Initial thoughts about guitar pickups

Since their conception, there was not much progress in the development of guitar pickups. Of course, efforts have been made, and we now have various methods of construction, for instance, the Fishman Fluence use a PCB “winding core” or the Lace Alumatone which uses an entirely different paradigm. There are, however, a few questions left unanswered. What does actually matter for the players, the people who use them on a daily basis? We should, of course, start by saying that the preference of a certain type of sound is a matter of personal taste, after all, the golden rule in audio is that “If is sounds good, it is good.” But that should not keep us from trying to put some numbers and dip our toes into some statistical analysis of the preference in guitar pickups.

Coming from an academic background and having finished a PhD in Psychoacoustics, this feels like the perfect project for me. I would like to start looking into what elements actually matter for perceptually better sounding pickups. What matters and what can we ignore? What measures are important for defining a perceptually relevant subjective preference? Is there a particular frequency response that is universally considered good? Maybe something like the Harman curve that is used in headphone research? And finally, what kinds of physical features do we need to consider when we are talking about “shaping the tone” of a pickup? Could we for example have a computer model which takes a frequency response as an input and outputs the physical parameters needed to get to it? Like a recipe, wind 7569 turns onto a unicorn-charged magnetic core, and you will get the exact frequency response as you desire? Surely that can be done. We need to start moving away from the “hand-wound-under-the-clear-moonlight-mojo-magic” way of thinking and realise that it is all about the maths. Once we fully understand guitar pickups, we can build them to the exact specification that a player requires.

The aim of this research is to take out some of the guess work that is predominant in the guitar pickup industry. We have the technology, we have the knowledge, all we need to do now is to actually start doing it. Slowly and surely.

Winding

In order to even start attempting any research in this area, we need to be able to wind guitar pickups. ACCURATELY. If we wind 10 pickups they need to be identical, same properties, same type of scatter, same wire tension. Only then we can start tweaking things and do the proper science.

The pickup winder project is nearly complete and is shown here. This winder aims to accurately and consistently wind guitar pickups in the shortest amount of time.

Goals:

  • Consistency
  • Accuracy

Measuring

Physical measures

  • Dimensions of the core
  • Number of winds
  • Length of wire (may be calculated from the two above?)
  • Wire dimension/material
  • Insulation dimension/material
  • etc.

Electrical measures (LCR).

  • DC Resistance
  • Inductance
  • Capacitance

These measures can be acquired all at once, and the values should be recorded and stored for each pickup that is bulit. As such, a database of electrical and physical measures will be stored digitally for each serial number. All info about a pickup should be accessible through the serial number or a QR code etched on the pickup.

Frequency response

  • The most important measurement for visualising how a guitar pickup behaves across the frequency spectrum.
  • A standardised measurement method needs to be devised and tested.

Note! Initial testing with an exciter coil and Room Eq Wizard showed successful results. (e.g, frequency responses shown in the cover picture of this post.)

Impulse response

  • Can be used for convolution with “dry” guitar signal

The challenge in this situation is to obtain a “dry” guitar signal: If you think about it, the sound of a guitar is already originally recorded through a guitar pickup, and its frequency characteristics are already in the recorded sample. So we need to first record the impulse response of the pickup that recorded the test sample and then deconvolve it. This procedure should theoretically leave us with a guitar sample that has the original pickup sound “taken away”. This guitar sample can be then used for further convolution with different impulse responses taken from different pickups.

Measurement Goals:

  • Understand how the physical measures affect the frequency response.
    • i.e., an increase of 0.5 in the inductance may result in a 2 dB increase in the resonant peak (I don’t know yet, this is what we want to find out.). We need exact values.
  • Compile a database of different physical measures of the guitar pickups and how they relate to the various frequency responses. With enough data we will be able to create a statistical model which can be used for designing pickups which precisely match a given frequency response.
  • Create a database of pickup impulse responses. This will again help with creating precise listening tests. Furthermore, it creates the opportunity to virtually showcase each pickup that comes out of the production line. (Measure IR -> convolve with “dry” guitar sound -> let players listen to the actual pickups and choose the one that suits them best.)

UPDATE!! A Proof-of-Concept of the virtual pickup selector can be tested here. This is a very crude implementation, however, it shows that the concept can work in a browser window. This can be very helpful if a customer wants to try a few different pickup models and even different serial numbers from the same model, and then proceed to ordering a precise pickup (remember how we said that we measure an impulse response for each pickup that we build? It is here where that reveals its full potential.)

Testing

Psychoacoustic testing. We need to conduct subjective listening tests in order to observe how different physical parameters of the pickups are translated into different subjective attributes.