Archives

Feature: Digital Signal Carriage: Pitfalls, Concerns, Considerations and Suggestions
By

In this second installment of this three-part series on digital data, Acterna's manager for broadband training John Downey looks at modulation schemes, explains how new digital service influences existing analog channels and provides tips on placing your haystacks (digitally modulated carriers).

Many engineers don’t realize that adding digital signals to the channel line-up could cause composite intermodulation noise (CIN). Because this is considered a distortion like composite triple beat (CTB), some believe that CIN adds on a 20-log basis. Increasing levels by 1 dB could make the CIN ratio worse by 2 dB. Doubling the cascade could make the CIN ratio worse by 6 dB. CIN will add to the regular thermal noise on a 10-log basis to give a worse carrier-to-noise (C/N) ratio for the existing analog carriers.

If we assume the digital channels will be between 550 MHz and 750 MHz, the simple mathematics of addition and subtraction would indicate that the worst area will be between 250 MHz and 400 MHz. This forces us to know the specifications of our transport equipment and to verify how it was specified.

There also could be out-of-band (OOB) noise to contend with, not to mention residual noise from "inactive" devices. Placement of haystacks and analog channels side-by-side could have a debilitating effect. It may be warranted to look at the device (modulator, modem, set-top, TV, VCR and so on) with a spectrum analyzer from 5 MHz to 1 GHz before installation. You may be surprised at what comes out of a supposedly inactive device (local oscillator leakage, harmonics, residual noise, and other spurious signals). Be sure not to overload your analyzer with too much signal level. You could interpret the readings as a problem that is not actually there.

Anomalies affect digital differently than analog

In the United States, the Federal Communications Commission (FCC) requires a 43 dB C/N for analog visual carriers on cable systems. Most systems are designed for 48 to 50 dB. You usually don’t get much better than 50 dB or so C/N out of the fiber node.

Many believe that if the analog channels look good, then digital will have no problems. Digital data is very robust, and there’s no noise hidden in the service like analog channels. It also can survive with a much lower C/N than analog, then reality hits, and you find that this ratio is very dependent on the type of modulation, adaptive equalization, forward error correction (FEC), and other factors.

For example, 256-QAM (quadrature amplitude modulation) requires a 34 dB C/N, assuming thermal noise, to achieve a 10-6 bit error rate (BER). That’s one errored bit out of 1 million bits transmitted. This BER is a typical requirement for digital set-top boxes to operate reliably. Also, this is not taking into account intermittent problems that may not be apparent on an analog picture, but could cause slight tiling on a digital picture.

Let’s look at one example: A 256-QAM digitally modulated carrier in an National Television System Committee (NTSC) system is 6 MHz wide and operates 10 dB lower than analog channels. The end-of-line C/N for the analog carrier is 48 dB. Digital data is added to the line-up and causes intermodulation noise that makes the carrier-to-composite noise ratio (CCNR) drop to 46 dB. (See Figure 1)

The analog channel level is +15 dBmV out of the tap, making the composite noise in 4 MHz (15 - 46) = –31 dBmV. If the digitally modulated carriers’ total power is 10 dB lower than the adjacent analog TV channels’ amplitude, it would be +15 - 10 = +5 dBmV average power. This C/N ratio is average power to root mean square (RMS) noise in the corresponding bandwidth of the carrier.

Because the digitally modulated carrier has filter roll-offs, it isn’t exactly 6 MHz-wide, because of the filter alpha. 256-QAM has a filter alpha of about 11 percent, therefore, 89 percent is the actual payload of approximately 5.4 MHz. The noise in 5.4 MHz would be; –31 + 10*log (5.4/4) = –29.7 dBmV. This makes the C/N of the data 5 – (–29.7) = 34.7 dB. This is very close to the 34 dB cliff we mentioned earlier. If we run the haystacks higher, they will have a better C/N, but there will be more laser clipping and CIN. Once again, we must know the consequences of our actions and know how to strike a compromise.

Standing waves and group delay

Another point to keep in mind is the effect of standing waves. Standing waves are created from impedance mismatches that cause signal reflections. These reflected signals will add in or out of phase with the original, causing a standing wave. Any defective component, low return loss connections, damaged cable, connectors, ground block, splitters, etc., can cause standing waves.

Most TVs have a poor return loss of approximately 5 dB to 6 dB, but the fact that we hit it with a relatively low signal (0 dBmV) actually protects us somewhat. Set-top boxes may buffer it also because they usually have better return loss than TVs and VCRs. Analog carriers have lived with this problem with little or no detrimental effect in most situations. Digital data is a different story. Remember, digital channels are either perfect, tiling or out. The difference between perfect pictures and none often is less than 1 dB.

For any analog type of problem (hum, ghosting, venetian blinds and so on) to show up on a digital channel, it must be in the house. Remember, the TV is more than likely still analog, and the digital signal is being converted back to analog at the digital set-top box.

Ghosting from reflections and direct pickup (DPU) have a different effect on digital data than on analog carriers. Most customer premise equipment (CPE) will have a channel 3/4 selector switch to eliminate direct pickup from a local broadcaster operating on one of those channels. You should never have a strong, over-the-air broadcaster on both channels 3 and 4 in your immediate vicinity.

