Double Bass Smart tech people: How complicated is a high pass filter?

[robgrow]

I'm curious, have you measured this? It would be interesting to perform a spectral analysis on one of these while it is loaded with typical working impedances.

No, didn't measure it. When I talked to the main technical guy at LR Baggs, he revealed that there is a fixed mid filter in the design of the Gigpro. Subjectively, the Gigpro always sounded muffled to me, regardless of the source.

 

[fdeck]

I'm not sure what you mean by a Q of 1/2. Q, for those who are not into geek-speak, is the center-frequency (CF) of a filter divided by its (usually 3-dB down) bandwidth (BW). Thus, holding CF constant, as BW decreases, Q goes up, indicating a sharper filter. The term comes from the old radio days in which the objective was to build very sharp filters into the radio-tuning circuits. The sharper, the better. These filters were referred to as having a higher "quality-factor," thus "Q".

Now, Q is not really a proper metric for a high-pass filter because the bandwidth is (theoretically) infinite. The filter passes all frequencies ABOVE a certain "cutoff." The two most important parameters of a high-pass filter are its cutoff frequency and its rejection slope. The latter is the rate at which frequencies below the cutoff frequency are attenuated. It is specified in terms of dB per octave.

So, a simple high-pass filter with a cutoff of 50 Hz and a slope of 12-dB per octave will pass all frequencies above 50-Hz. The response at 50 Hz will be 3 dB down. Each octave decrease will see another 12-dB drop, i.e., at 25 Hz it will be 15 dB down, at 12.5 Hz it will be 27 dB down.

I think it's a different definition for high pass and low pass filters. The way the Wikipedia article seems to use Q, it controls the shape of the curve around the cutoff frequency. I graphed a typical HPF with 3 different Q values -- see below. With Q=0.5, you actually lose a fair amount of amplitude above cutoff. Q=0.707 gives you the flattest curve above cutoff, and higher values of Q give you a hump.

And I think that if you combine high and low pass filters with equal cutoff frequencies using the Wikipedia-defined Q values, the result is a bandpass with Q as you describe.

 

[drurb]

I think it's a different definition for high pass and low pass filters. The way the Wikipedia article seems to use Q, it controls the shape of the curve around the cutoff frequency. I graphed a typical HPF with 3 different Q values -- see below. With Q=0.5, you actually lose a fair amount of amplitude above cutoff. Q=0.707 gives you the flattest curve above cutoff, and higher values of Q give you a hump.

And I think that if you combine high and low pass filters with equal cutoff frequencies using the Wikipedia-defined Q values, the result is a bandpass with Q as you describe.

I looked all over Wikipedia and could only find definitions of Q completely consistent with what I wrote. Perhaps you could direct me to what you found. From your plots, when you change "Q," you are, in some cases, adding a resonance to the high-pass filter. Q, in the tradional sense, is all about resonance. Note that, however you are using it, it is not an index of the rejection slope. I'd still like to see the article to which you are referring.

 

[drurb]

No, didn't measure it. When I talked to the main technical guy at LR Baggs, he revealed that there is a fixed mid filter in the design of the Gigpro. Subjectively, the Gigpro always sounded muffled to me, regardless of the source.

Thanks for the info. They probably did that to give it a "tube sound." I despise such design nonsense!

 

[fdeck]

I looked all over Wikipedia and could only find definitions of Q completely consistent with what I wrote. Perhaps you could direct me to what you found. From your plots, when you change "Q," you are, in some cases, adding a resonance to the high-pass filter. Q, in the tradional sense, is all about resonance. Note that, however you are using it, it is not an index of the rejection slope. I'd still like to see the article to which you are referring.

The reference is: http://en.wikipedia.org/wiki/Sallen_Key_filter

But now I see what you are saying. The Q is indeed not an index of the rejection slope. The only thing being controlled is the shape of the response near the corner frequency. But when you get down in the cutoff region, it is 12 dB/oct no matter what, like you say.

