Hi diy-comrades,
I've more or less read the thread and just have built my second driver. I took your advice and used 0.2 mm copper wire. It's a humbucker with 2 blades.
Dimensions: 1.4 x 16 x 57 mm.
Coil height: 5 mm
Windings: 2 x 60 turns
Resistance: 8,4 Ohms
My first driver was a single coil design and terribly sensitive to magnetic feedback (pigscreamer). The new driver seems to be able to do the job. I get string feedback without squeal. I've only done some outboard testing with the new one, but I have good hope it will work when it's properly installed.
Col you seem to want a more systematic, scientific approach
Electrical properties of the sustainer/driver
There is an easy way to measure the inductance of the driver. You need (1) a sine wave generator with a low impedance output (mine is 50 ohms) and a constant voltage output level over it's frequency range (2) an oscilloscope (3) a multimeter that measures frequency. My oscilloscope is a very old handyprobe 1, so nothing special. This setup also comes of handy when biasing a fet or setting up voltage levels needed for your opamp(s).
Create a notch filter.
For that purpose I connect in series: 180 ohms resistor, 680 nFarad capacitor and the driver. Connect this chain to the sine wave generator: resistor to signal, unused driver connection to earth.
Now probe the signal in between resistor and capacitor. Now you need to walk through the frequencies to find the resonance frequency where the voltage drops the most. Then measure the frequency (signal of sine wave generator, not in between resistor and capacitor where because of the voltage drop the measuring can be problematic).
Calculate inductance
L = 1/(4* pi^2 * F_resonance^2 * C)
In my case it measured 6000 Hz:
L = 1/(4* pi^2 * 6000^2 * 680e-9 ) = 0,001 H = 1 mHenry
The highest fundamental is about 1100 Hz.
X_C = 2 * pi * f * L = 2 * pi * 1100* 0,001 = 6,9 Ohm
Now we know the impedance at 1100 Hz. Real component R = 8,4 ohm, imaginary component X_L = 6,9 ohm
Z_1100Hz = (8,4 ^2 + 6,9 ^2 ) ^0,5 = 10,9 ohm
There are 2 reasons why at 1100 Hz the driver works less efficiently:
(1) higher resistance (10,9 instead of 8,4), lower current (I = V / Z), lower magnetic flux
(2) phase shift between voltage and current
arctan(X_L/R)= arctan(6,9/8,4) = 39,4 degrees (0 is in phase, 180 is out of phase)
Note that 90 degrees is the maximum phaseshift. At 90 degrees phase shift it's all conductance, the driver works as an ideal conductor. The driver won't dissipate energy, it stores energy and then spits it back at the lm386 (in case of the f/r. Therefore a zobel network is used to keep the impedance low at high frequencies and prevent instability.
The following is my DIY-theory. Maybe it's incorrect and completely useless, but my goal is to produce some kind of benchmark for drivers so we can compare one driver to another. Maybe there are better ways to express the quality of the driver. As long as there is a way of measuring it, otherwise i don't believe it.
It would be nice if I could combine (1) and (2) in one single number, a compensation factor for high frequencies. I will use the term specificaly for 1100 hz.
(1) 8,4 / 10,9 = 0,77
(2) use cosinus function normaly used for calculation of power: P = I * V * cos(phase shift between I and V)
In order for the sustainer to work the string should be helped. So when the string moves down the driver should pull and when the string goes up the driver should push. Just like a swing the timing when to push and when to pull is important. Phase shift, pushing too late or too soon, reduces efficiency.
cos(39,4 degrees)=0,77 (!) (1) and (2) are the same.
So the compensation factor should be 0,77^2 = 0,60. If there is a linear relation between current and magnetic flux then we could use this factor instead of phase shift and impedances. The magnetic flux at 1100 hz is 0,60 times the magnetic flux at let's say 50 hz.
So in order to get an even response the questions are:
(1) is a compensation factor of 0,60 enough?
(2) can we construct drivers with a higher compensationfactor (closer to 1). Maybe some of you have constructed a much more efficient driver, I only know what I have.
If this compensation says something essential about the driver's quality maybe we can draw conclusions about the rest of the circuitry. If the driver doesn't get a good benchmark but is sustaining flawlessly then the circuitry must somehow have compensated the driver's weaknesses (eq, more power, agc). And if a good driver doesn't sustain, we should suspect the circuitry and try to improve it.
Sorry for being long-winded, it must be the thread that makes me do it. Also, this contribution of mine seems to be a bit out of sync with what is being discussed at the moment. But it would be nice if every s******er builder has a way way to measure and calculate the inductance of his driver. Even more so if he shares his results with the other builders.