Designing a State Variable Filter for Low Impedance Pickups

[HD Evans]

I've been bouncing this idea around in my head for years, and I'm finally getting around to doing the dirty work of developing it.  The frequency response of a passive guitar pickup is that of a second order low-pass filter, and before anyone jumps all over this statement I *am* simplifying and considering the ideal case.  Depending on the number of turns of wire, magnet type/strength/placement, and the external load (input impedance, cable capacitance, etc.), one can reliably calculate the response curve of a given pickup.  DC resistance may well be the least useful commonly used piece of information regarding a pickup...  We would all be much better served by measurements of inductance, capacitance, magnet type/strength, and resonant frequency.  One university physics department (UIUC) has carried out some fantastic research on this topic and published their results online, and it's a great read for any true geeks.  Historically, the tone controls of electric guitars only shift the resonant frequency down...  What's so difficult about squeezing a little more flexibility out of the beasts?  If we desire a relatively natural response curve, it would be best to maintain a second order rolloff.  Aside from this, for ultimate variability the ideal device should have independent control of gain, resonant frequency, and Q.  One more thing...  while it would be easy enough to slap one of these filters onto any extant pickup, it won't be able to mimic as many other types because it will have a limited response.  Before a pickup reaches its' resonant peak and begins to roll off, it tends to have a relatively flat response curve...  And flat is exactly what this filter system thrives on.  To really make this brick fly, it will be necessary to design special pickups with much fewer windings and weaker magnets.  In the best case scenario, these pickups will need a resonant frequency of at least 20kHz so that when fed through a low pass filter it will give our ears what they expect to hear.  What have I just described?  A system utilizing at least one state variable filter allowing for independent control of gain, resonant frequency, and Q in conjunction with one or a plurality of coils(i.e. humbuckers) that will allow one pickup to mimic any other.  Single vs dual coil types will have somewhat different requirements, and a split coil humbucker will likely offer the best flexibility.

None of this is "rocket surgery," yet such a product does not exist.  The reason for this post is to get an idea if there is any real interest, aside from my own, in such a creation.  Custom wound pickups will be the most expensive component by several orders of magnitude, and I have started the process of soliciting manufacturers.  Circuitry will be cheap, even with an independent filter for each coil, input stages, mixers, etc.  Total price to get up and running, including all pickups and electronics should land in the $200-$300 range.  If anyone wants to help or has any thoughts to contribute, I will make all specifications and designs available freely although there will be use restrictions to prevent manufacturing (think of it as open source hardware.) 

Block diagrams/schematics will be online soon, once I have them halfway polished.  The basic idea is this:  Coil-->Input-->Filter-->Mixer-->Output-->Switching.  Stacked Volume/Tone controls could provide means to sweep the resonant frequency and Q.  In some cases, LDRs may become necessary to make this feasible, so I don't want to rule out using an external power supply.  This should be a lot of fun, and I look forward to giving the guitar world another cool toy.

[CynicalMan]

I think this is simpler than you think. Cable capacitance and resistive loading have a huge effect on guitar pickups, and removing them by placing a high-input impedance buffer in the guitar itself shifts the resonant peak and treble rolloff up by an octave or so. Once you do that, you can easily manipulate the resonant peak and treble rolloff just by adjusting the input resistance and capacitance of the input buffer. I've done it myself with great results for making humbuckers sound like single coils and vice versa.

[Ben N]

Your pickup requirement reminds me of the Les Paul Recording, reputed to have enormous tonal range & flexibility.

