in both products) .

The filter was inspired by (lifted from) parts of the circuit used in some early Vox amps (which was speculated to be parts of the early "Shadows sound"), and adapted for use just ahead of amp input.

It's a simple passive 4th order RC filter – and the steep roll off obviously emphasize midd/highs, but also makes it possible to use bass controls on the amp differently for alternative low end response not possible otherwise.

The filter has been a permanent part of my '60's instrumental rack' for years, and I have added the possibillity to bypass one/more RC stages for a less steep roll off - which in essence is percieved as a shift of the cutoff freq.

I understand the issues explained by RG/others here – but IME this filter works OK (sometimes it's an advantage not to know what you are doing, from a technical poins of view   ) I have it just before amp input (AC15/30) or in front of a vintage tapeecho sim, and there seems to be no appearent differences using different guitars (I do have a buffer stage earlier in the signal chain).

Hope this of interest.

lion

[John Lyons]

Thanks for that schematic lion!

[AudioEcstasy]

I found this schematic:

Google Image Result for http://ok1ike.c-a-v.com/soubory/tipy/HIGH%20PASS%20ACTIVE%20FILTER.JPG

It notes that it's for a 100Hz cutoff. I could just interchange parts to allow for different cutoffs, and wire them to a rotary switch, yes?

[artifus]

yes. double the cap value to halve the frequency. ish.

[AudioEcstasy]

So I can double the cap value without having to change the resistors?

Halving the frequency would be 50Hz, what about a circuit for 80Hz, 60Hz, etc?

I'm new, but always learning. Thanks for the help!

[artifus]

http://sim.okawa-denshi.jp/en/Fkeisan.htm

[AudioEcstasy]

I found this, seems simple enough.

http://www.muzique.com/schem/filter.htm

I can't seem to get anything as a result but NaN?

[artifus]

same math in both links. check your zeroes and decimal points - ohms, kilo ohm, mega ohm, farad, micro, nano, pico etc. math ain't my strong point either but you're gonna have to try if you wanna understand enough to do sh!t on your own. japanese link is more comprehensive - take the time to study the graphs, breadboard and listen to come to your own understanding of what's going on. be aware that components have 'tolerence' and trust your ears over any calc or scope. i don't mean to be obtuse but you wont find all the answers via the internet. you're gonna have to earn what you charge for.

[R.G.]

I found this, seems simple enough.
...
I can't seem to get anything as a result but NaN?

It's because you don't understand the math that the filter is trying to hide from you.

I have a long-standing dislike of packaged calculators. They often lead to an inability to think about the underlying problem.

Any single R-C filter has the same cutoff frequency. It's F = 1/(2*pi*R*C). Period. The relative positions of the R and C determine whether it's a high pass (signal hits the capacitor first) or low pass (signal hits the resistor first). Using a magic calculator will often prevent someone from *ever* learning the way to calculate that frequency.

Worse, it leads you to not understanding what the other parts that are not and never can be put into the calculator are doing to you. There is an impedance of the source that drives the filter and the input the filter feeds that also affects the filter frequency.

Beyond that, single-R, single-C filters are very limited in the tone shaping they can do; so are multi-R, multi-C as long as you stay passive.

And, as art said, if you want to understand enough to do it on your own and charge for it, you're probably going to to have to to the necessary learning.

[artifus]

There is an impedance of the source that drives the filter and the input the filter feeds that also affects the filter frequency.

any chance you could elaborate (just a little please - baby steps) as i still struggle with the math and lean heavily on calcs myself as starting points and would like to understand more. thanks.

*calcs not clacs - doh!*

ok, i've thought about it a little but may be barking up the wrong tree in the wrong forest yet again, but... impedance effects frequency due to inductance (still fuzzy on that too) which both r and c have, which in turn affects filter response? are we always just vaguely aiming for the ball park? i guess so in analogue guitar/audio circuits where it all gets quite subjective and down to personal taste beyond the point that it just works. i'm rambling again, prolly time for bed... aahhh - so if we go active we can control that input impedance and have a little more control over what's going on... bear with me, i'll get there one day. i'll stop posting so much and start thinking a little more...

[R.G.]

any chance you could elaborate (just a little please - baby steps) as i still struggle with the math and lean heavily on calcs myself as starting points and would like to understand more.

