"Cable Length Knob" - Negative Capacitance?

[PRR]

DrAlx> I'm not suggesting weakening the signal by a factor of 10.

Sorry, I can't sit long enuff to figure what you are proposing.

The classic 'scope "low C" probe is a 10:1 divider. Because the response changes where R=C, and 'scope C is approximate, we must trim the little cap for square square-wave response.

Your small-loss network, if passive, gives negligible reduction of C.... or am I missing something (perhaps drugs)?

Active gain plus a C "should" result in less C seen by the pickup. Devilish details.

[R.G.]

> 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.

Correct. It's not passive, it's the same or less complexity as using a buffer in the front anyway, cancellation is always tricky to get perfect, and it needs special cable if you're not also amplifying any noise induced on the "shield", which becomes an antenna in this case.

I kept waiting for someone to step in with the obvious answer after all the Zen "what is the sound of one ground clapping".   

Yep, wire, shield, shield is a good way and can make this work. But the way the industry has chosen, for several reasons, is to convert to shielded twisted pair driven balanced at the send and received and converted back to single ended at the receiver. This works, is reliable, and is all the recording studios will (... willingly, at least) use.

Can anyone appreciate how different the guitar world would have been if the folks designing the first guitars and amps had used a duotriode at the amp input as a differential amplifier and used a centertapped pickup on the guitar, going balanced from the first? We pay for that misstep every day.

[ashcat_lt]

Can anyone appreciate how different the guitar world would have been if the folks designing the first guitars and amps had used a duotriode at the amp input as a differential amplifier and used a centertapped pickup on the guitar, going balanced from the first? We pay for that misstep every day.

GN2 basically grew out of a site that was all about reducing noise from passive guitars, and we have lamented this point a number of times.  Over here at DIYSB, though, it would make our work a bit more difficult since all of our projects would need good Hi-Z differential inputs and balanced outs...

But I was just kind of wondering how this whole thing applies to balanced cabling.  The driven shield plus passive shield thing looks a lot like a balanced cable, and according to johnh's diagrams driving the "ground" side of the cap with an inverted signal should increase the effective capacitance.  Is this true in a balanced run?

[DrAlx]

DrAlx> I'm not suggesting weakening the signal by a factor of 10.

Sorry, I can't sit long enuff to figure what you are proposing.

The classic 'scope "low C" probe is a 10:1 divider. Because the response changes where R=C, and 'scope C is approximate, we must trim the little cap for square square-wave response.

Your small-loss network, if passive, gives negligible reduction of C.... or am I missing something (perhaps drugs)?

Active gain plus a C "should" result in less C seen by the pickup. Devilish details.

I'm looking at Figure 1 on this page:

http://www.nutsvolts.com/questions-and-answers/about-scope-probes

The crude picture I have in my head is that the cable capacitance can be bundled together with the 50 pF on the scope.
So I'm just ignoring the fact that the cable is a transmission line and treating that 50 pF at the oscilloscope end as the actual cable capacitance.

In that diagram, the probe consists of a large resistor and a small capacitor in parallel with it.
At DC, the overall gain is determined by the resistors only:  Giving 1M / (1M + 9M).  The resistors are in the ratio 1:9.
As high frequency, the overall gain is determined by the 50pF and trimable cap which form a capacitive voltage divider.  So if the trimable cap is set to 50/9 = 5.55 pF then you get the same gain at high frequency.
So unless something weird is going on at intermediate frequencies, the overall frequency response is flat.


All I'm suggesting is that you can use the exact same trick of matching resistive and capacitive voltage dividers, but instead of using  9M and trimable 15pF at the signal (i.e. guitar) end,   you use a small trimable 100 ohms instead of 9M and a large fixed 10uF cap instead of the 15pF trimmer. 

I don't know how well this would actually work in practice (if at all).  I just saw the pictures of those spice simulations and thought that if it really is valid to model the cable capacitance that way, then it suggested a simpler way of compensating for it.

