Greetings all.
@Passinwind asked me to chime in, because buffering is my speciality.
For any coil wound pickup you to sound good, you want to have it a moderate resonance peak at 2-3 kHz. Unloaded, it has a peak
far above that, which is not considered musical or useful.
You want a capacitor in the nanofarad region to achieve the lower resonance frequency. Dependent on the kind of pickup (speak: dependent on
its inductance and wire resistance) you need a parallel resistor in the 250-500k region (250 typical for single coils, 500 for Humbuckers)
to bring down the resonance peak (Q) to a musical height.
In a practical vintage (passive) system, this role is taken over by the volume pots, and a "voicing" capacitor is not seldomly added (particularly
on basses) to bring the peak frequency down.
So this is what every pickup wants to see.
Scenario #1:
If you want to isolate the pickups so that they don't interfere with each other, you intuitively would want to buffer each of them.
You would add a "taming" resistor to each of them to taste and if too shrill, a small capacitor. I would start with 200 pF and work my way up until
it sounds ok.
The Tone pot does not interfere much if in the off state, and in the low resistance state it connects the capacitor that is wired to it in parallel to the alleged voicing capacitor, which brings the peak down even further. So choose wisely. The stock values are surely too big for that.
For me, the first thing that comes to mind for such a buffer is a j-fet buffer, which can run on very low current and is benign in its overload characteristics. With a suitably high input series resistor, this circuit is also RF intrusion proof, just in case anything escapes the filtering action the pickup itself provides.
However, you cannot just join the outputs (in this case, the source terminals). You need some kind of a mixer stage.
Sometimes you see designs where they just mix signals via two resistors, but this is a very crude method and inappropriate in the lights of the effort you made beforehand.
So you need a mixer stage. Any inverting op-amp connection can serve as a mixer. The node on the inverting (-) input of the OPA, where the feedback resistor and the input resistor meets, is (for all practical means) at a (near) zero potential (the so-called "virtual earth", which it is NOT!), so you can add a second input resistor to this node and mix signals without any interaction.
You can use a variable resistor (a volume pot in rheostat connection) for this purpose, which in conjunction with the feedback resistor, gives a negative gain, i.e. an attenuation. This is how many simple mixers are done. The resistor values can be kept in the 10kOhm region.
Scenario #2:
You may of course skip the discrete buffers entirely and wire them directly to the mixer, but if you do that, you would need to raise the input resistor values to a fixed value in the MOhm range (disregarding the volume control momentarily). From a mixer perspective, this is perfectly valid and workable, but you have to move the RF protection somewhere else, plus you have to move the volume control somewhere else.
You could for example use volume pots in a traditional fashion alongside with tone pots (which works well), and then run into the mixer.
The first scenario keeps noise low due to low resistance values, the latter may be a source of noise. This depends vastly on the OPA chosen. More on this later.
If you want to do turn on individual pickups, this is most easily accomplished by switching contacts of choice. The signals are mixed without interference. In the off state, the switch would short-circuit the signal to ground, such as at the volume pot center connection.
The pickups will always be in parallel this way. Connecting them in series would be very complicated. That said, with "pickup" I mean "a physical pickup complex", as for example a humbucker or single coil represents one. You can always locally wire such a humbucker in series/parallel/single with a separate switch.
Note that this will throw your choice of voicing cap entirely, unless you made individual voicing caps for each passive setting.
Indeed, a passive parallel connection usually benefits form a voicing cap (and even more so a single coil), but this would be counterproductive in series setting.
You need to think about the OPA too. You may use a quad OPA to cover all the buffering and mixing, but you would need to be very specific in
finding an OPA that is low noise, low current, rail-to-rail, and potentially split-supply ready. You also want to know the overload recovery characteristics. Those demands are often technically contradicting.
j-fet and bipolar ones differ in the input noise characteristics vastly. The ones have current noise, the others voltage noise.
Frankly, I do not know what you want to gain from all that.
I have a bass guitar that has two humbuckers, and I interconnect the coils ad lib. The bass has a buffering amp on board, and there is *no* apparent degrading of tone. If I activate the
active/passive switch, the active always wins in terms of pristine signal.
For my Strats, I have made a crude j-fet buffer, which resides in a small enclosure with a 20cm cable on it. This is the shortest connection you can
make with an external box. The long cable to the amp and effects plugs into a connector on the box.
This is a HUGE improvement over the unbuffered version, and I never felt the urge for an improvement other than the fact of the somewhat clumsy external box.
IMHO you would be much better off with that. Using a buffer you always have a consistent and predictable signal, and if it all of a sudden appears shrill, then you just regained the treble you otherwise have lost on your cable. This can be most easily be corrected with a small move on the tone pot, or better, with a voicing cap inside.
Hope that helps.