By the way, I have gotten ready to try building one of these. Here is what happened. I had gotten complacent with using my magnetic upright bass pickup, and decided to try out my piezo with a new DIY speaker. The speaker went from being well behaved with the magnetic, to the cone visibly jumping out at me on low notes. Same volume level. I am assuming that the offending stuff is the subharmonic thump below 40 Hz, and my amp is only 3 dB down at 20 Hz. I think that a HPF tuned to around 40 or 50 Hz will help me control the cone without any other adverse effect.

Right now I am thinking of something like the attached circuit. It got a little bit elaborate, but the little parts are cheap compared to the pots, jacks, case, etc.

 

[drurb]

The reference is: http://en.wikipedia.org/wiki/Sallen_Key_filter

But now I see what you are saying. The Q is indeed not an index of the rejection slope. The only thing being controlled is the shape of the response near the corner frequency. But when you get down in the cutoff region, it is 12 dB/oct no matter what, like you say.

By the way, I have gotten ready to try building one of these. Here is what happened. I had gotten complacent with using my magnetic upright bass pickup, and decided to try out my piezo with a new DIY speaker. The speaker went from being well behaved with the magnetic, to the cone visibly jumping out at me on low notes. Same volume level. I am assuming that the offending stuff is the subharmonic thump below 40 Hz, and my amp is only 3 dB down at 20 Hz. I think that a HPF tuned to around 40 or 50 Hz will help me control the cone without any other adverse effect.

Right now I am thinking of something like the attached circuit. It got a little bit elaborate, but the little parts are cheap compared to the pots, jacks, case, etc.

Thanks for the reference. The Q referred to is, indeed, the resonance at the cutoff. Here is a nice link.

In my experience, I've never heard the shape of the rolloff referred to as Q. That is just a statement about my experience and not a judgment of whether it is right or wrong. I'm used to referring to the slope (rejection rate) and whether the shape is Butterworth, Cheychev, etc. I'll look into it.

Now, about the cone jumping around with a piezo pickup! Check here and here.

 

[fdeck]

DR, thanks for those links. I'm going with your judgement here. The circuit diagram shown above is Butterworth. But note that one of the resistors on the last op amp is labeled "Pot2." Now that I read that one link, it is worth noting that Pot2 smoothly varies the resonance from the lowest of those curves to the highest. I though of making that a trim pot, but have not figured out a good use for it yet. And I will try to take my HPF circuit for a spin soon.

 

[fdeck]

I looked at Link Removed, and found a 3rd order filter circuit that can be built using JFETs. It seems to model well in LTSpice, and should not be too hard to build by hand. The input is piezo-compatible, and has an additional filter pole, so the overall slope is 24 dB/oct. I can always decide later if the sharper knee of a Butterworth is worth the butter of going to op amps.

 

[bolo]

I've heard complaints that the Gigpro color the signal quite a bit. Can anyone verify or shoot this down?

Hey Chris, like robgrow, I thought the GigPro colored my sound. To my ears it made the overall tone much fatter, more tube-like if you will. I kinda thought it was scooping the mids, and it looks like maybe it was after all based on robgrow's comments.

After listening carefully one day, I decided to stop using it because to my ears it took too much off the initial hit of each note. Maybe it was scooping too much of those crucial mids and upper mids. But it smoothed out the attack too much for my tastes. It was almost like a compressor the way it seemed to affect the envelope or "boing" at the start of each note.

I know other folks dig the GigPro and other L.R. Baggs gear though. The GP probably reacts differently to different types of pickups and input sources. I was running a Rev SOLO into it. Perhaps other folks will chime in with their results using other types of p/u's too.

 

[drurb]

I looked at Link Removed, and found a 3rd order filter circuit that can be built using JFETs. It seems to model well in LTSpice, and should not be too hard to build by hand. The input is piezo-compatible, and has an additional filter pole, so the overall slope is 24 dB/oct. I can always decide later if the sharper knee of a Butterworth is worth the butter of going to op amps.

I'm confused. The design at that link is Butterworth. Does your modification using JFETs change the shape of the response? Anyway, 3rd-order Butterworth's are 18 dB/oct. Is yours really a four-pole. Can you make the cutoff adjustable? As long as we're getting carried away, while Butterworths are maximally flat in the passband, this comes at the cost of poor transient response. The best transient response is achieved by a Bessel filter.

 

[Chris Fitzgerald]

Hey Chris, like robgrow, I thought the GigPro colored my sound. To my ears it made the overall tone much fatter, more tube-like if you will. I kinda thought it was scooping the mids, and it looks like maybe it was after all based on robgrow's comments.

Yeah, that's a bummer, since if it was a clean signal path, it would be the cheapest and easiest way to carry a portable HP filter in my bass case. If anybody else finds small HP filters anywhere, post a link here!

 

[fdeck]

I'm confused. The design at that link is Butterworth. Does your modification using JFETs change the shape of the response? Anyway, 3rd-order Butterworth's are 18 dB/oct. Is yours really a four-pole. Can you make the cutoff adjustable? As long as we're getting carried away, while Butterworths are maximally flat in the passband, this comes at the cost of poor transient response. The best transient response is achieved by a Bessel filter.

Mine is an ill-conceived 4-pole because of the RC on the front end. But as we discussed in PM, I probably need to do more thinking than posting for a little while until I get a better understanding of active filters in general. Still, I am pretty sure that the op amp circuit shown above is a winner if one can stand the complexity. But I am off chasing something slightly different right now.

 

[Chris Fitzgerald]

Since all of you smart gentlemen have your thinking caps on already, let me throw another quick question into the mix:

The other day at the belchfest, one of the neat little devices there that made all of the piezo pickups sound better was a Summit Audio line level preamp that had variable input impedance and a phase switch. I'm sure that if I asked about the impedance thing, you guys would start quoting Einsteinian formulas that would make my head spin...but that little knob was amazing to help dial in the desired amount of "thickness" into the piezo sound. I'm not sure what to do with that except maybe buy one , but it was interesting to be sure.

But the other thing it had was a phase switch, and for the Full Circle pickup, I again noticed that just flipping this switch cleaned up the sound a great deal and removed a lot of the tubby artificial bottom that I use the HP filters on the Focus to do. So the question is, what the hell does the phase switch do? I mean, sure, it inverts the signal, but how? Is it just a matter of reversing the two wires so that "lead" becomes "ground" and vice versa? Could I accomplish this by simply reversing the wires myself?

 

[mje]

...Summit Audio line level preamp that had variable input impedance and a phase switch. I'm sure that if I asked about the impedance thing, you guys would start quoting Einsteinian formulas that would make my head spin...but that little knob was amazing to help dial in the desired amount of "thickness" into the piezo sound..

If you look at the input to almost any audio preamp, the first thing you'll see is a resistor going from signal to ground. That resistor sets the input impedance. (I'm oversimplifying, but not much)

As you vary the value of the resistor, more or less current is drawn from the pickup or transducer plugged in to the preamp. As you draw more current, the low end of the transducer starts to roll off. Variuable impedance lets you control that.

Also, piezos are resonant- they'll have a peak response at some particular frequency. An ideal transducer will have that resonance outside the audio specturm of interest, but I don't htink I've ever seen one that met that description. Variable impedance will also shift that resonance around a bit. Ideally, it would be nice to have a preamp with variable reactance too (capacitance and inductance) to tune this out. I've never seen this done for instruments, but it's a common issue in phonograph preamps.

But the other thing it had was a phase switch, and for the Full Circle pickup, I again noticed that just flipping this switch cleaned up the sound a great deal and removed a lot of the tubby artificial bottom that I use the HP filters on the Focus to do. So the question is, what the hell does the phase switch do? I mean, sure, it inverts the signal, but how? Is it just a matter of reversing the two wires so that "lead" becomes "ground" and vice versa? Could I accomplish this by simply reversing the wires myself?

