Friday 29 June 2012

Pickup Frequency Response Measurement. Humbucker, Single coil and Shuntbucker.


Well the shuntbucker has been in for a while now and I am still trying to get my head around how good it sounds. As it happened I did a pickup swap and quick setup for a PRS Tremonti SC (the cheap one) last week and was paid with the old pickups, which was more than I wanted. Anyway I now have a pair of ex PRS Korean generic humbuckers measuring about 10.1k DC resistance to play with.

I thought it would be the perfect test vehicle for a frequency response measurement comparing normal humbucker wiring, the normal shorting of the slug coil for single coil type response and the 47nF shuntbucker mod. First though I would need a stable and repeatable frequency measurement kit. It took a few swipes as you have to energise both the pickup coils with opposite magnetic polarity and it needed to work from low Hz to high kHz. Anyway to cut a long story short, I soon had a rig using a frequency generator, an OPA540 as the coil driver and an energiser coil. The output of the pickup was connected to a 100k resistive load and measured on an oscilloscope. This was a slight limitation as I would like to have measured at 500k but it’s the input impedance of the FET scope probe I used. To eliminate ambient noise the measurement was taken as an average of 100 consecutive measurements. The output in mV is plotted below. I have since reduced the drive amplitude as the output was higher than expected.

The blue plot is the normal humbucker. The red plot is with the slug coil shorted to ground which is normal practise for coil tapped pickups. In this configuration you acn see that the resonant peak has shifted to a higher frequency giving the expected single coil sound.  The green curve is the shuntbucker mod with a 47nF capacitor to ground across the slug coil. As this is directly injecting a magnetic field you won't see the inverse behaviour caused by fat strings producing a higher magnetic signature than say the thin top E string, which helps flatten all this out.


From this you can see that the 47nF worked as expected and has adopted the humbuckers low end frequency response. What is evident is that using a straight 47nF has produced some resonance with the coil which is why it appears to have a higher output compared with the humbucker configuration at 400Hz. You see the effect again at 800Hz as the kink in the otherwise smooth curve.  By adding a series resistance the resonance can be modified and a much cleaner transition from humbucker to single coil can be obtained. At the high end the shuntbucker follows pretty much the higher resonant peak of the normal single coil configuration. Hence a shuntbucker is like a single coil but with hairy balls.

 I think I will need to measure the spot frequency impedance of the humbucker and single coil configurations to work out the ideal shuntbucker capacitor series resistor. I could just try it out but I think the data will be useful. It could be I end up with two resistor/capacitor nets and possibly look at pulling the high frequency resonant peak about.

Monday 11 June 2012

Humbucker coil tap with passive shunt, or Shuntbucker.

I have been fitting the pickups to my guitar build and they came with five wires. A shield wire and green ground wire combined. The coil tap connection red and white together and black for hot. It occurred to me whilst wiring it all in that simply shorting out the bottom coil to ground to coil tap whilst giving a good single coil sound, was a bit simplistic and short sighted. There has to be some use for the output of the now shorted coil other then dropping it to ground.

The easiest mod would be to short it to ground with a capacitor. This way I should get the brightness from the top coil but with some added bottom end and a small amount of hum-bucking from the bottom coil. If you plot the impedance (AC resistance) of say a 47nF (0.047uF) and 100nF capacitor you will get the following impedance values.


If  I presume that the _average_ impedance of each coil would be say 3k then you can see that with the 100nF capacitor (red plot) half of the output of the shunted coil will be lost via the capacitor at 500Hz.  Without some complicated measuring I can't be sure exactly what the impedance of the coil will be at a particular frequency. I do know how the capacitor works though and it had to be worth a go. I went with 47nF (blue plot) in the end and am very happy with the effect on the bridge humbucker. Below is the diagram showing the two coils from a humbucker and how the capacitor shunts the bottom coil.
Shuntbucker with frequency dependent shorting of unused coil.

