[Ulas]

I do believe measurements have value, but I don't believe that they are absolute.

Would you please elaborate on what you mean when you say measurements are not absolute? Do you mean the physical properties being measured are not well defined and understood or do you mean the measurements themselves aren't sufficiently accurate and repeatable to suit you?

Perhaps the "EEs" in the other forums write what they do because they know it sends audio subjectivists, like you, into conniption fits.

[Occam]

...Sooooo...... how much capacitance would have to be in that interconnect fed by a 2k ohm source impedance to be down 3db at 40kHz?

[EDIT - The original qestion was "how much capacitance... with a 47kOhm load input impedance?",  prior  to Eico pointing out the error. The new correction of usings the source impeadance rather that the load impeance,  show that interconnect is of far less concern than what was previously presented pror to correction]

No takers?
Well, the answer is [assuming a best case perfect voltage source]  2,000pf
and for a more reasonable 25khz -3db down point, for those without  SACD/DVD-A sources and Murata supeertweeters it would be about 3,200pf.

We generally don't need to be quite overly concerned with capacitance in interconnects. The major occasion for concern might be highly capacitve cables connected to the outputs of some passive attenuators.

Here are some simple formulas for doing these simple calculations.
Capacitance, C,  is in picofarads, pf
Source Impedance, I,  is in kiloOhms, kOhms
Lowpass pole frequency, F, the -3db down frequency, is in Hz, cycles per second.

F = 159,155,000/ (C x I)
C = 159,155,000/ (F x I)
I = 159,155,000/ (F x C)

[Ulas]

Occam,
So, what do those calculations tell us? That no one can possibly hear the difference between any ICs whose capacitance is less than 135pF? After all, because normal human hearing, at best, is limited to 20-20,000Hz and it had been proven that humans cannot reliable detect amplitude differences less than 3dB, it stands to reason that no one can possible hear a 3dB difference at 25KHz. Have I got that right?

The problem I have with those numbers is the 3 dB assumption. I think limit should be .01 pB, as in one one-hundredth of a picoBel. Three dB may be the limit with test subjects who are asked to discriminate between pure, single frequency test tones but that’s not how the ear/brain relates to complex sounds. The humans may not be very good at evaluating the absolute amplitude of a single tone but they are exquisite at discriminating the relative amplitudes of the different frequency components in a complex audio signal. Who has not had the experience of recognizing a familiar voice over a noisy, distorted, and band-limited telephone connection, or in a noisy environment? Who can’t recognize the difference between a trumpet and a banjo, even if they are playing the same notes? I guarantee, the relative differences in the amplitudes of all the different partials that comprise the composite sound of each voice or instrument is less than 3dB!

[JoshK]

No takers?

I had intended to do the math and respond but I've been buried at work with a new product launch and haven't had much time to breathe.

Ulas, you raise a very good point!  I like the way you think, it is good to ask questions and challenge such reasonings because sometimes the context is very relavent.  I don't have the answer to that question, but it is a good point, worth dwelling on.

[Occam]

Ulas,

[Edit - Your question was, I believe, based your reading my incorrect post with flat out incorrect numbers, which severely overstated the 'dangers' of fairly typical interconnects. And with your interconnects which are particularly minimal capacitance, that is a non-issue. Apologies for any confusion I've caused]



I'd be very interested in your possibly measuring the LCR characteristics of your design, as it appears to be quite low in capacitance..

[Ulas]

I'd be very interested in your possibly measuring the LCR characteristics of your design, as it appears to be quite low in capacitance.

I can’t measure L with sufficient resolution to give meaningful numbers. If you read the description of the ICs at the beginning of this thread you can find C. Most wire gauge tables will tell you the R of different gauge copper wires.

[beat]

This is getting interesting,
How does one go about properly measuring capacitance/inductance of an interconnect? I'd like to see where some of my experiments have gotten me aside from just liking the sound.
thanks guys,
beat

[eico1]

Sooooo...... how much capacitance would have to be in that interconnect feeding a 47k ohm load to be down 3db at 40kHz?