[HD Evans]

Cable capacitance is what will shift the resonant peak, and input load changes Q.   Can tweaking those get you closer to a single coil sound using a humbucker?  Sure, but the humbucker must have reasonable properties to begin with.  You won't get there with an overwound, ceramic magnet 'bucker that has a resonant frequency of 2kHz.  Some of the old Fender single coils were as high as 13-14kHz, and generally had a softer peak (lower Q) than their modern counterparts.  Some of my personal favorite pickups are what most people would consider underwound, yet that is exactly why they produce great tone...  The upper harmonics come through because the response hasn't rolled off to -50db in their frequency range.  I appreciate where you are coming from, but if you like the results from the pre-amps you described this *will* take it even further.  Your pickups had to be in the ballpark of what you wanted, and the pre-amp effectively took the cable capacitance and input resistance out of the equation.  What I have described will allow you to effectively change the magnets and rewind a pickup by turning a few knobs.  It's as simple of a solution as I have thought of while allowing sufficient flexibility.

[HD Evans]

Ben,

The low impedance Gibson models are the closest thing I've heard of.  I wonder how the concept would have fared had ICs been more readily available at the time.  Instead, we use relatively low fidelity equipment that has never been fazed out for some reason.  This really is a matter of putting two and two together.  I personally don't like the argument of active vs passive tone, especially since most of us don't mind using a few active pedals in the signal chain.  There have been a few tinkerers online who have experimented with low impedance pickups using matching transformers and reportedly good results.

[CynicalMan]

Of course you can't do it well with ceramic humbuckers, but you don't need a custom-wound pickup with a 20kHz bandwidth either. If all you're doing is changing the resonance and rolloff of pickups, you just need a pickup with a decent bandwidth and a handful of parts. I can put up some spice graphs later, but it's a powerful approach.

Also keep in mind that there are many features of pickups that you're not going to be able to emulate. The timbre of a pickup needs more than a different frequency response to sound like a different pickup. The harmonic sequence, attack, compression (sustain), and magnetic effects on the string all vary a lot between different pickups, and your approach doesn't address those any more than mine does.

I'm not trying to shoot down your idea offhand, but it can easily be slimmed down for what you're getting.

[HD Evans]

Don't worry about offending me, if I didn't want some input and different points of view I wouldn't have brought the idea up in a place like this   As long as we otherwise respect each other, picking over the nitty gritty details and arguing over approaches are generally beneficial in my experience.  With that out of the way, let's get back to this flame war!!! LOL

Yes, I'm building in some overkill at least for this initial incarnation.  And it will work with any passive pickups, but will be limited by their bandwidth.  My example of an overwound ceramic pickup was overkill, and just a demonstration of why I believe a low impedance type will be more flexible.  Your approach is not without merit by any means, and does have the advantage of simplicity.  As I mentioned in the start of this thread, there are assumptions that have been made for the sake of simplicity alone.  The effect of magnetic pull on strings can be assumed to generally be negative, although still contributes to a characteristic tone.  Harmonic content should *mostly* be a function of the strings, although I have considered that since the custom pickups will be dramatically underwound it will likely be advantageous to change the bobbin dimensions to allow for similar coil width.  There will be at least limited control over attack and compression because the circuit has variable gain, but by no means perfect control.  This design is not intended to perfectly emulate all other pickup types, but it should offer improved fidelity and a much broader palette than most of us are accustomed to.  I suspect you may be selling this concept a bit short, because control over the resonant peak and its' height (Q) along with the ability to change their output (creating slight mismatches between humbucking coils for instance), and varying them all continuously will not be a subtle effect. 

Using a passive approach limits variation in resonant frequencies and Q to discrete points.  If a preamp is added you can take input impedance, along with cable resistance and capacitance out of the equation.  I don't doubt its' usefulness, but we're effectively taking the same problem and tackling it in differently.  Do you have something in mind to simplify my idea while still allowing continuous variation of all variable?

[PRR]

> change the bobbin dimensions to allow for similar coil width.

No. Pole geometry is already good. Bobbin should fill the space available.

Use standard parts, except half as many turns of wire half the area (0.7X diameter). That gets inductance to half. Buffer AT the pickup, that gets capacitance to half. You gain an octave.

You also have half the output voltage. IOW, 6db less S/N. This means the traditional 12AX7 or TL072 preamp will seem hissy. Noise-design will be important.