Adaptive equalizers are used in digital set-top boxes to mitigate reflection and in-channel tilt problems. Avoid the roll-off region of the amplifiers and diplex filters. Even though digital signals may work at a lower amplitude, they can’t compensate for too much in-channel response or tilt. Most digital equipment is specified for < 1 dB per 1 MHz.

Diplex filters and improper amplifier alignment contribute to group delay problems. Chrominance-to-luminance delay on channel 2 is one such problem caused by diplex filters. The color part of the picture begins to smear away from the luminance, or black-and-white part of the picture. Sharp diplex filter roll-offs create group delay, which is aggravated by cascading amplifiers. This will create BER problems with return path data carriers located near this roll-off. Most data equipment is specified for between 70 ns to 200 ns of group delay.

Some suggestions to mitigate these potential problems would be to place narrow carriers near the roll-off, place robust carriers in the low end, stay away from multiples of 6 MHz (8 MHz in PAL systems) because of common path distortion (CPD) and stay away from 27 MHz because of citizens’ band (CB). There’s not much left here, is there?

Once the decision is made to carry digital services, you must decide where to place the carriers. The haystacks could be placed in a region that causes CTB from the analog channels to fall in the guardbands. (See Figure 2) We would lose 1.25 MHz, but this may be good for any OOB noise, which could affect the adjacent channel audio or video. It seems there aren’t many systems taking this approach, though.

Understanding modulation

We need to understand how digital data is transported on an analog system. We also must understand the effect of compression and higher order modulation schemes on robustness. This will maximize efficiency and proficiency of the technician who must maintain these networks.

Digital compression (moving pictures expert group, or MPEG-2, for now) can greatly reduce the number of bits per second required to represent a digital video signal with more than adequate perceived picture quality. Typically, 3 Mbps to 6 Mbps have been found to be sufficient for most programming material, and new compression techniques (wavelets and fractals) may reduce these numbers even more.

Still, baseband digital transmission would require between 3 and 12 MHz to transport these signals. The excessive bandwidth problem of digital video services is solved through the use of higher order digital modulation techniques. Many don’t realize that the word "modem" is actually an acronym for modulator/demodulator. We modulate (manipulate) an analog carrier in amplitude, phase or frequency to represent binary digits (bits), then we demodulate at the other end to convert back to analog for conventional TVs and VCRs. This is the most efficient and economical way to send digital information on an analog plant.

QPSK in the time domain

Looking in the time domain as shown in Figure 3, you can see the four phase states of quadrature phase shift keying (QPSK) modulation. Each phase shift is 90 degrees. Every cycle represents a symbol. In QPSK, there are 2 bits per symbol. This would give a theoretical throughput of 2 bits per Hz. Other modulation schemes like QAM utilize amplitude and phase manipulation to achieve more bits for every cycle. This can be confusing because baud is symbols per second, which is different than bits per second bps.

In the case of QPSK modulation of an analog carrier, we may use 6 MHz for the whole haystack, which may actually be 5 MHz of payload because of filter skirts. This would be 5 MHz of bandwidth or 5 Msymbols per second (Mbaud). This also could be interpreted as 10 Mbps because QPSK has 2 bits for every Hz. The only time baud and bps are equal is when the modulation used only has one bit for every cycle (Hz), such as binary phase shift keying (BPSK), frequency shift keying (FSK), amplitude shift keying (ASK), etc.

As shown in Figure 4, 64-QAM has 3 bits that go to the in-phase (I) side of the modulator and 3 bits that go to the quadrature (Q) side of the modulator. "Q" is 90 degrees out of phase from "I," hence it would be the Y-axis and "I" would be the X-axis. One of those bits is taken care of by a 180-degree phase shifter, and two of the bits would be covered by a four-level linear attenuator.

What we have is a four-level linear attenuator on the I and Q channel, and a 180-degree phase shifter on the I and Q channel. Bring them back together and we have a signal, or symbol, representing 6 bits. This is a cycle of an analog wave that would be phase and amplitude modulated and would look very distorted. Six bits equal two to the sixth power (64) with regard to symbol landings or different combinations of ones and zeros.

Understanding the effect this new digital service will have on the existing analog channels eliminates potential pitfalls before money and time are wasted. There are many anomalies in our networks that may affect this new service differently than the analog service. To avoid some of these problems, systems may try a few tricks like placing the haystacks off-set from a standard frequency plan.

The TV set traditionally has been used as a piece of test equipment. Technicians could observe the analog picture or audio quality and make a diagnosis. Now with digital pictures being perfect, tiling or out, we can’t use the TV set anymore because there is no in-between. We must use some type of test equipment that can demodulate the digital signal and give good, quantitative measurements, such as a QAM analyzer. These devices show the health of the signal that a TV set cannot.

There may also be problems on the network that have been there for a long time, but have been masked by the relaxed requirements of analog TV sets. A problem may appear on a digital channel that was never noticed on an analog channel. Now we enter another frontier in CATV troubleshooting.

John J. Downey is the broadband training manager for Acterna. He can be reached at .

 Back to December Issue

Access Intelligence's CABLE GROUP
Communications Technology | CableFAX Daily | CableFAX's CableWORLD | CT's Pipeline
CableFAX Magazine | CableFAX databriefs | Broadband Leaders Retreat | CableFAX Leaders Retreat