"Cable Length Knob" - Negative Capacitance?

[ashcat_lt]

So we've been having some discussion on GuitarNutz and something came up that I thought might be of interest over here.  If it actually works, I think it could be useful in a number of places in our stompboxes, and I'm kind of hoping some of the smarter folks over here can help settle the argument between the smarter folks over there, or at least add some info.

So, the idea is that we can create the effect of a variable capacitance to ground by driving the "ground" end of the cap with a copy of the input signal.  Depending on the gain applied to that signal, we can even end up with what looks like negative capacitance, which can be put in parallel with some other capacitance to ground and actually reduce the effective total capacitance.  The context there is to counteract/negate the effects of cable capacitance on the passive pickup signal from a guitar.  Whether it's installed at the guitar end, or in the cable itself, or even as the first stage of guitar pedal, it allows you to have a knob to alter the effective "length" of the cable.

This gives a pretty concise description of the theory:

And this is the results of a 5Spice  simulation:

Note that the stuff in the blue box on the left is the pickup and volume control in the guitar.  No tone, to keep things simple.  In the box on the right is the cable capacitance and the impedance of the amp or pedal to which it's connected.  That capacitor can just as easily go on the other side, though, and it works the same.

So, what do you all think?  Can it work?  Is it useful to us?

[DDD]

The idea is good as it is.
It works good, and one can find verified schematics in the link below (please look through the article up to the end).

http://www.sugardas.lt/~igoramps/article51.htm

[R.G.]

This gives a pretty concise description of the theory:
[...]
So, what do you all think?  Can it work?  Is it useful to us?

Interesting question.

Here's another interesting question. Figure A shows a ground symbol. I don't see any ground symbols in B through E. So the voltage in at "V0" is relative to ... what...?

There's always a reference point somewhere. Where is it on B - E?

Remember that in electronics, there can be no functioning monodes.

[ashcat_lt]

Well, those are kind of just conceptual drawings.  I think the actual schematic in the spice sim has plenty of grounds. 

[merlinb]

It's not negative capacitance, BTW. The current still flows from the point of higher potential (the opamp output).

[johnh]

Hello! Just thought Id call in here, to join this discussion. Those are the sketches and sim that i did on GN2.

Yes, there's no actual  negative capacitor, just a normal one being charged in the opposite direction.  But the concept of the circuit acting like a 'negative capacitor' is much more appealing and seems like a fair description of how it interacts with the rest of the system.  Maybe it could catch on in popular imagination, much like the 'flux' kind?

[merlinb]

Maybe it could catch on in popular imagination, much like the 'flux' kind?

The same principle is used in some phono preamplifiers where it is called 'load synthesis'. 'Synthetic capacitor'? 

[DrAlx]

Interesting, but is there not a way of cancelling out cable capacitance without active components ?
I'm thinking along the lines of how the trimmer on a scope probe cancels out cable capacitance.

The load in your example has (1M || 1nF) between its input and ground.

Lets say you stick (100 R ||  10u) in series with the circuit output.

At DC, the load sees a resistive voltage divider with values 100 and 1M.    (The ratio of R values is 10000).
As high frequencies, the load sees a capacitive voltage divider with values 1n and 10uF.  (The ratio of C values is 10000).
So overall frequency response is flat.  You can still tune stuff (make the 100R variable).
Or am I missing something?

[amptramp]

There were some high-impedance measuring probes that used a double-shielded cable where the outer shield was ground and the inner shield is driven by a shield driver.  The idea is that since the capacitance of the triaxial wire core to the inner shield sees the same voltage on both capacitance terminals, this capacitance appears to disappear.  For high impedance meters, the leakage current also disappears since the voltage is the same on the core and inner shield.  For this to work, you need the shield driver to be on the input end, meaning an active device in the guitar.

[Perfboard Patcher]

As I stated in GNutz thread   

My take:

The innovation seems to be an extra (middle-) shield in between the core (signal) and outer shield (ground).

The middle shield is connected to the opamp's output. C2 is the cable capacity between core and middle shield.
C3 is connected incorrectly, the cable capacity between middle and outer shield is located between the opamp's output and ground.