You can go further with 1/3 or 1/4 the turns of fatter wire. The limit will be how you you can go with first-stage hiss.

Alternatively you can do as Les did, wind for 150 ohms and incidentally infinite bandwidth (100KHz?). However noise will be killer unless you use a step-up "mike" transformer to the first stage. Hi-Z transformer windings have the same problems as pickup windings, though to different degree.

> I suspect you may be selling this concept a bit short

I feel you have sold yourself a bit long. Underwound pickups and short cords exist. Buyers go for the "hot pups" and 30' cords. Onboard active buffering will significantly raise the resonance, and optionally flatten the bump. Sales are steady but very small. IMHO, 90+% of players like a restricted treble. There are good musical reasons for this. Steel strings in guitar length/gauge are inharmonic above a few KHz. Many fingers are sloppy and can't hit cleanly, or squeak. Drummer's cymbals are All Over the 6KHz-up range. Even if we wanted highs, the Guitar Speaker has a severe (20+db/8ve) hi-cut above 3KHz-6KHz. Tweeters have never been favored.

[HD Evans]

Wow, I had no idea Les was going for coils that light!  Even if the goal is a perfectly flat response through the human range of hearing, that just seems crazy from a noise perspective.  Noise is definitely a major concern with this design, since not only is S/N ratio taking a hit, SVF topology is on the noisy side as well.  It's entirely possible that this concept is crazy and won't accomplish anything useful, but either way I'll learn something.   I fully agree that most sales are for hot pickups, but at least to my ears I've found great results from what most people would consider underwound.  Do my 12" speakers roll off well below the hump of these pickups?  Quite possibly, but I can hear other things happening.  It's particularly useful for clean tones, and as the gain increases I favor pickups with much lower resonant peaks.  The sound is much more like that of a steel stringed acoustic, and the most obvious difference is in the harmonics particularly for open strings.  They are much less forgiving, similar again to acoustic guitars.  Thanks for your input.  So far nothing has convinced me that I'm completely crazy, or that this is by any means impossible.  But I don't think anyone can be entirely certain of how the end result will sound.

[PRR]

> that just seems crazy from a noise perspective.

Les had good mike preamps. In older studio technology, it was easier to find a mike input than anything suitable for guitar. Mike inputs are generally very near ideal noise figure.

> SVF topology is on the noisy side as well.

Gain first, then filter; attenuate into "normal" guitar-amp inputs or line-interface to studio gear.

> So far nothing has convinced me that I'm completely crazy

Yes, by the facts/ideas presented, the extent of your craziness is unproven and may be very mild. In fact you may need more "craziness" to push the idea to users. Buzz-words and graphics have more impact than theory.

[HD Evans]

Hey now, there's crazy and then there's marketing crazy...  Symptoms are identical to large doses of crack and have long term cognitive side effects!  Enough time is spent in my day job keeping marketing types in check, and I would imagine any personal attempt at such practices would be blatantly sarcastic...  I suppose that could be a good thing.

I wish you hadn't mentioned winding for microphone type impedance ranges.  Now I'm curious to see how I would fare using ultra low noise op-amps vs. matching transformers.  The good news is that the same circuit will work for both approaches.

Speaking of low-noise op-amps:  Does anyone have any particularly great recommendations?  I've found several in the $5 to $7 per dual package price range.  To be totally honest, money isn't much of an object since this isn't a commercial product... yet.  My only concern is doing it right by the best possible means.

[Ben N]

Speaking of low-noise op-amps:  Does anyone have any particularly great recommendations?  I've found several in the $5 to $7 per dual package price range.  To be totally honest, money isn't much of an object since this isn't a commercial product... yet.  My only concern is doing it right by the best possible means.

