Educate me: AES cabling

Re: Educate me: AES cabling

Glenn: you did appear to assert that cable characteristic impedance is a fiction... and it's not.

For that matter, /connectors/ have a design characteristic impedance as well, a point we all gloss on DMX wiring.

RS-485 PHY was /designed/ to be stupidly tolerant of bad wiring; if uou do everything perfectly, you can probably exceed the DMX spec and still jace everything work ok.
 
Re: Educate me: AES cabling

mackerr said:
Maybe you could explain your previous post then, because the way I read it it contradicts your assertion.

Mac
Click to expand...

Like this line?
"Cable does resist a signal, called impedance, depending in its length, thickness, capacitance and frequency being transmitted."

Many devices have the terminator built in. Not so with DMX and I use RS485 in my day job all the time. I have runs of over 2,000 feet to PLC's in factories all over the world. "NO T TAPPING".
DMX has to use a terminator at the end of a run normally on the loop out of the last device. The receiver side of the signal and not the source side, hello!
CCTV 75ohm devices normally have the end of line built in with a switch so you can lift devices loop thru as do many PLC's, Security panel, Fire panels, many still using RS 485 for communications loops. And a LOOP is not a T TAP.

The main reason for the PAD at the end of any wire is to increase the drive voltages for longer distances. The pad knocks the voltage down to a usable level for the device input. As in the easy to see case of CCTV the drive voltage can be 2 volts, the input circuit will distort with a drive voltage that high. It wants to see 1.1 volt Peak to Peak max. Doing this helps signal to noise, surely a familiar term here. As the input circuit pads the signal down and the noise picked up along the way lowers as well.

Yes, cable maters and mostly because high capacitance cable will turn a square wave into a camel hump quickly. Most data grade cable will have low capacitance. You can see the leading edge on a scope, "| = best, / = not so much"
 
Re: Educate me: AES cabling

Jay Ashworth said:
Glenn: you did appear to assert that cable characteristic impedance is a fiction... and it's not.

For that matter, /connectors/ have a design characteristic impedance as well, a point we all gloss on DMX wiring.

RS-485 PHY was /designed/ to be stupidly tolerant of bad wiring; if uou do everything perfectly, you can probably exceed the DMX spec and still jace everything work ok.
Click to expand...

If it sounds that way it was not my intension. I wanted to point out that the cable is not 75 ohms, or 50 or 110 ohms. The cable (resistance to DC) or impedance ( resistance to AC) is not 50, 75, 110, 600, but it is the terminating load used at the end of a signal cable run. I do know manufactures print these numbers on the jacket but it takes a long long cable to equal 75 ohms of resistance.
 
Re: Educate me: AES cabling

Glenn Adams said:
If it sounds that way it was not my intension. I wanted to point out that the cable is not 75 ohms, or 50 or 110 ohms. The cable (resistance to DC) or impedance ( resistance to AC) is not 50, 75, 110, 600, but it is the terminating load used at the end of a signal cable run. I do know manufactures print these numbers on the jacket but it takes a long long cable to equal 75 ohms of resistance.
Click to expand...

Glenn, RS 485 specifies both terminating load AND a characteristic impedance/capacitance for the cabling.
 
Re: Educate me: AES cabling

Max Warasila said:
Now - the cable itself also must have a certain characteristic impedance and capacitance within a set specification for the signal to be sent without degradation over a distance.
Click to expand...

For a transmission line terminated in its characteristic impedance the capacitance (capacitance per unit length, actually) is irrelevant. Only the characteristic impedance matters.

The capacitance matters for lines that do not have matched terminations. In the case of electrically "short" lines (generally less than 1/20 wavelength at the highest operating frequency) the line is accurately modeled as a lumped capacitor that, in combination with the parallel combination of the source and load resistances, forms a first-order low-pass filter. This is the situation we run into with very long mic cables and perhaps is where the confusion comes from.

--Frank
 
Re: Educate me: AES cabling

In fact, the capacitance per foot still matters. The cable acts as a lowpass filter, rounding off the digital signal more the longer the cable gets, as noted above.

I have never read anything which suggests that that effect is ameliorated by termination at the end of the cable, do you have a source?
 
Re: Educate me: AES cabling

Jay Ashworth said:
do you have a source?
Click to expand...

OK, I'm not going to try to give an on-line course in electrical engineering -- I was merely trying to add some aids to visualization and clear some confusion.

But since you asked, the references I have handy are just my old books:

"Electromagnetics" John Kraus, Keith Carver, McGraw Hill 1973 (Introductory college text)
"Transmission Lines, Antennas and Waveguides" King, Mimno, Wing, Dover (reprint) 1965 (Very readable introduction for the Armed Forces in WW2.)
"The Handbook for Radio Amateurs" ARRL 1994 (Very well written more-or-less math-free introduction.)