I reported as well that I built the circuit and that it doesn't perform as the spice model predicts, but for some reason they don't believe me.

[TejfolvonDanone]

First of all the cable has two wires: signal and common. THe capacitance of the cable is between these two wires. If you plug the wire into the input of a regular pedal you will ground the common wire. So if the output of the gain block in diagram B to E are ground referenced you just ground the output of the gain block.
So in the schematic you have to connect the lower end of C2 to ground.

Also i'd like to add that a cable has a small series resistance which is maybe negligible but we should keep it in mind that it is there.

If you think about it:
In the small diagram D you have Ic = j*omega*C * (Vo-2Vo).
((It's only true for sinusoidal signals.))Which is the same as if you've put out -Vo in the first place. Flipping the phase will solve the capacitance issue? It isn't that easy.

[ashcat_lt]

As I stated in GNutz thread   

I reported as well that I built the circuit and that it doesn't perform as the spice model predicts, but for some reason they don't believe me.

I certainly appreciate your input.  I mean, you're the only one of us who bothered to take the time to try to build it.  It's not so much that I don't believe you, but:

1) you built it on a breadboard, which introduces its own set of concerns, and at least in my experience makes all results just a little questionable,
1a) you're still talking about that three wire system, so I'm not sure you built what John drew,

B) best I can tell, your test procedure involved strumming a guitar and listening, which makes the results at least partly subjective.  In general, it doesn't seem to be a particularly well controlled experiment, though I guess it's closer to the actual practical application, and empirical evidence is at least as important as theory,

III) you reported that you seemed to be hearing more of higher frequencies (what I expect the thing to do) and then proclaimed it didn't work.  I kinda can't get my head around that.

Plus, I really kind of want it to work. 


@TejfolvonDanone - I'm not sure you're actually understanding the scheme that John drew.  It shows all of the grounds.  Those are all already connected, by the cable or the chassis or whatever.  OK, it doesn't show the power supply, and things might have change a bit on a single-ended supply.  Take a look at the page DDD posted.  It's in some Cyrillic language, but just scroll down a bit and you can see an example very much like this thing and how it connects to the input stage of an amp or pedal.  This does the same thing with the cable coming into it just like any other amp or pedal, then the circuit.

As I mentioned over on the GN2 thread, it's not that far off from how the classic wah pedals "fake" a variable capacitance.

[TejfolvonDanone]

@TejfolvonDanone - I'm not sure you're actually understanding the scheme that John drew.  It shows all of the grounds.  Those are all already connected, by the cable or the chassis or whatever.  OK, it doesn't show the power supply, and things might have change a bit on a single-ended supply.  Take a look at the page DDD posted.  It's in some Cyrillic language, but just scroll down a bit and you can see an example very much like this thing and how it connects to the input stage of an amp or pedal.  This does the same thing with the cable coming into it just like any other amp or pedal, then the circuit.

1) If in the posted schem C2 represents the cable capacitance it's connected bad because the the lower end of the cap should go to ground. If C3 represents it then the schem isn't what is in the diagram.
2) My statement regarding the idea that it's just flipping the phase still stands.
3) In the link all the schems have a part where IN goes to an amp section which is in fact shorted from its input to its output. For example in pic 8 there's a complementary pair FET stage with the "Alter" knob. C6 its input is connected to its output C8. So it raises a lot of questions about their validity. I don't understand anything in that article because i couldn't even tell what language it is in but the schematics provided won't work as they are or as they are intended. Also they have one input and one output which makes me question that they are working on the same principal as your original diagrams (i maybe wrong on this but i have no proof to support either statements).

I don't want to sound like i know everything and whatever i say is right but nothing you provided stands as a proof. I'm just curios about who is right and for the sole purpose to know whether it works or not.

[johnh]

In the hand sketches, it only shows the action of the amp section and 'C', which represents C3 in sketch A (no opamp), and C2 in B to E. the sketches are not a full circuit, and all signals would be referenced to ground

In the sim, C3 is the cable capacitance, between the cable hot wire and ground, as expected. The sim captures the entire system to a level that is usually good enough for most guitar circuits, but in this case may be missing something as it approaches the critical balance point between caps.

