A six coil pickup and on board electronics for individual string distortion.

[Mike Sulzer]

The purpose of this project was to make an electric guitar pickup system with distortion on individual strings, that is, with low intermod distortion on multiple strings. A constraint is that the guitar must work with a standard cable and use standard amplifiers. This is accomplished by making a pickup with six individual coils and on board electronics for generating the distortion, combining the outputs, and filtering the combined signal.

A coil like the ones in the pickup is shown here: http://www.naic.edu/~sulzer/coilPrototype.jpg. The coil is longer than it appears due to the shortening from the viewing angle. The core is made from two ferrite beads, each .2 by .437 inches. It has 5000 turns of #43 wire. The inductance of the coil is about .5 Henries. The material is so-called #73 with a relative initial permeability of 2500. (I might have mistakenly said this was 5000 in an earlier reference to this project.) The magnet is 1/8 inch in diameter and 1/32 inch thick neodymium. The field strength provided by this magnet backed by this core, 3/32 above the magnet, is larger than that of a typical humbucker, but less than a typical ajnico single coil.

These coils can be used in a variety of ways. For example, six connected in alternating polarity in series makes a "single coil type" bridge pickup with high hum cancelation. The resonance is at about 5KHz with a normal guitar cable. The ferrite is lower loss than alnico, and so the pickup must be loaded with a lower total resistance than a standard single coil. The pickup delivers more than 1.5 volts p-to-p with a six string chord.

As an individual string pu, loaded as explained later, the output is about 0.2 volts p-to-p. An earlier version of this project used a seventh coil to provide hum cancelation for each of the six coils. This is unnecessary; reversing the polarity does the job, even though the effective gains of the non-linear circuits can be very different. (All that really matters is that there be good cancelation when there is no, or very little signal present.)

Also in an earlier version, a cancelation network was used to get the signal leakage between adjacent coils down from the inherent 20 db to 30 db. This is also not necessary.

The six coils are simply glued into a strat pu cover. This is shown here: http://www.naic.edu/~sulzer/6coilPUonG.jpg. This system is only useful for a bridge pickup (string bending). It should be possible to make a 12 coil system for a neck pu. (But that is another project.)

In later posts, I will discuss the pickup some more and describe the electronics. The circuit diagram is shown here: http://www.naic.edu/~sulzer/sixCoilPre.png

[David Schwab]

Great project Mike!

I've been wanting to make a hex fuzz ever since I played an ARP Avatar guitar synth back in the 70's.

A buddy of mine wants me to make him a pickup like this for using with six Pro Co RAT fuzzes.

That's a nice looking guitar, what is it?

[Magnut]

Very cool project. I would be interested in hearing it. I had worked on a similar project and really enjoyed the sound.

That looks like a Black Korina body guitar. Nice tone wood! I have one too.

[Mike Sulzer]

Thanks to you both! For this project I bought an unfinished body and neck from Warmoth. (I am a slow wood worker; I would still be carving!) The body is their VIP model, black Korina, yes, chambered body with a black Korina top as well. The neck is mahog. This body style has a large control cavity; so there is plenty of room for fancy electronics.

[David Schwab]

I have a white limba (korina) neck through fretless bass. It's a nice sounding wood. The black variety is nice looking wood. Apparently it's all from the same tree.

[black_labb]

i'd be tempted to have a outboard box for the distortion controls. having a control per string could be difficult to fit on the guitar, or are you planning to have the distortion all preset without a difference between the individual strings, but keeping the distortion from crossing over.

i'm interested in this project!

[Mike Sulzer]

i'd be tempted to have a outboard box for the distortion controls. having a control per string could be difficult to fit on the guitar, or are you planning to have the distortion all preset without a difference between the individual strings, but keeping the distortion from crossing over.

i'm interested in this project!

The second: no individual distortion controls. Strings 1 and 2 have a different gain from strings 3-6, but this is built into the electronics. The gain difference was chosen to get the sound I wanted. It is both a gain and frequency response difference.

(That schematic I posted has one error that I know of; there is a 33K resistor in series with the distortion control pot. Other than that, what I am playing is what is on the schematic as far as I know. I expect that there are a few more errors!)

The electronics is not so simple; I am going to try to get the first part of the explanation done today.

[Mike Sulzer]

The electronics module in the guitar is shown here: http://www.naic.edu/~sulzer/sixCoilEmod.jpg. The selector switch is not on the main aluminum sheet so that it can be unscrewed and taken out the back while the rest stays in place. This allows resistors mounted on the switch to be changed easily for tweaking the design.

