In Reply to message #372149 by aclementsiv
Old School Member Buffalo Chips is not online, or is invisible.
1/23/2017 2:00:27 PM
Buffalo Chips Member #: 7190 Registered: 1996-2001
Posted: 471 View all posts by Buffalo Chips
Company: Interactive Broadband Technical Services Occupation: HFC Engineer/Consultant Location:
Re: Over equalizing - NEVER EVER EQ RETURN NOISE!
Haven't been on the Bar in a while but I had to add my 2-cents!

I agree with "aclementsiv" on having a techinican FIX the noise problems at the lower frequencies in the return bandwidth and NOT adding equalization in order to attempt to make the noise floor look flatter.

There is a very big difference between frequency response "flattness" (peak-to-valley) and a clean noise floor. BOTH must be addressed in order for the HFC network to operate correctly and efficiently! Not doing so will only cause the system to not work to its full potential and slowly deterioriate causing modem issues.

The return input level to each hybrid should be as flat as possible to maintain a constant carrier-to-noise ratio for "maximum power transfer" as per design. This is done from the reference point in the headend through the fiber link (optimization) and to all RF amplifiers in the system. This can ONLY be performed with reliable SWEEP equipment to monitor the actual frequency response, diplex roll-off and any other connector defect that causes abnormalities in the RF path. Only after a confirmed "good" response is determined should the CORRECT equalizer be selected.

If there are "spikes" or "humps" that are in the bandwidth response observed on the sweep, these issues should be corrected and not "over equalized" in order to compensate for a problem. Fix the problems, do not hide them!

Going one step further, noise spikes can be seen in extreem cases on a sweep response but only a SPECTRUM ANALYZER can accurately locate and help correct the problem! I have seen for many years hundreds of technicians incorrectly "fixing" noise, high transmit, low transmit and no responder problems from both the in-house and contractor side of the business!

The most common and incorrect way that they "fix" the problems mentioned is by changing pads and equalizers because their test equipment gives them a better SNR reading than they had. Just like the Maytag repairman, parts get changed until it starts to work and the customer is back online! They never think that changing the pad or EQ just screwed up the entire street, block or node area because the main goal was to get the single troublecall customer working.

I am drifting from the initial EQ issue...sorry!

Getting back on track, noise at the bottom end of the return spectrum is unwanted. The noise is not constant and depending on may variables will have more amplitude on some days and less on others. The amplitude is measured in dBmV or actual signal level. This signal, unwanted signal, is input power! Yes, actual signal level that MUST be controled by eliminating or reducing it below the threshold that will cause harm to the other signals present on the spectrum. This problem can cause return optical laser transmitters to "clip" and generate beats and additional noise only seen in the optical link.

The use of a SPECTRUM ANALYZER and good SWEEP equipment is critical to the optimum efficiency of the HFC network. This includes TRAINING of the technicians and contractors who will be doing this work.

If someone "over equalizes" the noise hump in the amplifier it does a few things. First it causes the amplifier to be not "flat" as per design disrupting the "unity gain" concept. Second, it "hides" a problem. Third, as noise changes in level (as it does), the equalizer will need to be changed daily or hourly depending on the amount of garbage. Fourth, the entire design past the "over equalized" amplifier location will changed causing other modems to react accordingly. There are many more reasons but this is a quick list.

Combined nodes at the CMTS is another issue that can cause problems. If nodes are noisy and combined with clean ones, the modems on the clean nodes have just been contaminated with noise. This will cause high transmit and possible no responders as well

I apologize for the "soap box" but it is in my blood from the old days when I used to give training and seminars as I still do on an as needed basis. The biggest problem I have seen in the CATV industry over the past 37 years I have been in it is a lack of proper training. Many technicians and contractors guessing how to repair a problem only causes more problems. Not understanding how to use the test equipment is another big issue. Things like huge workload quota that cause poor workmanship to individuals who rush through work to get more production numbers only add to the madness!

Best 2017 to everyone!

-Greg Tobin

This member is a Regular Member.