Sort of. Problem is that one lead of your transducer/pickup is connected to the shield- if you just flip the wires, the shield isn't grounded anymore. What the preamp does is electrically invert the signal- what was going positive goes negative and vice versa. Simple to do with a single transistor or op-amp.

I suspect what happened when you flipped the polarity of the signal is that the speaker was just far enough away so that you got some cancellation between the output of the instrument itself and the speaker. WHen the sound at certain frequencies reached your ears, it was out of phase and cancelled out, causing a big dip in response. The problem is that this won't sound the same to someone in a different position, or if the speaker is farther away.

 

[drurb]

If you look at the input to almost any audio preamp...

Great job, mje! Varying the input impedance, of course, varies the effective RC time-constant and changes the response. I wouldn't describe it as being about current-draw, per se, but that is a fine point.

 

[mje]

Great job, mje! Varying the input impedance, of course, varies the effective RC time-constant and changes the response. I wouldn't describe it as being about current-draw, per se, but that is a fine point.

I didn't want to get into piezos having reactance, and in restrospect, the remark about current was sort of heading down the wrong path...

I have yet to see some company offering an integrated piezo-preamp combo specifically designed to compensate for the reactance in the piezo. Maybe that's be my next are of experimentation. Right now I'm playing with embedding piezos in bridges (courtesy Rob Wilson, who supplied me with a few cheap bridges removed from instruments).

 

[fdeck]

I didn't want to get into piezos having reactance, and in restrospect, the remark about current was sort of heading down the wrong path...

I have yet to see some company offering an integrated piezo-preamp combo specifically designed to compensate for the reactance in the piezo. Maybe that's be my next are of experimentation. Right now I'm playing with embedding piezos in bridges (courtesy Rob Wilson, who supplied me with a few cheap bridges removed from instruments).

Well... if you want to try a variable input capacitance, it would be hard to do with a continuous control because variable capacitors in the nano-farad range don't exist. But I think you could get there with a selection of capacitors in 1/2/4/8 sequence and a binary encoded rotary switch. It would not be expensive.

 

[dbas]

Having not perused the Talk Bass web site in a long while, this question caught my attention. As a Bass player I get the esthetic part of ones sound and how much it affects my playing, but my "day job" is with a company that specializes in electronic filters. We chiefly focus on signals in the Audio and Sub audio ranges.

For this and any other question about how filters are designed - even ones in our amps, preamps and anywhere else, please visit our site: www.freqdev.com

If you find you still have questions, please call me at X130

Don

 

[mje]

Well... if you want to try a variable input capacitance, it would be hard to do with a continuous control because variable capacitors in the nano-farad range don't exist. But I think you could get there with a selection of capacitors in 1/2/4/8 sequence and a binary encoded rotary switch. It would not be expensive.

I was thinking of coming up with matched pickup/preamp combos. Breadboarding a prototype you'd just try different values. You might want a trimmer to tweak a specific setup. (Although I did a search and found compression trimmers up to 2nf.)

 

[dbas]

Yes, you are correct that a source such as peizoelectric film has an inherent "reactive" resistance. What you may be missing is the purpose of the little preamp circuit attached to your pickup, and what it's function is. I don't know the level of electronic skill here in this group yet but the term "feedback" on the input amplifier (often called gain but it can be a gain of less than one too), is designed to remove the reactive component of your transducer and produce a linear response to the stimulus of the transducer. That linearization of the preamp, and how well (or badly) it deals with the dynamics of a real source is the frequency response of your preamp. Sort of like what you need to take into account when designing a power amp that has to deal with the back EMF produced when your 18" sub is pushing 400 Watts of a low B string and the cone has to go back to the rest position. In that case it is how to design for good damping factor of the amp.

A pretty good description aout the peizo film transducer can be found here:

[Invalid or Expired Link Removed]

- D