A Clean Audio Installation Guide™

3.0  INTERCONNECTION SYSTEMS
3.1   Interconnect Bandwidth
3.2   Interconnect Slew Rate
3.2.1  System References
3.3   A Cable Problem
3.4   A Second Cable Problem
3.5   System Frequency Response
3.6   Interconnect Rules
3.7   Amplitude Advantages

3.0 INTERCONNECTION SYSTEMS
While not yet universally adopted, the voltage sourced balanced interconnect system is really the only viable alternative for the audio professional. The single ended IHF interconnect system found in consumer and "semi-pro" equipment is not usable because of its inability to reject normal power line related voltage differences. Although we will do everything in our power to minimize these voltages by carefully tying the various pieces of equipment together, voltages will still exist to some degree at various frequencies.

The 600 ohm power matched system developed by the Bell System for the telephone industry is falling from favor, and should have long ago. With modern amplifier technology it is no longer necessary, nor desirable, to terminate audio lines with a "matched" low impedance, except with very long interconnects (1/10 or greater). The foil-shielded audio cable that we use today does not have a characteristic line impedance of 600 ohm but rather near 100 ohm, or below. The true audio transmission line is rare and typically will only exist with the telephone company, very large networks, stadiums, and various military installations. It is possible to see the effects of unterminated cable at one tenth of a wavelength, at the highest frequency of interest (20 kHz) - i.e. approx. 3250' if the velocity-of-propagation constant is 0.65. In practice, with runs of 2000' or more, it might be best approached from a power matched transmission line perspective. But to do so, the correct impedance of the cable should be determined (usually between 50-90 ohm, and proper interface provided.

3.1 Interconnect Bandwidth
The voltage sourced interconnect system with low, 50-60 ohm, source impedance and relatively high (100k ohm or higher) input impedances, is becoming the accepted practice. An output impedance of 50 to 60 ohm has been found to be the optimum drive Z for today's foil-shielded audio cables by ABC⁽¹⁾ in New York, and by Deane Jensen.⁽²⁾ The advantages are threefold:

• Less power drawn at low frequencies from the source equipment (normally no 600 ohm loads), and therefore, less heat generated;
  

• Lower (14 dB lower) noise pickup by interconnect lines because of the lower equivalent line impedance; and most of all,
  

• Five to ten times the cable length that may be driven for a predetermined small signal high frequency interconnect cut-off.

To understand this last statement, one needs to see that the output impedance of a piece of audio equipment combines with the capacitance of the interconnect cable to form an R-C low pass filter.

For example, assume a circuit with 1000 feet of shielded pair (oft-times found with TV remotes) whose capacitance is 32 pF per ft between conductors, a drive impedance of 600 ohm and a bridging (100k ohm input. The high frequency small signal cutoff of the interconnect is 8.34 kHz! Not exactly HI-FI. (See figure 2)

With the line terminated in 600 ohm, the situation improves to 16.68 kHz but is still nowhere near what we are looking for in the demanding world of high quality audio. By dropping the source impedance to 60 ohm, the small signal bandwidth of our interconnect moves out to 83.4 kHz, an even better situation. If we use 1000 feet of Mogami 2944A cable at 6 pF per ft then our small signal bandwidth will move further out to 442 kHz. This is a greatly improved condition, but it is still not the total story, as we shall see in a minute.

The above calculation for F_c is found by;

[1.0]

Where:

R is the output impedance (60 ohm),
C is the cable capacitance, and
F_c is the small signal, high frequency cutoff.

Our tests indicate that an 80 to 90 ohm source yields a maximally flat response. Lower source impedances will result in high frequency peaking because of series inductance in the cable, not accounted for in our simplified equivalent circuit. However, the best compromise between bandwidth and high end response, even though some peaking takes place at the very highest frequencies (above 100 kHz), is 60 ohms.

3.2 Interconnect Slew Rate
In addition to the small signal bandwidth limitation set by the interconnect low pass filter, an additional problem is created by the cable's capacitance. If we are to drive our cable to any appreciable voltage swing at high frequencies, we will need current from our amplifiers to feed the cable capacitance. The amount of current that is needed is in direct proportion to the total cable capacitance and output slew rate at the highest frequency to actually be sent over the cable. Slew rate is the rate of voltage change per unit time. We are, in our usage of the term, referring to the "linear" slew rate of an amplifier - that is, the amplifier is not allowed to go into saturation. Linear slew rate is defined as:

[2.0]

Where:

SR = Slew Rate = , a change in voltage with respect to time,
f = the minimum upper frequency for full voltage response, and
V_p = the peak output voltage from the amplifier.