In general, frequency compensation is needed both before and after distortion generating stages. It seemed to me at the beginning of this this project, that it should only be necessary to compensate afterwards when generating the distortion on each string individually. The reason, it would seem, is that with a single string one can only generate harmonics (not cross modulation products), and it should be possible to set the best ratio of harmonics with compensation only afterwards. My ears have told me something different.

There might be several explanations, but the one that appeals to me, and seems to agree with what I hear, is that there really is cross modulation, but not what you might expect. All strings are anharmonic; light gauge guitar strings are more so that most. The harmonics generated by the distortion from the fundamental are locked to multiples of that frequency. Since the strings are anharmonic, the higher harmonics in particular are significantly shifted from those frequencies. Thus harmonics generated by distortion from the fundamental and first few harmonics (which are not much shifted from the fundamental) are not at the same frequency as the higher natural harmonics. This produces ugly beat frequencies, or at least that is how I hear it.

The solution is simple: filter out the high frequencies, before distortion. But do not take out too much, or the sound is not "alive". Since this must be done for each string individually, there is a premium on doing it in a simple way. Therefore this circuit uses the pickup inductance as the frequency selective element, together with a resistor. The op amps are used in the inverting mode so that the resistor that loads the inductor is one that would be required in the circuit in any case.

The circuit has also be set up so that strings 1 and 2 have more gain than the others at low frequencies, but also the gain starts rolling off at a lower frequency. This is to compensate (at least as I hear it) for a thinner sound in those two strings.

[Possum]

...

Velvet Hammer made such a pickup called the "Stereo." But it took half the poles in an even/odd selection out to 2 outputs. I've never seen one, Velvet Hammer is back in business but not sure if they are making those particular ones.

[Mike Sulzer]

Velvet Hammer made such a pickup called the "Stereo." But it took half the poles in an even/odd selection out to 2 outputs. I've never seen one, Velvet Hammer is back in business but not sure if they are making those particular ones.

Then they likely had no hum cancellation.

[Mike Sulzer]

Velvet Hammer made such a pickup called the "Stereo." But it took half the poles in an even/odd selection out to 2 outputs. I've never seen one, Velvet Hammer is back in business but not sure if they are making those particular ones.

Then they likely had no hum cancellation.

[Mike Sulzer]

Here are some requirements for the distortion circuit:
1. The level of distortion must be voltage controlled. This means that a large number of distortion circuits can be controlled together, either by a pot. or by a switch. The actual circuit uses both: five levels from clean to very distorted are selected on a switch. A pot allows finer levels of adjustment, and also allows compensation for changes in the battery voltage with use or age. This avoids the need for a regulated supply. One biasing circuit feeds all six distortion circuits.

2. The circuit must have a gradual onset of distortion rather than abrupt clipping, at least in the intermediate positions of the switch. One could design for symmetrical distortion or not, or perhaps making it selectable. This circuit is symmetrical in order to keep it simple. It is thus roughly analogous to the distortion of a push-pull output stage rather than that of a single-ended triode.

3. Since we need six distorters in a small space, it must use few parts. The goal was one op amp, two diodes, and a few resistors and capacitors.

The circuit (http://www.naic.edu/~sulzer/sixCoilPre.png) is a variation of the standard diode clipping circuit. The diodes can be forward biased using voltages selectable by switch and pot. When the forward bias at its maximum, the dynamic resistance of the diodes is low, and so the gain of the op amp circuit is low. The ac signal from the guitar string is low compared to the dc bias current, and so the circuit is very close to linear. As the bias current is reduced, using the switch and or pot., the ac signal increases due to the increasing gain from the increasing diode dynamic resistance. The ac signal increases relative to the decreased dc current, and so the circuit becomes non-linear. How non-linear depends on the bias current. When this current is reduced close to zero, the circuit behaves very much like a standard clipper. However, the biasing resistors have a component in series with the diodes, and so the circuit never clips completely flat.

The waveforms for the first four switch positions are shown here: http://www.naic.edu/~sulzer/sixCoilWaveforms.png. The input is a 200 mv p-to-p sine wave.

The eight op amps used in the circuit are two LMV654 chips. These are a new generation of low voltage, low current chips with good characteristics for audio use. They come only in the tiny TSSOP14 case; I had them soldered onto adapter cards (very expensive). They are very linear right up to very near the power supply rails, and so the five volt maximum limit is not a problem for use in a guitar. I use three AA alkaline cells as a power supply; these cells are inexpensive, and they are the cockroach of the battery world, available in quantity wherever there are people. A set should last nearly 1000 hours.

[bbsailor]

A variation on your design

Mike,

Your hex pickup is very impressive. I will suggest another way to feed the individual string channels.

