High Pass Filter Question

[BuddyPrince]

I'm working on fine tuning a clean boost circuit, the Fetzer Valve, for bass.  I'm wondering if the 22n and 100k potentiometer (volume) are acting as a high pass filter.  I know that a capacitor followed by a resistor to ground is an HPF.  I'm assuming a variable resistor is no different, even if it's for volume control.

I'm getting some low end loss in my pedal, which has an overdrive and active 2-channel mixer.  I'm aware the low frequency loss may be at any of these other circuits too, but when I look at the passive HPF's in those circuits, they should cover bass frequencies to <30Hz.

Currently I have a 50k pot in place of a 100k pot because that's what I had on hand.  A 22n capacitor and a 50k resistor give me a cutoff frequency of ~145 Hz. A 100k pot works out to a cutoff of about 72Hz, still not low enough for bass.  I could drop a 100n cap in there with the 50k pot and get down to 31 Hz or a 220n and get down to ~15 Hz, well below electric bass frequencies, that is if my assumptions are correct.

[GibsonGM]

Hi Buddy, answer is "yes", and it varies with the setting of the pot, and what follows the circuit.

Volume all the way down, the cutoff frequency is 72Hz., so anything higher passes.

At say, 10k, almost fully "up", that becomes 723 Hz! 

These numbers, however, are affected by what is following the volume pot; the next "thing's" input impedance.    You probably have something more than 10K following this (I hope!)...could even be 1 Meg.   So the cutoff frequency is shifted lower, more energy can pass, and we don't actually *hear* only frequencies above ~700Hz! 

In order for the 'filter' to have that crazy effect, you'd have to buffer it so as to not be affected by things downstream...hope that makes a little sense....for 'fine tuning', I'd literally connect it to what you will usually play with after it, or your amp, and choose your output cap.   Not much value, IMHO, to doing a bunch of math as that won't tell you what it SOUNDS like, and it is complicated because there's an input cap for the next ckt in the way, too. 

[BuddyPrince]

It's almost never a simple answer with electrical circuits, is it...There seems to pretty much always be a "Yes, but..." answer. 

Thanks GibsonGM, makes sense.  Sounds like the best thing to do is some trial and error with everything plugged in and see what works the best.  That's easy enough to do...

Thanks!

[GibsonGM]

You're welcome!   And yes, you're right, it's often "not that simple".  But a great question!  Made me go open LT Spice and try a few things to make *sure* I was telling you the way it is, which is good!

Like most things involving the output of an effect...the higher the following *thing's* input impedance is, the lower the effect that pot will have.  Makes sense, it swamps that pot, being in parallel with is as far as AC is concerned. 

And impedance is just "AC resistance (reactances) plus DC resistance", to your AC guitar signal, so the concept is kind of like resistance. 

[PRR]

The impedance of the un-loaded potentiometer-connected pot will not drop as you turn it up. It is always 100K (on the original plan).

In real life you connect stuff to the wiper. When wiper is full-up, yes the load drops.

Let's pretend you used a 100k pot, and that a typical guitar-cord load may be 100k. Full-down, or way-low, this is 100k; full-up is 50k.

And when you only have a 50k pot, bass is less.

A full analysis should also include what is driving the C-R network. In this case the JFET Drain is near infinity shunted by the DC load resistor. Plan says 100K pot, trimmed for ~~4.5VDC. I'll ballpark a typical setting as "22K".

Agree: you want to be looking in your hundred-nFd drawer.

[GibsonGM]

The impedance of the un-loaded potentiometer-connected pot will not drop as you turn it up. It is always 100K (on the original plan).

In real life you connect stuff to the wiper. When wiper is full-up, yes the load drops.

This is what I was getting at...not too many of us have a way to monitor audio without putting some kind of load on there!   The load following the cap/pot output network is going to interact with it. 

But usually not in a bad way, providing the input Z of the following is high enough.

[antonis]

The load following the cap/pot output network is going to interact with it.

In a "peculiar" way, I should dare to add, Sir.. 

Despite load's value, output pot doesn't behave like its original taper (Lin, Log, Rev Log, etc) anymore due to permanent load setting in parallel with pot's wiper..

[GibsonGM]

The load following the cap/pot output network is going to interact with it.

In a "peculiar" way, I should dare to add, Sir.. 

Despite load's value, output pot doesn't behave like its original taper (Lin, Log, Rev Log, etc) anymore due to permanent load setting in parallel with pot's wiper..

Absolutely!    And if we do not know what will follow the circuit, we can't really know how that pot will REALLY behave.  BUT - we can make assumptions and they are close enough.

