Application Note
WEBSITE:

www.jdsu.com
Using JDSU Equipment to Test and Troubleshoot CPD,
Impulse Noise, and Ingress in the Return Path
History of CPD
Common Path Distortion (CPD) is created by non-linear mixing from a diode junction created by corrosion
and dissimilar metal contacts. Its not just dissimilar metals, but dissimilar metal groups. There are 4 main
groups of metals:
1. Magnesium and its alloys,
2. Cadmium, Zinc, Aluminum and its alloys,
3. Iron, Lead, Tin, & alloys (except stainless steel), and
4. Copper, Chromium, Nickel, Silver, Gold, Platinum, Titanium, Cobalt, Stainless Steel, and Graphite.
CPD is second and third order intermods from the forward channels intermixing and creating distortions,
which fall everywhere. CPD will make CSO/CTB worse for forward performance.
Separation depends on forward channel plan. NCTA, HRC, and IRC plans that use NTSC, 6 MHz spacing
will have beats every 6 MHz. PAL could be every 7 or 8 MHz.
The original culprit was the old feed-through connectors. Dissimilar metals from the copper clad, aluminum center
conductor and the stainless steel seizure screw. Application Note: Return Path Troubleshooting
2
Housing terminators are notorious now because of the
higher levels to mix and intermodulate, not to mention
a few bad varieties that were manufactured.
Colder weather makes CPD worse because the diode
works better. Electron funneling is better with heat
so there isnt as much non-linear mixing. Because of
contraction and expansion, CPD could become worse
with heat.
There is another impairment that manifests itself like
CPD, but the separation is a little dierent; it is called
transient hum modulation. An RF choke can saturate
with too much current draw and cause the ferrite
material to break down. The same thing can happen
in customer installed passives unless they have voltage
blocking capacitors installed.
Troubleshooting CPD
Pull a forward pad to see if the return cleans-up. This
is denitely CPD, but very intrusive when doing this
and may disrupt CPD temporarily.
Try not to disturb anything in this tracking process.
Vibrations and movement can temporarily break
away the diode/corrosion causing this CPD.
Voltage surges can also destroy the diode. At least long
enough to warrant a return visit!
The test point locations will determine the outcome.
If CPD is on any of the downstream output TPs of an
amplier, it may be the output seizure screw or con-
nector. Otherwise, continue down that leg. Look for
housing terminators.
If CPD is on the Fwd input TP and not on the output
TP, it may be the input seizure screw or connector. The
reverse amplier provides isolation that prevents CPD
from appearing on the output if created on the input.
It could still be downstream though, because the levels
on the reverse input test point may be too low to see,
which may warrant a pre-amp. Otherwise, attach to
the reverse output and terminate reverse input pads
one at a time to determine the oending reverse input
leg.
If you view the reverse spectrum from a bi-directional
test point with an analyzer, you could overdrive the
front-end of the analyzer with too much forward
path signal and cause intermodulation within the test
equipment. To see the reverse ingress, the instrument
is in its most sensitive mode. Both forward and reverse
signals are going directly into the mixer input. The
high level forward channels will cause intermodulation
products in the front-end of the meter. This will hap-
pen on any type of analyzer.
Use a low pass lter to block all the forward channels.
You could use a diplex lter, but its cumbersome. The
insertion loss may not be calibrated, and it may not be
dc blocked.
This is why newer units have a built-in, switchable,
lowpass lter to block out the forward channels.
It may be advantageous to troubleshoot CPD from the
end-of-line back toward the node. This will eliminate
disturbing the fault until you get there.
Note: Be sure forward input levels to the Stealth headend
transmitter (Tx) are between 4 and 12 dBmV. If levels
are too high, distortions will be created in the Tx,
which appear as CPD when viewing the Noise
mode.
Tracking Down Ingress
The first step is to verify it is truly on your network and not
self-induced. Use some type of spectrum analyzer to view
the anomaly. Cross reference with frequency charts that
identify different ingress sources to get a best-guess idea.
Noise and transient ingress above the diplex filter region
is probably laser clipping or induced at the node. You may
also want to view the frequencies below 5 MHz to verify
its clean. Noise below 5 MHz could still affect the lasers
dynamic range.
