updated: 7/10/00
If you have test equipment with gated measurement capabilities, such as the Avantron AT2000R, HP 8591C with option 107, Tek 2715 or newer Wavetek Stealth meters, all you need to do is ensure that really "quiet" quiet lines are available for the test. The 4100 and 4200 In-Service Test Processors can provide the necessary quiet lines. Additional information is available under "Products for Efficient Testing"
Even if you don't have the latest test gear with gating capabilities, you can perform C/N tests without service disruption but inserting a notch filter into the video bandwidth of the modulator. See the tech article "How to run non-disruptive tests even if you test gear doesn't do gated measurements".
When running C/N tests in the headend (not required for proofs), you may need to disrupt service in order to measure much beyond 60 dB C/N.
By using non-disruptive test methods and conducting the tests at the output of a set-top converter, there's no need to avoid C/N tests on scrambled channels. The only problem we've seen is that some scramblers generate excessive noise during the vertical blanking interval. So, you will need to insert a quiet line after line 21. In practice, line 25 works well. The 4100 and 4200 In-Service Test Processors can perform this function for you. We use line 25 because some scramblers use lines 22 and 23, and we like to put a multiburst test signal on line 24.
See the tech article "Running Proofs on Scrambled Channels" for details.
Some measurement devices automatically insert or remove input attenuation in 10 dB steps when performing C/N tests. Typically, the instruments will also calculate "noise near noise" correction factors. So, corrections for changes in attenuation are automatically applied. However, at the threshold where the attenuation is automatically changed by 10 dB, the change in correction factor is quite large. Measurement accuracy suffers when large correction factors are necessary. In some cases accuracy may be improved significantly by adjusting the input level by a dB or two (enough to reduce the anaylyzer's input attenuation).
To test at the tap, or the converter. THAT is the question.
How to run non-disruptive tests even if your test gear doesn't do gated measurements