Most any guitar effect will have an input impedance ("Z") over 10k,  and really over 50k is most common, FF and a few others excepted.  Modern effects using opamps and Mosfets, JFETs (and triodes) in the input could be 1 Meg or more!      So, we assume it will be "high".    Most all the time, our output will behave very much like we designed it to, because our ears cannot hear MUCH of a difference (once in a while, though...).   

It is safest to use good design practice, and then blame the effect that someone plugs in after for having low input Z      Seriously though, trial and error based on what you see that others have done is a great way to size those output caps and pots.

[antonis]

our ears cannot hear MUCH of a difference ..

To be more specific, no difference less than 3db (Voltage ratio of 0.707 or Power ratio of 0.5).. 

trial and error based on what you see that others have done is a great way to size those output caps and pots.

Hmmm...

So, I inaccurately thought plagiarizers only prosper in tropical climates.. 

[GibsonGM]

Oh, no, Antonis!  We do not like to re-invent basic things....if it has been done for you, you must take advantage of this!!   

We do things, and we don't even know WHY we are doing them, ha ha...

[amptramp]

The effect of the high-pass can be heard well below and above the nominal -3db (power, -6db voltage) frequency.  The effect of phase shift can move the peak amplitudes of harmonics so the pluck of a string can sound indistinct.  Since there are no subharmonics generated by this circuit, excess bass response is not a problem.  A 100 nF capacitor would be good but I might go with 220 nF.  It is standard practice among high-fidelity builders to place the high-pass frequency at least 3 octaves below the minimum frequency you can hear.  In this case, you only need the lowest frequency you can input which from a bass guitar is 41 Hz so 5 Hz is good for fidelity.

[ElectricDruid]

If it was me, I'd stick a 470n or even a 680n film cap in it and not worry. Those are the largest values that are both cheap and not polarised. 470n is a good top limit in many ways, and I often design around that. I don't want to *require* the use of 1uF non-polar caps if I can avoid it.

T.

[antonis]

It is standard practice among high-fidelity builders to place the high-pass frequency at least 3 octaves below the minimum frequency you can hear.  In this case, you only need the lowest frequency you can input which from a bass guitar is 41 Hz so 5 Hz is good for fidelity.

But what about 50/60Hz mains noise, Ron..??

[GibsonGM]

Yes, it's well-known that high fidelity folks will do this (3x below min freq)...but isn't narrowing the bandwidth really what guitar amplification and effects are all about?  Our speakers don't give faithful reproduction, and our amps aren't designed to reproduce *everything* that we can generate.....and what we CAN generate is pretty limited, in the real world.   The more we spread out, the more chance we have to experience problems and instability in overdriven and distorting circuits...

It's a sound principle to cut out as much low as possible/reasonable to reject mains noise, as Antonis notes, and also to reduce 'low end flub' (and blocking distortion in triode circuits).    On the other end, we want to tame the shrill, too, of course.

Unless I was shooting for a hi-fi type of reproduction, I'd stay away from the low or really high extremes!   I set a lot of hi-pass things at like 150Hz, 200Hz, and count on the slope to pass "some level of lows below that", maybe down to 100Hz at lessening energy levels, of course.  I don't notice any absence of bass, but I do note the absence of "boom"...once in that 150Hz neighborhood, I tune by ear using caps....YMMV

[amptramp]

There are several good reasons for extending the low-frequency response.

1. A larger capacitance connects the source in parallel with the stompbox input making the thermal noise and particularly the 1/f noise lower.

2. You cure feedback instability by having an extended response, not by restricting it.

3. Normally the human ear is not sensitive to phase but there are exceptions and one is a sudden step function like plucking a string.  Restricted low-frequency response shifts the phase of the signal so that the release of a string which should start all harmonics at their peak point instead move the starting points so they don't coincide.

4. You can reduce a lot of the problem with low-end blocking and flub with proper biasing in triode circuits and by splitting the frequencies in non-linear distortion/fuzz circuits.  After all, you don't need to generate higher harmonics of signals that are already high and clipping low-frequency signals also clips any high-frequency content riding on it, so the low frequencies can be passed with minimal distortion so you don't get muddy sound.

[GibsonGM]

There seems to be disagreement in this area, Ron...I've read sources that detail what I wrote above...which is different than what people in the hi-fi would are after, or course.  They're VERY interested in energy out at 15k, higher....not a whole lot for guitar up there, or below 80Hz.   Maybe the drop-D crowd would balk, LOL...but even then, the energy output of the bass strings dwarfs that of the higher ones...

