How to run non-disruptive tests even if your test gear doesn't do gated measurements!

Updated 6/28/00


Its really nice to have the latest test gear. However, if you don't, you can still run carrier to noise ratio (C/N) and in-channel response (ICR) tests using nearly any analyzer and some signal level meters. It won't be quite as convenient as using gated measurements, and it may take a little more technical savvy on your part, but it is possible using older or non CATV-specific analyzers and even some of the newer meters.

There are more benefits to non-disruptive testing than you may have realized:

    1. Of course, fewer service disruptions means fewer subscriber eruptions!
    2. By using in-service test methods, the amount of time spent running tests at night can be minimized.
    3. Testing with channels in service means you're testing under the same conditions that your subscribers are seeing. Turning channels off for C/N and ICR tests, then testing at the tap, doesn't necessarily reflect what your subscribers are getting.
    4. By running C/N and ICR with channels in service, there's not much left for the tech in the headend to do during proofs. As a matter of fact, for about $200 worth of commonly available devices, you can eliminate the need for a tech in the headend during proofs.

In-service C/N tests using any analyzer

First, let's look at what we need to do. The rules say that "The ratio of RF visual signal level to system noise shall be as follows: ….. As of June 30, 1995, shall not be less then 43 decibels." If you dig a little deeper, you'll find the definition of system noise in 47 CFR 76.5; "…. System noise is specified in terms of its rms voltage or its mean power level as measured in the 4 MHz bandwidth between 1.25 and 5.25 MHz above the lower channel boundary of a cable television channel."

Another way of saying all this is that C/N needs to be at least 43 dB as measured in a 4 MHz bandwidth starting at the visual carrier and going upward in frequency. This also means that making the measurement in the guard band between channels doesn't really meet the requirements. (There are additional reasons for not testing in the guard band, but that's a topic for another paper.)

Part 76 also tells us how to measure system noise. One way is to actually measure noise over the entire 4 MHz bandwidth specified (this can't be done with the channel in service due to the presence of the visual carrier). Another method is also discussed: "If it is established that the noise level is constant within this bandwidth, a single measurement may be made which is corrected by an appropriate factor representing the ratio of 4 MHz to the noise bandwidth of the frequency selective voltmeter." The latter is the way nearly all C/N measurements are performed. Of course, spectrum analyzers and signal level meters are both "frequency selective voltmeters".

So, we need to measure the visual carrier, then measure noise somewhere between the visual carrier frequency and 4 MHz above the visual carrier frequency. Assuming that the noise spectrum is flat, C/N can then be calculated. Of course, spectrum analyzers that were designed for cable TV measurements, but don't have gated capabilities, perform the calculations for you. Examples include the Calan 1776 and Tek 2714. With other "general purpose" analyzers, you may need to perform the calculations.

A common way of performing the test is to measure the visual carrier level, then remove the video input to the modulator so system noise can be measured with the specified range. This, of course, disrupts service.

Rather than remove the video, we could install a notch filter at the input to the modulator. If the notch is set to the frequency we use for the noise test, we can perform the test with the channel in service.

For example: If we normally measure the system noise 2.7 MHz above the visual carrier frequency after removing the video input to the modulator, we could do the same thing by installing a notch filter, tuned to 2.7 MHz, at the input to the modulator. The filter would effectively remove the video signal — but only at frequencies near 2.7 MHz. The channel would remain in service (although it would suffer some distortion). 2.7 MHz is a convenient frequency because the impact on the both the luminance and chrominance is minimal. Here are a couple of sample images with and without the 4100 In-Service Test Processor's filter enabled.

Without filter

With filter

Image compression used to condense the images for this web page makes the effect less noticeable.

Click here to see larger pictures with and without the filter

The most apparent effects are on closely spaced text. For most "scene" types of images the effects are hardly noticeable (especially when they are not side by side as shown here).


The spectrum with the noise filter in place looks like this:

Display on a Tek 2714

Display on a Calan 1776

Note: The test conditions for the two samples displays are different. For the display on the left the C/N was approximately 55 dB. For the display on the right is was approximately 45 dB.


To run the C/N test, simply use the same method that you normally use, except be sure to measure noise 2.7 MHz above the visual carrier frequency.

 

In-service ICR tests using any analyzer

Again, let's check the rules: "The amplitude characteristic shall be within the range of ±2 decibels from 0.75 MHz to 5.0 MHz above the lower boundary frequency of the cable television channel, referenced to the average of the highest and lowest amplitudes within these frequency boundaries. ….. As of December 30, 1999, the amplitude characteristic shall be measured at the subscriber terminal."

Restated, this is ± 2 dB from 500 kHz below to 3.75 MHz above the visual carrier and needs to be measured at the output of the set-top converter starting in the year 2000.

If gated measurements aren't available, most systems run this test by replacing the video input to the modulator with a full field test signal such as multiburst. The amplitudes of the various multiburst packets are then measured in the field and the peak to peak variation (divided by 2) is recorded.

Another approach is to insert a high level multiburst test signal into the vertical blanking interval and use the analyzer's peak display mode to capture the signal. Because the inserted signal is high level (100 IRE), it will normally produce more energy than the video program at each of the test frequencies and will, therefore, be displayed.

There are two common methods for displaying the test signal. Use the analyzer's peak hold mode, or use the peak display mode and slow down the sweep speed to roughly 1 second per division. Either method will produce the desired results. However, slowing the sweep rate is usually faster than waiting for the trace to "fill in" using the peak hold method. Some instruments, like the Calan 1776 and the newer Wavetek Stealth receivers don't allow independent control of the sweep rate, so you must use the peak hold mode with them. The key is to select long dwell times for those instruments.


Click here to see displays captured on various analyzers and a Stealth receiver.


When using this technique, watch out for a couple of "traps"

1. Just because the test signal is 100 IRE (and no part of the video signal should be larger), there's no guaranteed that you will always be able to see all the multiburst packets. The reason is that the display is a complex combination of:

For example, if the video happens to have a lot of 500 kHz spectral content, it is possible to exceed the energy that occurs during the multiburst’s 500 kHz packet (due to the limited amount of time that the packet is available). The 4100 and 4200 In-Service Test Processors were specifically designed to minimize this possibility. You may notice that there are no gaps between its multiburst packets — allowing maximum duration for each packet. If you do see a packet being swamped by video, its usually a matter of waiting a few seconds for the video content to change.

2. Also be aware of the effects of the analyzer's resolution bandwidth (RBW) filter. The design of the resolution bandwidth filters of some analyzers (like HPs) are a gaussian shape as opposed to a more rectangular shape used by others (like Tek). There are pros and cons to both designs and neither is always better than the other. However, when using an analyzer with gaussian shaped filters, you may not be able to see the 500 kHz sidebands using the default RBW setting (300 kHz). If that's the case, simply switch to 100 kHz RBW to make the packets visible.

Back to Technical Papers Index

Home

Products

Tech Papers

Contact Us

Gary Andrews
Television Measurement Services
garya@tvms.net
www.tvms.net