[R.G.]

But I was just kind of wondering how this whole thing applies to balanced cabling.  The driven shield plus passive shield thing looks a lot like a balanced cable, and according to johnh's diagrams driving the "ground" side of the cap with an inverted signal should increase the effective capacitance.  Is this true in a balanced run?

It depends on the impedance of the driver and receiver, and how distributed the capacitance is along the cable.

First, cable capacitance is distributed, and to the extent that the cable itself has resistance and inductance, there are "capacitors" you can't reach with active drive because they are isolated by the cable resistance and inductance itself. This is the bane of simple simulation, although mostly it's an issue at higher than audio.

To some extent, wrapping a high impedance line in a shield, then driving the shield with the inverse of the signal does increase the effective capacitance, but it also introduces additional complexities as Paul alluded. If you drive the "balanced shield" with the inverse of the signal how much it acts like a bigger capacitance depends on the driving impedance, and the shield impedance itself as wire, from above. If you drive the "balanced shield" with the same signal as on the wire, there is no voltage difference (conceptually) between the shield and high impedance wire, so the wire sees no effective capacitance. Driving the "balanced shield" with a bigger version of the signal on the high impedance wire forces signal back into the input wire, and is then a form of positive feedback, with all the dangers that implies. It gets complicated.

Shielded twisted pair cancels external interference along the cable by introducing it equally to both sides, where it can be cancelled by the differential receiver. The wire/shield/shield setup lends itself to this, but the likely difference in impedance between the internal wire and the driven shield make this difficult to see as a really balanced pair. Balance means balanced driving impedance too.

Beyond that, there is the issue of the One True Ground. A wire/shield/shield lets you use one shield as a common ground between the guitar and the receiver. A wire/driven shield relies on the rest of the universe to act like a "ground" in making all the loop currents and voltages sum out right. the impedance of the rest of the universe is variable as seen from any one place, and sometimes it'll work OK, sometimes not. It's quite difficult to model "the rest of the universe" in a simulator.

[merlinb]

So unless something weird is going on at intermediate frequencies, the overall frequency response is flat.

The frequency response is flat but the input impedance is not.

you use a small trimable 100 ohms instead of 9M and a large fixed 10uF cap instead of the 15pF trimmer. 

That would not work, since the upper arm in your divider has low impedance, so there is not much separating the relatively high impedance pickup from the cable. It would only work if the upper arm of the divider has a much higher impedance, but that also means a lot of attenuation, i.e. noise.

[Perfboard Patcher]

I do believe the circuit is interesting enough to experiment with.
It's easy to build and with a given combination of capacities (C2, C3+Cable) the potmeter can be used to find that single spot where the resonance peak seems to be transposed up like if  the cable capacitance has been lowered.  1M seems too much for R4. I achieved higher resonance frequencies by lowering R4. What else? A buffer, a rotating switch with pairs of capacitors to bring the resonance peak down.

Ashcat_lt, no breadboard involved.    I used a 15H inductor instead of a guitar pickup to do the tests.



Could the shielding of the guitar cable be persuaded to carry a signal? The schematic is merely to support this conceptual idea, I don't know if there are some safety concerns  involved! If it is really that bad the image can be removed. T1 is a buffer and furthermore there is a poweramp with an attenuator consisting of  a 10 and a 1 ohm resistor.

[DrAlx]

So unless something weird is going on at intermediate frequencies, the overall frequency response is flat.

The frequency response is flat but the input impedance is not.

you use a small trimable 100 ohms instead of 9M and a large fixed 10uF cap instead of the 15pF trimmer. 

That would not work, since the upper arm in your divider has low impedance, so there is not much separating the relatively high impedance pickup from the cable. It would only work if the upper arm of the divider has a much higher impedance, but that also means a lot of attenuation, i.e. noise.

Didn't account for source impedance.  Thanks Merlin.