I won't 'argue'  much for 'my position', only put it out there for others to check out; design in whatever way makes you comfortable.  Many modern designs are 'light on the bass', one can look that up on their own....boosting it (22u bypass caps etc!) are early Fender days, and are a relic now, that started with hi fi.   If you were to introduce a higher order of complexity to a circuit, of course you could do what you like with bands.   Me - I don't LIKE a whole crap-ton of level below 100Hz (or even higher)...to me, that sounds like poo.  But some ppl surely do, playing cleaner or what have you.

How does extending the response work to cure feedback???  That goes 1,000,000% against everything I've ever read or tried.  Not talking about intentional FB loops, I am talking about unwanted coupling etc.  Cranking up the amplification on freq's outside bands we can reproduce (well) or even hear...recipe for disaster, no?

Designing for frequencies our speakers can't reproduce, and well outside where most of the energy of our instrument is generated, seems counter-intuitive, at least to me.    Yes, you CAN split frequencies and treat them all very independently, that's a whole very interesting area (!)...basic stompbox, tho - not that level of complexity.   Same for triode bias...don't see that happening very well in 4 stages, going to need some more I think...

YMMV, clearly it's working for you.

[amptramp]

If you take a good look at the Big Muff Pi:



you can see band-limited clipping at work.  C6 and C7 define the lowest frequency to be clipped.  C11 and C12 define the highest frequency to be clipped.  These are not brick-wall filters but they make higher inputs necessary outside of the midrange frequencies.

Similarly, the Tube Screamer has antiparallel clipping in an op amp feedback stage but since it is a non-inverting clipping stage, the gain does not go to zero (like it would with a DOD250 with diodes to ground in the signal path), it goes to unity gain so even in deep clipping, it has unity gain and the original signal rides through, preventing excessive muddiness.  The capacitor across the two diodes has the same effect as in the Big Muff Pi, it decreases the high-frequency content and raises the signal level required to clip as frequency goes up.  These are examples of circuits designed to clip within a certain band without a great deal of extra complexity.

As for frequency response in feedback amplifier stages, you need a feedback lead capacitor across the feedback resistor to ensure there is a dominant pole that drops the gain below unity before any other poles kick in.  Similarly, coupling capacitors have to be large enough that the pole from the output transformer (tube) or capacitor (transistor with single-ended power supply and no bridging).  If you have any rise in the frequency response at either end of the spectrum, the amplifier is marginally stable or completely unstable.  Trying to limit bandwidth within the forward loop will push the amplifier towards instability.

I agree that you don't need a lot of frequency response for a guitar - a treble guitar has nothing below 82 Hz and the high end can be reduced successfully - if you look at "No Jacket Required", it was a great album from 32 years ago that had nothing above 5 KHz in it.  But higher fidelity may help the signal become more crisp when a single string is plucked.

[PRR]

> A larger capacitance connects the source in parallel with the stompbox input making the thermal noise and particularly the 1/f noise lower.

1/f rumble "should not" be a problem in guitar work. (It once caused me real problem in a phono preamp.)

> (22u bypass caps etc!) are early Fender days, and are a relic now

Fender started when there were no good bass amps. The guitar's belly had to assist the kickdrum, upright bass, or tuba. Ampeg worked for bass and eventually the 300W two-15" bass rigs flooded the bottom of the band. Now the guitar should shave bass to avoid mud and allow more energy in the tenor/alto notes. So Things Change. History is not a sure guide because makers sometimes cling to old patterns (especially when they get bought-out by clueless capitalists).

[GibsonGM]

Thanks for the post, Ron...but the BMP sort of seems to be supporting what I was saying...that there is a 'range' selected for clipping...lows attenuated, highs controlled...clip in a certain band.   I may not have been clear enough before - what I was saying was intended for stages that will experience at least some degree of clipping.   I have very little use for stages that don't have a CHANCE of clipping, LOL.   

Of course, a trem, phaser, etc...you don't want that...but I still would trim the low out of it for clarity.   If I wanted some PURELY clean boost, I COULD justify boosting some bass - I would make it variable....for use with the Les Paul center pickup setting, 50s style jam   

As for amps using feedback, what you said re. poles.  I don't much like feedback, personally, but many many do, very Fender-y and classic, I know.   "Stability" I was talking about is re. high gain preamp stages...going broadband with them is, in many circles, considered a no-no.   Just like the BMP clipping within a certain band... 
But however one 'gets there', that is great, as long as you arrive!

Yes, Paul - a lot of designs, when you see 'em - you can tell whoever 'added on to them' didn't know what the parts they retained actually do, ha ha...running pentodes with gain 10, no grid stoppers, etc etc...lifted right from 50s designs.

[ElectricDruid]

Who'd have thought a simple clean boost circuit could cause such anguish?!? Only amongst dedicated followers of the art of pedal building...lol Us lot!

Don't mention Politics, Religion, Football, or the frequency range required for a guitar, apparently.

T.