--Frank
 
Re: Educate me: AES cabling

Jay Ashworth said:
In fact, the capacitance per foot still matters. The cable acts as a lowpass filter, rounding off the digital signal more the longer the cable gets, as noted above.

I have never read anything which suggests that that effect is ameliorated by termination at the end of the cable, do you have a source?
Click to expand...

As Frank has alluded to, the length of the cable and the wavelength of the signals matter in how you model the cable. For cables that are "short" relative to the wavelength of the signal, one typically models the cable as a lumped impedance (with R, L, and C values from the cable properties, typically length-dependant). This is the case for audio cables in almost all cases we run into in this industry.
For cables that are not electrically "short" relative to the wavelength of the signal, one typically models the cable as a transmission line with loss. In these cases, things like characteristic impedance and proper termination matter, and it is no longer valid to model the cable as a simple lumped impedance.

Cable capacitance can indeed alter the rise and fall times of digital data, but this is typically taken into account in the design of the digital system (and is one of the reasons for maximum cable length specifications).

The lengths at which a cable is no longer "electrically short" are given below for some frequencies of interest. All assume that the cable has a velocity factor of 0.6667, and that the maximum length for something to be electrically short is 1/10 the wavelength.

4khz (telephone-quality audio): 5km
20khz (high quality audio): 1km
250khz (DMX): 80m
6.144Mhz (AES audio): 3.25m
 
Re: Educate me: AES cabling

And as long as the cable is electrically short, the accumulated capacitance is immaterial? I'm fine with that part.

But it sounded to me like Frank was alleging that an /otherwise significant/ parallel capacitance /stopped being significant/ if you properly terminated an otherwise open cable run.

That's the part I challenge.
 
Re: Educate me: AES cabling

The Rob Timmerman post #28 is the correct answer.
Although I have an nit to pick about:
"the maximum length for something to be electrically short is 1/10 the wavelength"
In my view it's about ¼ to ½ wavelength.So in a transmission line system, if the output stage, the interconnect cable and the input stage all have the same characteristic impedance, then the system will appear to be an infinity long cable. (that is no reflections)

Note that at audio frequencies (or under 100kHz) a cable does not have a single characteristic impedance, as the impedance will change with frequency.
 
Re: Educate me: AES cabling

Jay Ashworth said:
And as long as the cable is electrically short, the accumulated capacitance is immaterial? I'm fine with that part.
Click to expand...

To the contrary, it is with short, mismatched lines that we are typically concerned with cable capacitance. This is the "long mic cable" case where the -3dB frequency is 1 / (2 * pi * R * C) where R is the parallel combination of the source and load resistances and C is the total cable capacitance.

But it sounded to me like Frank was alleging that an /otherwise significant/ parallel capacitance /stopped being significant/ if you properly terminated an otherwise open cable run.
Click to expand...

This is indeed what I was alleging. In a matched-termination transmission line the distributed capacitance is cancelled by the distributed inductance and so the capacitance, in itself, does not matter. As Rob pointed out, the frequency limit for a given length of transmission line is typically governed by loss, which, in other than air- or vacuum-insulated lines, occurs mainly in the dielectric. Another consideration in broadband use might be dispersion, in which different frequencies have different velocities, which smears pulses. I'm not enough of a long distance data transmission guy to know, off hand, under what circumstances this becomes a problem.

With respect to what constitutes a "short" line, the lambda/20 figure I gave is conservative and will work in almost all situations -- better safe than sorry. In many circumstances we might get away with using a lambda/4 piece of the "wrong" cable but in RF work lambda/4 is huge. It's the length of a 1/4 wave antenna, or a 1/4 wave stub, for example.

On bandwidth required for binary data transmission, depending on just how the signal is detected at the receiving end, we might want to look at the edge rate (rise and fall times) rather than clock speed. This is true for much of digital hardware design.

Hope I've managed to communicate, at least a little. --Frank
 
Re: Educate me: AES cabling

Frank, you've made more sense out of the topic than I expected to find. Granted, I had to go read up on stuff to make full sense of your posts, but it's certainly better than being told to use digital cable because mic cable isn't actually a twisted pair or shielded. :blush:

FWIW, the rise/fall time of DMX 512 is supposed to be less than 1uS, equating to 1MHz. This is, however, useless to the OP.
 
Re: Educate me: AES cabling

Jay Ashworth said:
In fact, the capacitance per foot still matters. The cable acts as a lowpass filter, rounding off the digital signal more the longer the cable gets, as noted above.

I have never read anything which suggests that that effect is ameliorated by termination at the end of the cable, do you have a source?
Click to expand...

The terminator of the transmission line reduces reflections. To make it simple, reflections act like comb filtering in audio by canceling some data resulting in data errors.

In DMX, the commands are sent continuously so if one instance were messed up, the next might fix it before you noticed. However, as the number of addresses in use increases, the time between commands for a given instrument gets longer and you may observe unintended behavior.


Sent from my iPad HD
 
Top Bottom