I don't think you can calculate the roll-off frequency without knowing the output impedence of the source.

Knowing the load impedence is usefull if you are concerned about the cables inductance interaction.

steve

[Ulas]

Beat, you can measure L, C, & R with a bridge or a meter. A few DVMs include the ability to measure C and fewer also measure L. Unfortunately, such DVM rarely have enough resolution to be useful measuring audio cables, which are usually measured in pF an nH.

Once you have a meter with the needed resolution you also have to remember that whenever you are measuring very small values you have to subtract the residual capacitance, inductance, or resistance of the test leads and of the meter itself. To measure ICs I use a jig that is a RCA jack attached to a block of wood. I attach the test leads to the solder lugs of the jack and measure the capacitance of the jig. Then, without moving the jig or the test leads, I plug an IC into the jack and measure the capacitance. The capacitance of the IC is the second reading minus the first.

It’s the same with inductance except you have to short the far end of what ever you are measuring. The inductance measurement requires a closed loop. The inductance of the jig is measured with a shorting plug in place. The IC is measured with a shorting jack at the other end.

[Occam]

I don't think you can calculate the roll-off frequency without knowing the output impedence of the source.

You don't? Would you please post the corrected calculations. or a url, reference book.... to justify your conclusion?

TIA

[eico1]

The equation is F=1/2piRC, 2pi being about 6.28.

So in the picture below, a 10k output impedence with a 16nF cable capacitance rolls-off starting at 1/6.28*10,000*.000000016 = 995Hz!

As you add a resistive load across C, like the input impedence of an amp would be, the roll-off point does not change as the attenuation is fixed across all frequencies.

steve

[Occam]

Eico1,

You are absolutely correct, and I'm quite embarassed. No excuse other than I was looking at the reactive voltage divider arse backwards. As Eirco1 said, it is the ouput impedance of the source along with the shunt capacitance of the cable which determines the highpass rolloff.(+the series R of the cable and input capacitance of the amp, etc,... contribute but are generally not material)

Given Eico's correction, and the fact that source impedances are a few thousand Ohms, at worst, the level effect is far less than would be with the typical load input impedance.

I'm going to edit, after the fact,  the my posts  to correct the conclusions and which impedance is of concern..

Thank you Eico1.

[PT914]

Ulas,

I tried your first IC since I have alot of hollow poly rope, very difficult to follow weave without kinking wire, so I tried your second IC instead.  I substituted 1/4 inch poly rope for the teflon tubes, just pasted the wire through the center and weaved the four ropes as you did.  Then I pasted the weaved bundle into the center of the 3/8 inch rope, tight fit but do able.  Can't compare it with your version but it sounds good.  It has alot of poly around the wire.  Thanks for sharing your hard work.

Philip

[JoshK]

For weaving the rope, I can't help thinking of what my mother would do, being a knitter.  Take a small knitting needle, thread the end of the wire through the eye and bend it over once and then use the needle to guide the wire through the weave without kinking it.  Then will all done you can cut the tiny bit of bent over wire off.  Viola, or atleast sounds good on paper.

[Ulas]

…at least sounds good on paper.

When I first tried to weave the wire into the rope I used a needle, as you described. It’s doable but trying to pull the wire through two or more tucks at a time leads to breaks, kinks, and snarls.  Even doing one tuck at a time leads to kinks and is very slow. There are other ways and I am sure once you have the rope and wire in your hands and realize that it’s not like your mother’s knitting, you will find a solution that works for you.

For me, the attraction of DIY is solving problems; like how to weave a tiny wire through many feet of rope without kinking and provide adequate strain relief with an Eichmann RCA plug. Assuming that other DIYers are similarly attracted to the hobby, I didn’t overburden the recipes with details about how I solved the problems. I leave it to each maker to devise a method that suits him. But, if you have tried and have trouble fusing the polypropylene or can’t weave the wire through the rope without kinking, contact me off-line and I will give you some hints.