It has a distinct change when tapped as you would expect but the pickup has a lot more balls, more P90 than strat sort of sound. For the bridge pickup this is ideal for me as I never found a use for a single coil bridge position. After trawling the net I did find that this is not a new idea, coil cut was a term I found amongs some others. Considering the simplicity of the modification, and the potential in using more complicated passive networks I am a little lost as to why the vast majority of sites will just short out the one coil and concentrate on additional phase and series parallel switching.

I think the major pickup manufacturers have also not actively brought to the fore this simple modification or even re-voiced there own pickups with a passive shunt so that when taped you get something better sounding. Combining a capacitor, inductor and resistor network across the lower coil could tailor curves into the pickup and tailor the response beyond wire gauge, turns and ad-hoc parasitics. It seems a vastly unexplored path, maybe it needs a name people can identify with, so maybe coining the term Shuntbucker would help it stick in the conscience.

EDIT:


I have found that as per the follow on article “Pickup Frequency Response Measurement.  Humbucker, Single coil and Shuntbucker” this really needs a series resistor with the capacitor to make this work. The issue is that the capacitor and coil inductance are too resonant and the transition results in the mid-range being swamped.  I have found that a 47nF and 3k3 resistor across the bottom coil works well for me.
 



HSG Mod front end design thoughts.


The HSG mod has a slightly different approach to the real valve gain setting and positioning of the voicing RC network attenuator compared to the original Soldano design.  Though not the final design this is a dialogue covering the thoughts I had in approaching the modification.  I will concentrate here primarily on the input stage and what I attempted to do to make more sense of the setting.  All values are from SPICE simulation using Microcap 9 evaluation edition from Spectrum software.  This is a free download and contains the 12AX7 / ECC83 models as standard from Duncan Munroe. 

The standard stage is set up pretty much from the off as a high gain amplifier. The valve is operated at a high gain and then droped with a passive resistor network with 1Meg as the gain pot This generates a lot of resistor noise and susceptibility to interference being a high impedance point then amplified by the rest of the preamp. The gain of the input valve is a combination of the anode resistor (220K) and the load resistor (1Meg pot + series resistor) over the cathode resistance.

 For the Soldano design below, with a 100mV input at the point marked VG1 the voltage at the V1A anode is 6.26V pk-pk or a gain of 62.37. However at point A at the input of the second stage and after the 1Meg series resistor the overall gain achieved is now 27.6 with the gain pot set to 100%. The quandary is that whilst 62.7 is pushing the valve fairly hard the overall gain achieved is modest, and I still have an issue with all that impedance and resistor noise on the input of the second stage.


In changing the GAIN pot to 100k in lieu of 1Meg it will reduce the resistor noise into the second stage but lower the gain of V1A. I can re-jig the gain by shorting out the attenuator R53. With a 100k anode resistor the valve gain is now 35.6 and due to effectively removing  R53 it remains at 35.6 at point A.  There is no blocking at the point between V1A and V1B.

All this doesn’t come without cost, the AC frequency response curves for both circuits are quite different.  However the low frequency  is recoverable by increasing C27 from 1uF to 4.7uF. The final curves are shown below. The red is the standard circuit with 100% gain at point A, the blue is modified including C27 = 4.7uF. 
The Soldano circuit also has a low frequency response that flattens and extends way below bottom E and into hand-on-string and bump noise and I didn’t see that helping either.  The modified circuit has a much wider frequency plot, so I can easily tweak the circuit to lose or re-voice this. The original circuit is pushed so especially at the high frequency end it is what it is.
 I worry that I have missed something here as I seem to find 1 meg-ohm as the default go-to value for the gain pot, and to me that doesn’t make sense.  

I should note that the dynamic range of the modified circuit is reduced because of the bias being unmoved but it will still take a 2.5V pk-pk signal before clipping the first stage which should be enough. If your humbucker is hotter than this then you are probably after clipping anyway. The important thing is that the modification can be tried without removing parts from the PCB and just placing parts or wires in parallel with the existing circuit. By reducing the gain of the valve I also hoped to remove some of the valve to valve dependency that other uses have reported. In the original design the high frequency end is dominated by the valve and not by circuit values.