[johnh]

DrAxl - I think that in a purely passive system, the true voltage divider that defines the output signal relative to a theoretical source needs to include the impedance of the pickup (approximated by R L and C in a simple model). Once those are represented, the addition of the extra R and C in-line as you describe would not be expected to do much.

(Edit: Appologies, I see this has already been addressed above)

[PRR]

> how different the guitar world would have been if the folks designing the first guitars and amps had used a duotriode at the amp input as a differential amplifier and used a centertapped pickup on the guitar

Bah. A simple effective interface was obvious. Wind the pickup to a few hundred (not thousands) Ohms. Run cable to *transformer* to first tube grid.

This is of course how "all" dynamic microphones were run from early days to now. And telephony before that, any time an electronic amplifier had to be inserted.

Les Paul Recording guitar is like that also.

The source need not be center-tapped (except extreme cases).

The line may be floating, balanced, or for short benign runs can be unbalanced. (I've run weak mic signal unbal 300 Ohm 150 feet in chapel, though jack cleanliness was an issue with an AM transmitter just upriver.)

Only low-cost PA and guitars/etc try to wind-up the transducer over 10K Ohms to get enuff signal voltage to get over the grid-hiss. The excuse was that cables were short and clients were un-fussy.

But in guitar-land, we not only wind-up for "sufficient" voltage, for some decades we have been using more-and-more turns for a "hot" output, albeit very ringy then drop-off at the top. If you only play with distortion, this may be good: dist adds overtones, and limiting the input reduces IM hash.

Then again we have the FuzzFace (and 999 copycats) with a naked junction input stage which is very fussy about the source impedance it sees. It "won't work right" with buffer in front, works different for every cable and vol-knob setting, yada.

[merlinb]

> how different the guitar world would have been if the folks designing the first guitars and amps had used a duotriode at the amp input as a differential amplifier and used a centertapped pickup on the guitar
Bah. A simple effective interface was obvious.

If only, if only, at the very least, stereo jack plugs had become the standard for guitar rigs. Then at least we could have had a separate shield, whether the interface was balanced or not. Buggrit.

[amptramp]

> how different the guitar world would have been if the folks designing the first guitars and amps had used a duotriode at the amp input as a differential amplifier and used a centertapped pickup on the guitar
Bah. A simple effective interface was obvious.

If only, if only, at the very least, stereo jack plugs had become the standard for guitar rigs. Then at least we could have had a separate shield, whether the interface was balanced or not. Buggrit.

You have to drive the shield from the source end, so you would also need power wiring.  You could do a carrier current output.  Hey maybe that's what you need - an outer shield, a driven shield and a core carrying DC, maybe as part of a carrier current or 4 - 20 mA instrumentation driver.

[ashcat_lt]

You have to drive the shield from the source end, so you would also need power wiring.  You could do a carrier current output.  Hey maybe that's what you need - an outer shield, a driven shield and a core carrying DC, maybe as part of a carrier current or 4 - 20 mA instrumentation driver.

I think we've been talking about just plain balanced wiring, then merlinb was talking about separating what we at GN2 call "signal return" - the "ground side" of the pickup coil and other electronics in the guitar - from the "shield grounds" like the backs of the pots, the switch frame, any foil shielding in the guitar and the cable shield itself*.  None of that really needs power.

I'd like to mention that the diagrams JohnH drew in my OP are distinctly not a driven shield design.  The opamp drives a simple capacitor which appears to be in parallel with the cable capacitance.  Caps in parallel add, so normally you can't get two in parallel to make a value smaller than either.  That would require putting a cap in series with the cable cap, which it should be obvious is not possible.  This whole thing is to make that cap in parallel "look negative" so that the total gets smaller.


*Another important aspect of that would be that we could more easily implement a "safety cap" in series with the touchable stuff connected to the shield to stop DC - and reduce AC - leakage caused by amplifier faults and/or faulty wiring all without affecting the actual tone of the guitar.

[Perfboard Patcher]

If you're not a tweaker, don't bother to read.

This kind of circuit is was referring to: