More on 16-QAM

Archives January 2003 Issue
RON HRANAC It’s possible to at least double the raw data throughput of your upstream digitally modulated carriers by making the switch from quadrature phase shift keying (QPSK) to 16-QAM (quadrature amplitude modulation). For instance, if you’re now using 1.6 MHz bandwidth QPSK digitally modulated carriers, the raw data rate is 2.56 megabits per second (Mbps). Switching to 16-QAM will give you 5.12 Mbps in the same 1.6 MHz of RF bandwidth. If you use 3.2 MHz bandwidth digitally modulated carriers, your raw data rate with 16-QAM goes up to 10.24 Mbps!

Despite the success stories regarding the use of 16-QAM in the upstream, there are a lot of skeptics out there. First of all, 16-QAM does work in real-world cable systems, and has been deployed in several locations. This column highlighted a successful deployment in Time Warner Cable’s Wilmington, N.C., division, where 16-QAM has been in use for the past two years (see “16-QAM Success Story” in the September 2002 issue of Communications Technology, also available on-line at www.cabletoday.com/archives/ct/0902/0902_broadband.htm).

CMTS configuration

Making 16-QAM work reliably requires attention to several details. First, be sure your cable modem termination system (CMTS) configuration has been optimized for 16-QAM. One important parameter to configure correctly is modulation profile settings. Modulation profile settings define how upstream data is transmitted from cable modems to the CMTS, and affect burst guard time, preamble, modulation (QPSK or 16-QAM), and forward error correction.

If you don’t have modulation profile settings optimized for 16-QAM, you may see upstream packet loss that at first glance appears to be plant-related. Tweak modulation profile settings, and things will work the way they’re supposed to. Cisco’s John Downey has written a very comprehensive white paper on modulation profiles. It’s available on-line at http://www.cisco.com/offer/sp/pdfs/cable/cable_land/ModProfile_wp51.pdf.

OK, modulation profile settings have been optimized and the switch made to 16-QAM. Now what? Make certain your network meets or exceeds the recommended parameters in the Data Over Cable Service Interface Specification (DOCSIS) Radio Frequency Interface Specification. This is critical, especially with regard to the upstream 25 dB minimum carrier-to-noise, carrier-to-ingress power, and carrier-to-interference ratios. These ratios, which I refer to collectively as carrier-to-junk, are measured at the CMTS upstream input port.

Before you accuse me of improper caffeine levels, rest assured that it is possible to meet or exceed the 25 dB spec. Time Warner’s Wilmington, N.C., division manages to keep its around 30 dB. AT&T Broadband had a corporate spec of 40 dB (20 to 40 MHz) in systems where primary line voice service was being provided. Yes, 40 dB was a real number–I verified this level of performance during visits to a couple of its systems!

Improving plant performance

How can one keep the reverse plant at or better than 25 dB carrier-to-junk? It’s nothing more than getting back to basics: Proper forward and reverse alignment, leakage and ingress management, and good installation practices.

Broadband sweep is the best way to align the plant (make sure you have suitable guard bands programmed around downstream and upstream digitally modulated carriers to eliminate the possibility of sweep transmitter interference). Effective ingress management starts with a good signal leakage program. Where there’s a leak, there’s bound to be ingress.

If you find a leak, fix it. Period. Figure on keeping leakage at or below 5 microvolts per meter. If you’re doing flyovers, target 98th or 99th percentile performance rather than the FCC’s 90th percentile requirement. Subscriber drop installations? Do ’em right. No rocket science here.

Another important parameter is the DOCSIS upstream 200 nanoseconds/MHz group delay spec. QPSK is pretty forgiving, but 16-QAM goes bonkers when group delay gets out of whack. The easiest way to avoid group delay problems is to keep your digitally modulated carriers away from band edges and diplex filter rolloff areas. In the 5-42 MHz reverse spectrum, group delay starts to go in the tank above about 35 MHz, and really gets nasty above 38 MHz. If all else is perfect–optimized modulation profiles, no ingress or impulse noise, good carrier-to-junk ratio, and proper signal levels–and you still have unexplained bit errors, the culprit may be group delay. Move your digitally modulated carrier to a lower frequency, and the problem will likely disappear.

More on upstream group delay

While I’m on the subject of upstream group delay: If you’ve ever wondered how to measure it, one of the best tools I’ve seen is Holtzman, Inc.’s Cable Scope®. This versatile test system characterizes upstream frequency response and group delay from the subscriber premises all the way back to the headend. More information is available on-line at www.holtzmaninc.com/cscope.htm.

One last thing to keep in mind: Your cable modems will lose a little of their upstream dynamic range when you make the switch to 16-QAM. DOCSIS specifies that cable modems must support an upstream transmit level range of +8 dBmV to +58 dBmV for QPSK, and +8 dBmV to +55 dBmV for 16-QAM. If you have any modems that are currently operating at or near their maximum upstream transmit level with QPSK, you may find that their output level isn’t quite enough for reliable operation when you switch to 16-QAM. In most cases you’ll find the maxed out transmit levels are related to excessive upstream attenuation between the modem and first upstream active. The first thing I’d check is the subscriber drop at the affected location.

There really isn’t any magic to making 16-QAM work. It does require attention to detail and doing things right, but for the most part it’s nothing more than what Mike Connelly, vice president of engineering for Time Warner Cable’s Wilmington division, calls “cable 101 stuff.”

Ron Hranac is a technical leader, Broadband Network Engineering for Cisco Systems, based in Englewood, Colo., and senior technology editor for Communications Technology. Reach him at [email protected].