Even if this idea were to have commercial application, no doubt that figure would drop substantially when you are ordering multiples of 1000. Then again, don't forget the offsetting savings on magnet wire! But, seriously, this is the kind of system that would go into a relatively high-end, limited production guitar. $3-4 (less than the cost of strings) for a low-noise, high-headroom, high-fidelity preamp chip that makes the whole concept feasible should not be a deal-breaker. Of course if you need two of them...

[mhelin]

Over 120 days old topic but still interesting. How did it end up?  I would recommend using low-noise mic-pre IC like THAT1512 at least if you will build 150 ohms impedance pickup. Maybe you could go for using phantom powering as well. Put the VCF to external box which also doubles as power supply. Use some RF (BT or any other) interface for controlling the volume, Q, frequency and tone in guitar (http://www.ti.com/tool/ez430-rf2500 is one cheap solution).

[HD Evans]

I'll have some proper updates in the not too distant future.  Shortly after this thread the project ended up on the back burner for a while.  I founded a business with a longtime friend (why just get paid to consult for the big guitar manufacturers when I can do it myself?) and we now have a nice CNC system operating in our shop.  As we are preparing to launch a small product line pickups have now re-entered the picture.  I have designed a CNC coil winder and have been actively experimenting with a variety of low-z pickups.  Traditional construction and coil geometry have been emphasized, while awg selections varied to find the right combination of properties.  The filter design is where things are getting really fun.

Here's the biggest challenge:  Most players would be happy with a couple of tones to handle sparkling cleans and something with some punch to it.  For the sake of simplicity, all of the electronics will be on-board unless a special request is made for an external unit.  Having a set of knobs for gain, resonant frequency, and Q for each pickup is not at all feasible.  Instead, current plans call for LDRs to allow for huge tonal shifts with the flip of a switch.  There very well could be a simple solution, but we haven't arrived at one yet.  I will personally play with an external unit extensively, since all of those knobs will be okay on a big stompbox.  The weird part in that case is that I plan to use a midi cable, or something similar, to carry signals from each coil to the unit for shaping.  Your idea of wireless switching might not be out of the question...  Those are low enough power devices that it could be rather elegant.

If anyone has a great idea to control this system I would love to hear it.  Hopefully, I will have some sound samples in the not too distant future.

[PRR]

> idea to control this system

Anything more than a switch and one knob: iPad/Android has become THE controller.

[gritz]

MIDI does sound attractive from a cross-compatibility point of view (control your sounds from your DAW software...). TouchOSC via the aforementioned Apple whatnots is also a possibility. Still, it takes two hands to play a guitar (for most of us mere mortals) and that leaves both feet free...

Using the MIDI protocol would mean that users wouldn't be constrained by proprietary hardware and so might be less reticent about buying into your product than if there was a risk of built in obsolescence.

Vactrols are great for low distortion and zero zipper noise, but they do have transfer functions as bent as Nixon's nose, so resorting to tricks like running a servo vactrol in a feedback loop might be necessary. I'm not exactly sure what your plan involves now, but state variables might be better implemented with something like the Coolaudio V2164 chip - it boasts four fairly unfussy and low noise VCAs that have a log response to control voltage, so filters are a piece of cake.

Anyway, sounds like fun - and I suppose that amplifying low impedance pickups is not a mile away from handling dynamic microphone output.

[HD Evans]

As I see it, there will be a few players would want absolute control over every coil in their guitar who won't even blink at shelling out an extra $150 to $200 to do it.  For the remaining 90%+ (what? marketing guys sometimes conduct research, and then skew the results to meet their needs...  I'm skipping a step and ballparking it) I think including a bank of resistors that can be switched using jumpers could do the job.  With a manual explaining which jumpers to set to achieve common voicings I can remove all of the pots.  There might still be a better way, and I'm fine with tinkering to get it right.