Hi PfBP, I really want to thank you for trying the circuit. Your results had too many clues that it was working! eg " The sound gets brighter when increasing the gain but it doesn't sound as a shift of the resonance peak but more like boosting the ultra highs", but I certainly accept that as you increase gain, it will deviate from the sim due to effects not captured by the model.

[FiveseveN]

For what it's worth, there's an English(-ish) version of the S.A.G.E. article: http://www.sugardas.lt/~igoramps/article57.htm
The variable capacitance part seems to work in SPICE, going to have to try it some time.

[amptramp]

Figure 29 of this app note is what you want:

http://www.ti.com/lit/an/snoa725a/snoa725a.pdf

It shows an LH0033 being used as a shield driver.

[R.G.]

Simulators are *very* sensitive to what you call ground, and where the actual references are.

So is Mother Nature, although perhaps not in exactly the same way, as simulators by their nature are a simplification of the real world to make it manageable on computers.

[PRR]

> the trimmer on a scope probe cancels out cable capacitance.

No, it does not. Find an explanation.

The naked 'scope is typically 1Meg||30pFd.

The 10X probe changes this to 10Meg||30pFd. Less capacitance, but not "cancelled" capacitance. The scaling is same-as the reduction of signal (10X probes no good for teeny-tiny signals) and the increase of resistance.

You don't want 10X reduction of your precious weak geetar signal just to get a little more highs through. The new-found highs will be buried in universal hiss.

> driving the "ground" end of the cap with a copy of the input ..... .... Can it work?  Is it useful to us?

Well, yeah, and all the devilish details when you attempt to cancel stuff in a complex poorly-defined real circuit.

It is NOT "passive". That amplifier stage is smacking at your signal as it goes by. How is this better than just sticking a buffer on the front? (OK, maybe you don't need power AT the source if you cap-cancel at the far end.)

Ron gives a perhaps "better" plan. Core, shield, shield. Middle layer is driven to follow signal. The cable is a little tough to find, especially in sexy colors, and probably not made curly. But such driven shields are routine in several types of signal acquisition work.

[DrAlx]

> the trimmer on a scope probe cancels out cable capacitance.

No, it does not. Find an explanation.

The naked 'scope is typically 1Meg||30pFd.

The 10X probe changes this to 10Meg||30pFd. Less capacitance, but not "cancelled" capacitance. The scaling is same-as the reduction of signal (10X probes no good for teeny-tiny signals) and the increase of resistance.

You don't want 10X reduction of your precious weak geetar signal just to get a little more highs through. The new-found highs will be buried in universal hiss.

I'm not suggesting weakening the signal by a factor of 10.  That would clearly not be a smart thing to do.
If you check out the values in my example you see the signal is only weakened by a factor of 0.9999. 
My point is that the load in the spice example looks like a resistor in parallel with a capacitor (where the capacitor is actually the cable capacitance) so the load impedance is effectively

     R_load / ( 1 + s * R_load * C_cable)

So the load has a pole at s = - 1 / ( R_load * C_cable )

I'm talking about putting a small R in parallel with a large C in the signal line at the guitar end like this

Code Select Expand


                   ---- R ----
   V_signal --->---|         |--------> cable --------> V_load
                   ---- C ----


Where V_signal is the pickup output that would normally go straight into the cable.
The cable is modelled as a capacitor to ground and the load has resistance R_load to ground.

If you choose R and C so that
      
      ( R x C ) = ( R_load * C_cable )

then you end up with all the frequency dependent terms cancelling to give

      (V_load / V_signal) =  R_load / ( R_load  + R )

So there is no dependence on C or C_cable.
You have negligible signal loss since R is much less than R_load.
You do increase noise by adding R but since R is easily 10 times smaller than the pickup resistance the noise increase should also be insignificant.

[Cozybuilder]

Welcome Back Paul- we missed you!