Consider that each string to be the dynamic element such as in a ribbon microphone. You can do a quick experiment by measuring the impedance of each individual string. They vary from a fraction or an ohm to over 1 ohm depending on its diameter for solid strings and the core size for wound strings.
Just use a miniature ourput transformer across a string with a 4 or 8 ohm side and a 20K to 100K side for a high turns ratio. Alligator clips will work for a quick experiment on one string. Generally, you want the transformer input impedance to be 10 times the individual string impedances for the maximum voltage transfer. By using an input transformer with a cente tap, you can optimize the balance between wound and solid strings a little better. You can even use a 70V speaker line matching transformer to try this out.

The laws of physics dictate that the amount of voltage induced into the string is higher when the length of the magnetic field under the string is increased. I use a .5" wide X .125" thick X 7" long rubberized/flexible magnets under the strings from the neck to the bridge. This ensures that a full range of harmonics are being induced into the string.

With this type of pickup, each string is fully isolated from each other. On one of my early prototypes, that I made about 25 years ago, I took an old Applause guitar, removed the fingerboard and placed a copper strip over the truss rod and extended the end into the guitar body in the heel of the neck as a ground return. I used a brass nut and soldered the copper strip to make the connection behind the brass nut.

Each string had the common ground at the nut end of the string and the six hot connections behind the bridge. The Applause guitar had the strring ball ends secured behind the bridge rather than inside the body. I just put a copper rivet in the bridge for each string to go through and soldered a wire to each rivet. I routed a hot wire from each rivet to the low impedance side of an 8 ohm to 50K ohm transformer. I enclosed all six transformers in coper foil and mounted it on the heel of the neck. By alternating the phase of each alternate string I was able to minimize hum similar to a humbucking pickup. I routed a common ground from each transformer and one hot connnection from each transformer to an eight pin microphone connector (using only 7 pins, one for ground and one hot for each string). I made breakout cable with six 1/4" plugs that I fed into a six channel Radio Shack Mixer. I had individual string level, EQ, and L-R panning control. At that time I mounted a large ceramic magnet on a wood block that I suspended in the sound hole. But, longer magnets, putting more string length in a magnetic field, work better as it produces a higher output and produces a greater amount of higher harmonics. Just put the long magnet closer to the string near the bridge where the string motion is less. I believe you will get an invividual string output very close to your reported 0.2V P-P output.

So what is the point to all of this? This design produced a wide bandwidth pickup that sounded more acoustic than electric with a greater amount of higher harmonics, as I did not loose all the upper harmonics beyond the resonant point as would happen in high impedance pickups. In your case, I believe you mentioned that each string coil has a 5KHz resonant point.

Since you are going to the effort to add active EQ and distortion in your design, I thought that you might want to see what happens when you have more upper harmonics on each string to play with. You can even produce many common electric guitar sounds (both single coil and humbuching) by active EQ shaping.

Since it only takes two alligator clip wires and a junk box transformer, I thought you might want to quickly try this. I believe this method has some potential that might contribute to the next generation of guitar pickups.

If you try it, let us know what happens?

Joseph Rogowski

PS. Here is an additional thought. I even had success by using a 3000 turn, low frequency torroid current transformer by wrapping a different amount of low impedance turns through the torroid ring from each string. This is a simple way to make combined output with a single transformer.

[Mark Hammer]

Maybe it was already mentioned, but there are two distinct kinds of hex pickups. One is the kind like what was used in the Ripley guitar, or the Bartolini type, where the intent is to provide for a more interesting stereo image. here, the coils don't have to sense ONLY one string, but simply have to sense that string more than the others.

The other type of hex pickup is that used for guitar synthesis, where the purpose of the individual coils is to make sure that the sensing circuitry receives no information from "irrelevant" strings, such that it can generate a trigger, gate, and proportional control voltage for only the note being played on that string at that time.

The second kind imposes more stringent constraints than the first.

[Mike Sulzer]

Joseph,

That is a great idea, using the string as the conductor, that is, one side of the loop around which one integrates to determine the induced voltage. I like the hum canceling, too. The loop for hum pickup is bounded on the top by the string and then follows the return path for the current. Thus it is quite large, but most of it goes away completely with your hum cancellation because the return path is common to all strings. (That is, B dot da cancels when the six loops are considered with the sign inversions.)

It would be interesting to try a sequence of small neo magnets along the string. One could probably get the limiting field (that is, limited by "string pull") without too many.

The 5 KHz resonant frequency for my coils is for using them in series to make a "single coil" pickup with a cable of about 500 pf. The inductance for such a pickup is about 3 H, very much what one would expect. When these coils are used individually right into a preamp (much lower capacitance than a cable), the resonance is much higher than 20 KHz; so one can have flat response in the audio range. However, what I wanted to do was to cut high frequencies by loading the coil.