Most of my business revolves around artists using 8-strings and some 7-strings.  The complaint I've been hearing to no end is that the guitars come loaded with active pickups, typically EMG, that are designed to create exactly one tone.  Not everyone playing these instruments wants a super-hot signal at all times, and finding a passable clean tone is virtually impossible.  The response is too dark, there's a big bump in the mid range, and the dynamic range is rather compressed (not so much on the SA, humbuckers are the main offenders.)  I'm not saying EMG make terrible products, I have a set on one of my guitars that I absolutely love, but they don't innovate.  They could have created what I'm working on now years ago, but instead they have effectively the same product line as they did 20 years ago. 

In simple terms it would be very safe to say that I want to have one system that is totally on-board, and switches between voicings at the flip of a switch...  And a highly-tweakable floor unit  covered in knobs for power users.  I know enough about electronics to be dangerous, but the survey course given to Mechanical Engineers is a bit lacking.  Growing up around a EE helped a bit too.  I would need to run it past my business partner:  but if anyone who is highly experienced, can assist with circuit design and implementation, knows the ins and outs of pc board layouts, and is interested in getting involved...  send me a PM.  I'm sure we can reach an agreeable arrangement, varying from sweat equity to direct compensation.  My background is heavily geared towards mass production, and we have endorsements in place based on reputation alone.  But we could definitely use an electronics guru. 

[Cliff Schecht]

So how are you currently implementing the preamp? I'm not asking you to post any proprietary information FWIW, just curious what IC/opamp you went with in your design.

[HD Evans]

When all is said and done IP will be protected as needed, thus far there is a lot of prior art... But the refined implementation of combined technologies is what stands the best chance of obtaining patents trademarks and copyrights.  Oddly enough, publicly posting proprietary plans/ideas is not only safe: it's a decent idea to prevent anyone from stealing your work (as long as you state your intention to protect your IP.  Since my intent has been stated, if another company beat me to market I have proof right here that my concept has been under development at least since my first post in this thread... So I'm happy to share what's going on, and absolutely encourage people to tinker with it as well.  I just don't want a manufacturer to start producing and selling anything based on my ideas.  To that end I haven't ruled out using some form of open license, allowing individuals to work with the system while preventing mass production by unlicensed parties.  Now, on to your question:

The current front runner is the ultra high-fidelity, low-noise, low-distortion, quad op amp: LME49740.  Each one is used to implement a state variable low pass filter with independent control of gain, resonant frequency, and Q.  They are adjusted only using potentiometers right now, although alternative methods will likely be used later on.  Especially the on-board system will need a simplified scheme, and there are dozens of options on the table at the moment.  At least one filter will be connected to each pickup.  One of the next big questions that needs to be addressed is whether or not humbuckers benefit from being equipped with one filter per coil... It might even just end up being offered as an option.  Such a simple version will at least feature a mixing stage to facilitate coil balance manipulation.  This is definitely a situation in which one test is worth a thousand expert opinions.  

For the current stage of prototyping I'm being lazy and using components from my spare parts bin.  At the moment that means another quad opamp: LM324.  A perfectly decent little opamp, and it facilitates laziness because it doesn't require a bipolar power supply... In fact, it runs just fine connected directly to a 9V battery.  I have a couple dozen NE5532's, but the LM324's will tell me what I need to know for now.  I'm working on a deal to get some PC boards that were originally made for a different project but has a layout that will accept the circuit I've worked up around the LME49740.

Also, I have a set of bobbins ready to wind two humbuckers and a single coil.  I've run all of the design calculations and plan to wind using 36 awg and 37 awg wire.  These will be the set of low-z pickups I need to carry out the rest of the R&D while verifying my designs.  

For the onboard system, I will most likely need assistance making a proper PC board unless I get really good at using automated tools... Actually, the same goes for the stomp box.  I know enough to be dangerous designing circuits, and know much more about control systems, mass production, and quality control.  Also, the onboard circuits will need to use surface mount components.  

It's getting fun.  Suggestions and ideas are welcome.  So far everything is on schedule to get this system into production this year, which will be amazing.

[artifus]

http://www.cycfi.com/projects/six-pack/