The amount of current that is required to feed the cable, or in fact any capacitance, is given by;

[2.1]

3.2.1 System References
At this point, we need to digress a minute to define our system reference nomenclature. In this paper we will be using the term dBu. 0 dBu is defined as 0.7746 volts. This is the same voltage that would be found on an audio transmission line operating at 0 dBm on a properly sourced and terminated "600 ohm" line. The use of this voltage as a reference is very desirable because of the vast accumulated experience in 600 ohm systems and because of the readily accepted volume indicator commonly called a "VU meter". The VU meter is, of course, a voltage measuring device, where 0.7746 volts will give a "0" indication when fed with a steady tone.

While the dBu may not be a broadly recognized standard, it is an official standard among the Nordic countries in Europe. It is found in Nordic N-10 standard. Also, its common usage, particularly in Europe, causes us to accept it as the most logical way to define the voltage reference that relates to the power matched system. Occasionally, in some of our older documentation, the term dBv will be seen. This has the same meaning as dBu. Other authors will use dB/.7 or dB/0.775 to indicate the same voltage reference. One Japanese manufacturer uses the term dBs. And now back to our cable problem.

3.3 A Cable Problem
If we have 1000 feet of cable as above and we expect to be able to pass 30 kHz at the maximum output amplitude of our distribution amplifier (we will use the maximum output for most equipment of recent design of +26 dBu min., and of +30 dBu max.), what is the required drive current from our DA that will enable the above criteria to be met? Remember, O dBu is a voltage reference of 0.7746 volts.

[3.0a]

Therefore:

[3.0b]

Solution 1
From the above, +26 dBu is equivalent to 15.46 volts RMS output. The peak output voltage is 1.414 x the RMS value, or 21.85 volts peak. Therefore:

Solution 2

+30 dBu out = 24.50 volts RMS, peak output of 34.69 volts. Therefore:

Now that we know the slew rate needed to meet our desired conditions, let's calculate the current necessary to feed the line. We recall from our earlier work that 1000 feet of cable has a capacitance of 32 nanofarads (32 pF per foot), and,

[3.1]

Therefore:
@ +26 dBu out,

@ +30 dBu out,

Multiple cables fed from a DA, of course, multiply the total current drain from a unit.

Let's suppose that we wish to drive our line with SMPTE time code and that a 100 kHz full voltage output is deemed necessary to preserve waveform integrity.

@ + 30 dBu

In this case our line length must be limited to considerably less than 1000 feet. If we want a full output to 100 kHz, then we really need a small signal interconnect bandwidth of at least 300 kHz. This limits the total cable capacitance to 8.84 x 10^-9 Farads. At 32 pF per foot, the maximum cable length we can use is 276 feet. The current required is:

@+26 dBu out,

@ = 30 dBu out,

If longer cable lengths are required, low capacitance cable is the only alternative. Mogami 2944A shielded pair at 6 pF/ft allows almost 1500 feet of cable to be used and still meet the above criteria.

3.4 A Second Cable Problem
If a piece of equipment uses an NE5532 as an output amplifier, what is the maximum cable length that can be driven to the full output amplitude capability of the equipment at 30 kHz?

Solution
From the manufacturer's data sheet we find that the output current limit of the NE5532 is approx. 40 mA. Let's assume the use of ±15 volt power supplies and, therefore, the maximum output amplitude capability of the device is approx. +26 dBu.

We already found that +26 dBu at 30 kHz represents a slew rate of 4.110 x 10⁶ volts per sec; rewriting an earlier equation:

[4.0]

Therefore:

At 32 pF per foot we find a maximum of 303.5 feet of cable is permissible. With Mogami 2944A @ 6 pF per foot we find 1619 feet of cable may be used. If the equipment is capable of + 30 dBu out, then only 6.127 nF, or 191.5 feet of cable or 1021 feet of the Mogami 2944A, may be used.

3.5 System Frequency Response
All of the above assumes a 30 kHz interconnect full voltage capability. That is our recommendation for minimum system performance. It is also our recommendation that a 200 kHz small signal interconnect bandwidth capability be the design goal to achieve the flattest response and minimum phase shift at high frequencies for good overall system performance.

It is important to recognize the difference between the small signal bandwidth of an interconnect and its slew rate limitations.

