EoW November 2009

technical article

Most analogue cables, rarely tested or verified beyond 1 GHz, have simply unknown performance for HD signals and beyond. Many of these cables have very unstable impedance and exhibit high return loss (reflections). This would be invisible to the installer or end-user until the cable was actually in use. Figure 1 is the output of a standard HD device. As Table 1 shows, a device with 1080p/60 output has double the clock frequency, now 1.5 GHz. The second harmonic would be 3 GHz, and the third, 4.5 GHz. Problems in testing The initial problem with 4.5 GHz, and other high frequencies, can be easily determined by contacting a test equipment manu- facturer or distributor; test gear beyond 3 GHz at 75 ohms simply does not exist. The reason is less obvious. Most test equipment manufacturers would claim there is no demand to go past 3 GHz. At least one cable manufacturer, wishing to test cable beyond 3 GHz, has resorted to having custom-matching networks made for their network analysers [note1] .

Since the third harmonic of the 1080p/60 clock is 4.5 GHz (1.5 x 3 = 4.5), both test gear and matching networks must be verified and tested to those high frequencies.

There is no industry standard for this, but the wider the bandwidth, the greater the amount of data is captured.

A hard filter? Many might question why one needs to go past the SMPTE minimum of the second harmonic. Figure 1 explains this. Figure 1 shows the typical output of an HD device, such as a camera. Even though this, and every HD device, has a hard filter at the second harmonic (1.5 GHz) this does not mean that there is no output, no data, beyond that point. It can be clearly seen that there is significant data out to 3 GHz, even out to 4.5 GHz. Transferring this entire wideband signal, and making sure devices and cable are measured beyond the 1.5 GHz second harmonic, can assure more signal, and therefore a more robust bit stream. This is not to say that an ordinary 75 ohm coaxial cable would be unable to carry this signal, but it is neither tested nor measured to these high frequencies; the user has no indication if the device or the cable in between will work or not.

What to test? While there are basic tests to cable, such as attenuation, there are other tests that will give more meaningful results in the high frequency domain of HD and beyond. Key among these tests is return loss. Return loss tests a device, cable, connector, or any other passive part for its consistency of impedance. 75 ohms was chosen for all passive devices because this impedance offers the lowest loss with small, low-voltage, signals such as video. Therefore, everything should be precisely 75 ohms. However, nothing is precise. While a traditional way might be to look at the impedance, a more powerful way is to look at the reflections that are caused when the impedance is not perfect. Figure 2 shows the return loss graph of a piece of cable. Figure 2 shows the return loss out to 3 GHz. Note that the actual return loss varies with frequency, in this example better than –30 dB for most of the spectrum, occasionally passing –40 dB, with a few spikes at around –25 dB. The two lines above it are, first, the manufacturer’s guarantee, and the shorter line on top, the SMPTE limit line as mentioned in broadcast standard SMPTE 292M. In this example, the manufacturer’s guarantee is that any cable intended for HD will be no worse than –23 dB from 5 MHz to 850 MHz, and –21 dB from 850 MHz to 3 GHz. Such a guarantee should always be sought for passive components such as cable, connectors, patch panels, patch cords, adaptors and similar devices. The SMPTE limit line above that is –15 dB return loss. Table 2 shows various values of return loss against the match achieved: the percentage of signal reflected and the percentage successfully delivered. Howmuch reflection? Even with very poor impedance control, and a return loss of –10dB, the reflection is only 10%, so 90% of the signal is reaching its destination. However this reflected signal is not leaving the cable, as in the effect of resistance turning a tiny part of the signal into heat, but is returning to the chip that is sending out the signal. Early chip designs had real problems with reflected signal, and levels less than 10% reflection were often enough to stop the chip from working.

Figure 1 ▼ ▼ : High definition output

Frequency (MHz)

Figure 2 ▼ ▼ : Return loss on a piece of coaxial cable

Frequency (MHz)

68

EuroWire – November 2009