Listening to Ingress for Identification of the
Source
The second step is to demodulate the ingress, if possible, to
identify the type of ingress. Reverse path ingress is usually
amplitude modulated (AM), but could also be FM. Listen-
ing to the ingress helps to identify the source.
FM demod for the audio of forward channels and
certain shortwave radio.
AM demod for most reverse interference and ingress,
such as CB, Ham, and shortwave radio.
This may give you some insight into the location of the
source or at least the nature of the source. You may be
able to get the call signs of a ham radio operator
or a mile marker from a truck driver using his CB.
This could aid in pinpointing the ingress location. Application Note: Return Path Troubleshooting
3
A single source of interference is easy to track down. If its
constant, just use the divide and conquer theory to dissect
the system. Observing how it reacts and changes could indi-
cate different sources such as a trucker or home user. A CB
level changing quickly or slowly could indicate this source
quickly.
Multiple ingress sources, bursty noise, and electrical tran-
sient noise are a totally different story and are very difficult
to pinpoint. Remember that the lower value taps contrib-
ute more noise and ingress than the higher value taps. The
lower attenuation from tap values of 14 and below coupled
with the low attenuation in the cable at lower frequencies
creates an easy path for noise to funnel back.
Return Path Power Addition
Many people dont fully understand power addition and
become discouraged when trying to perform noise miti-
gation. A little decrease could be more than you think.
Understanding power loading for return path ingress is
essential to help aid in troubleshooting.
For example: I observe a CB signal at 27 MHz in the
headend at 20 dBmV total power. I disconnect one leg
by removing a reverse input pad and the level drops to
18.8 dBmV. I disconnect the second leg and the level drops
to 17 dBmV. After disconnecting the third leg, the total
power drops to 14 dBmV. After disconnecting the last leg,
the ingress at 27 MHz is eliminated. So the question that
remains is, which leg has the largest amount of ingress?
The answer is none. All four legs of the node are funneling
equal amounts of noise to the headend of 14 dBmV! Two
14s equal 17. Three 14s equal approximately 18.8 and four
14s equal 20 dBmV. Remember, every doubling of power is
3 dB.
Test Location Considerations
Because the return path signals are low in level, it may
be warranted to use a preamp.
The preamp is used to raise the signal above the noise
oor of the test equipment. This is especially a problem
on the return signals that are read from high loss test
points.
The newer units have a preamp built-in and compen-
sate all measurements accordingly.
If a problem is observed at the output seizure screw of a
tap, continue on.
Some new probes from SignalVision and Gilbert create
a good ground and quick connect.
Note: One caveat to this is a probe will always be bi-direc-
tional and will cause an impedance mismatch itself.
This is something to keep in mind when troubleshoot-
ing. Sometimes an in-line pad can be attached to
decrease the amount of energy tested, which in turn,
may create a better match. Be careful when probing
seizure screws, though. The AC present will harm in-
line pads and certain test equipment. The equipment
is AC blocked for ~ 100 Vac.
Start with 14 dB taps and lower. If the problem is at the
input of the tap and not the output, then the problem
is from one of the drops or farther upstream possibly
from a cracked cable before the next amplier.
Look at one drop at a time to determine the biggest
contributor.
Noise Readings
Be careful with spectrum analyzer, noise level readings.
2 dB/div is a good scale for sweeping and 5 or 10 dB/
div is best for the spectrum mode.
The level displayed is based on the RBW setting and
will be very dierent from one setting to another.
A -20 dBmV noise oor with 30 kHz RBW is really
1.2 dBmV in a 4 MHz bandwidth and theres usually a
correction factor associated with it.
Note: The Spectrum mode is not the same as a true spec-
trum analyzer. The RBW is set at 280 kHz and a VBW >
1 MHz. This is optimized for analog carriers and burst
noise measurements. It has a peak noise detector so
the noise reading may be significantly higher than a
normal spectrum analyzer with the same RBW set-
ting.
A pad on the analyzer will lower the level as well. At-
tenuation and gain aect noise and carriers equally.
Measurements with no point of reference are very
misleading. If theres a reference carrier present, you
can make a relative measurement, such as desired-to-
undesired ratio (D/U). One fault with this, though, is
RBW settings aect noise and continuous wave (CW)
carriers dierently. A CW carrier is