Small signal bandwidth sets the 3 dB cutoff of the interconnect filter, which in turn describes the flatness and phase response back at 20 kHz. It must be remembered that every element of an audio system will contribute it's 3 dB cutoff and associated phase shift to the overall performance of the system. Every element must be viewed as one section of a large multi-pole low pass filter; and while one element may have adequate response to 30 or 40 kHz, it is the cumulative effect of these filter sections that is of major concern. At first glance, the proclaimed need for a wide bandwidth of 200 kHz in both the equipment and the interconnect may seem outlandish. However, when you realize that television stations and networks, for instance, may have from ten to twenty or even more pieces of equipment in the audio chain, each contributing its cutoff characteristics, we begin to realize the magnitude of the problem in achieving adequate high frequency performance through the system.

If a piece of equipment or an interconnect has a 100 kHz bandwidth and has a simple single-pole, 6 dB per octave roll off (often only true for interconnects), the following chart can be used to estimate the system roll off.

Manufacturers of audio equipment have for years mistakenly considered an upper bandwidth of 20 to 30 kHz to be totally adequate for their equipment. This narrow viewpoint, of course, fails to see their equipment as an element in a long chain, and potentially the limiting element. While we will actually never "use" - that is, put a signal into that upper portion of the 200 kHz bandwidth - it must exist to achieve the necessary 30 to 40 kHz system bandwidth.

We have often been told that "an audio chain is only as strong as its weakest link" - would that it were as strong as its weakest link.

Slew rate limitations, on the other hand, are large voltage swing limitations and again, have to do with the actual current output required of a stage that is driving a capacitance. It is important that no amplifier be allowed to slew limit. To do so produces high frequency intermodulation distortion. If an amplifier can provide adequate current to a cable to allow full output swing to 30 kHz at low THD, the chances are practically nil that it will ever slew limit with normal audio.

3.6 Interconnect Rules
So then, to maximize the interconnect performance in a system:

• Move to a low output impedance of 60 ohm (balanced) on all equipment. Many pieces can be modified by changing their build-out resistors, however, current demands must be considered.
  

• Keep the cable runs as short as possible.
  

• Where long lines are unavoidable, use low capacitance cable (Mogami 2944A is normally in stock at Benchmark Media Systems, Inc.); also, Belden Data Cable at 12 pF per foot is an alternative to the Mogami 2944A.
  

• Make sure the device that is driving a long line can supply the necessary current at high frequencies.

One word of caution concerning distribution amplifiers is in order. A number of manufacturers of DAs have been changing over to a 60 ohm output impedance. However, in many cases the conversion consists of simply removing 300 ohm build out resistors and replacing them with 30 ohm resistors without regard to the possible output current demands that can occur.

If we establish as a criteria the need to be able to support a third of a DA's outputs in a short circuit condition, remembering that a DA is an  insurance policy, then, for a DA that uses resistive splits and 10 outputs, 3 shorted outputs represents a 10 ohm load on each of the two output legs. As a result, upwards of 3 amps peak output current capability is needed from the DAs. This drive capability is in addition to that required to feed cable capacitance. It is a rare DA that can handle this type of current demand. One shorted 60 ohm output should be reflected as only a 0.02 dB drop in amplitude at the other outputs operating at maximum output amplitude. If a distribution amplifier will not provide this degree of isolation, the question must be asked, "Is this truly a distribution amplifier?", for is not the whole purpose of a DA to provide isolation?

3.7 Amplitude Advantages
Another benefit of the 60 ohm output impedance is the 0.8 dB amplitude difference between a bridging input and residual 600 ohm inputs. The advantage of this may not be immediately obvious, but let's consider a not unusual TV situation where the nominal system reference (O VU) is +8 dBu. Next, we must realize that the normal peak to average ratio (crest factor) of most audio is 8 to 10 dB, but, in fact, may be as high as 16 dB, or even higher on very percussive material. Now, let us assume that we are feeding a bridging input from a 600 ohm system (the output voltage will be 6 dB higher without a 600 ohm "termination"). And lastly let's add up our amplitudes:

When you understand that a unity gain differential (op-amp) input stage running from ± 15 volt supplies (typical for many pieces of equipment) clips at about +21 dBu, and the input clip of a properly designed input stage operating from the same ± 15 volt supply rails is +26 to +27 dBu, you begin to see the need for reducing that last 6 dB term.

Go to: Section 4.0