Absolutely. We all start with baby steps.

First, let's take a step into the imaginary world of perfection. Imagine a perfect (or ideal) voltage source. This is kind of like an infinitely big car battery. It puts out the specified voltage on its terminals no matter what is connected. No matter what. If you have (or imagine) an ideal 10V voltage source, and connect a 1M resistor, then ten microamps of current flows. If we connect a 1 ohm resistor, ten amps flow. If we connect a thick copper wire with only 10 milliohms of resistance, then 1000 amps of current flows. If we connect a 1micro-ohm resistor, then ten million amperes flows. You see where this is going. The voltage is *constant* on it output.

Then there is an ideal current source. It adjusts the voltage on its output to make the specified current flow. If we have a 1A source, and connect a 1 ohm resistor, then the voltage is 1V. A 1M resistor still gets 1A, but the current source puts out 1,000,000 volts to make it flow.

There are no ideal voltage or current sources in the real world. But we can use the ideas to model real sources. Real-world voltage sources act very much like ideal voltage sources with an internal resistance inside them where you can't get at it. So a real world voltage source will not provide infinite current, but only some maximum current, and its apparent output voltage sags as you pull more and more current. A real-world "9V" battery might have an open-circuit voltage of 9.3V if it's fresh. If you connect positive to negative with a fat copper wire and measure the current in the wire, you might find 2A flowing. The battery limits its output current internally. It acts like there's a resistor in there, dropping the output voltage as current goes up. In this case, the internal resistance is 9.3V/2A  = 4.65 ohms.

That internal resistance is the source impedance of the battery.

All real-world voltage sources act something like this. They have internal resistances. This accounts for why you can load them down. Their internal resistances make the output voltage drop lower as you pull more current.

ok, i've thought about it a little but may be barking up the wrong tree in the wrong forest yet again, but... impedance effects frequency due to inductance (still fuzzy on that too) which both r and c have, which in turn affects filter response? are we always just vaguely aiming for the ball park? i guess so in analogue guitar/audio circuits where it all gets quite subjective and down to personal taste beyond the point that it just works. i'm rambling again, prolly time for bed... aahhh - so if we go active we can control that input impedance and have a little more control over what's going on... bear with me, i'll get there one day. i'll stop posting so much and start thinking a little more...

Let's hold up here and be sure you have the "source resistance" idea down, then we'll go on to inductance, frequency response, etc.

So - is the idea of real voltage sources, including batteries, AC wall sockets, guitars, CD player outputs, whatever, having an internal resistance where you can't get at it making sense?

[artifus]

So - is the idea of real voltage sources, including batteries, AC wall sockets, guitars, CD player outputs, whatever, having an internal resistance where you can't get at it making sense?

yes, excellent - getting there,  thank you. i like analogies. i'm finally getting around to filling in some of the gaps (gaping holes in places) in my knowledge - a lot of that 'boring stuff' that i used to skip over and try to work around is actually becoming pretty interesting. i'm even starting to get the calculator out occasionally! thanks again - i appreciate you taking the time.

[R.G.]

OK. So every voltage source in the real world acts like it is an ideal voltage source with a resistor and possibly other junk attached inside it before it gets out to the real world.

The "other junk" gets important for AC signal sources. A guitar pickup for instance. Guitar pickups are coils of wire with something like 4K to 18K of wire resistance, but also 2 henries to 4 henries of inductance, because they ARE a coil of wire making an inductor. The string wiggling around the magnetic field makes the signal voltage we want in the coil of wire, but then the signal has to get through the resistance and inductance to get to the amplifier.

So one simple electronic model of a guitar pickup is the guitar signal as an ideal signal voltage generator, but with a resistor and an inductor in series with it. It turns out that this model, with the big inductance, is why we need 1M of input impedance to properly amplify a guitar signal. The impedance of the inductor increases with frequency, so that when the signal voltage is up at about 5kHz-7kHz, the inductance is about 100K or so.

But that's an extreme example. In most cases for audio, we can simply think of the impedance of a signal source as an ideal voltage generator in series with an internal resistor. We (guitar electronics nut-cases  ) happen to deal with the extreme case all the time.

If you're OK with that, we can dive into why the impedance affects filters.