|
Post by aquin43 on Sept 21, 2022 11:50:18 GMT -5
The ideal pickup model should as accurately as possible reproduce both the impedance and velocity transfer function of the pickup at least up to the open circuit resonant frequency. This requires that both the response to an exciter and the complex impedance be measured. In measuring the impedance, any extra loading on the pickup is to be avoided since it will lead to errors. Phase in particular is important. The method described by ms requires the simplest possible measuring fixture and provides a robust measurement. Phase is not usually measured at all in the frequency response Bode plot but it is desirable that the loading should be the least possible and should be known. The fitting method used in the Gitec Pickup Wizard program can produce a multi-coil circuit that closely approximates the impedance of a real pickup coil using usually two but sometimes one or three non-coupled inductors. This is a behavioural model, the actual coil is more complicated than that. It turns out, though, that by splitting the input exciter voltage in the correct proportion between the coils the response of the coil to the magnetic velocity signal from the string is well reproduced, at least for pickups without excessive losses in enclosing metalwork. Other pickups need a supplement to the frequency response and it seems that an extra single low pass step filter is sufficient even for a very lossy pickup. Here is an Armstrong "Johnny Smith" pickup modeled in this way with eight parameters. This is an extreme example of a pickup with a cover that deliberately attenuates the treble but which isn't tightly coupled to the coil. The brown line is the overall response, the red line the added step and the orange line the core response of the coil. Most pickups don't need anything like as much correction but it is encouraging that such an extreme example can be made to work with a single low pass step. The exciter voltage is applied to the coil sections in proportion to their inductances as fractions of the total inductance. The coils are not coupled so the total is simply the sum of the parts
The method of introducing the voltages uses a minimum of Spice components. There are other equivalent circuit configurations that are more amenable to non-Spice calculation.
|
|
|
Post by ms on Sept 23, 2022 6:57:26 GMT -5
Putting a part of the exciter voltage in series with each of the model coils is a good idea. It has the physical justification that each turn of the pickup coil has a small voltage induced around it that can be simplified to a single voltage source for for as many turns as are simply connected in series. I am wondering if the the exciter voltage should be divided by fraction of total inductance as you do, or by the square root of inductance, which would be closer to the relative number of turns.
|
|
|
Post by aquin43 on Sept 23, 2022 11:52:12 GMT -5
I am wondering if the the exciter voltage should be divided by fraction of total inductance as you do, or by the square root of inductance, which would be closer to the relative number of turns. I was following the suggestion in the Gitec Pickup Wizard where the signal is introduced as a 1/f current driven into the output terminals so that the voltages split in proportion to the inductance values. My circuit is a sort of Thevenin equivalent which removes the need for the 1/f variation of the drive.
I think that you are probably correct in suggesting using the turns or square root of inductance ratios. If, say, the string is considered as a very loosely coupled exciter coil, the mutual couplings and hence the voltages will depend on the square root of the inductances.
The change to the model is trivial to make and produces little change in the output in most cases. The extra step is still required.
|
|