DOCSIS 3 Cable Modems

The internet is being relied upon for many different functions. One thing that I am see more of is STL via the public network. There are many ways to accomplish this using Comrex Bric links, Barix units or simply a streaming computer.

We often can take for granted the infrastructure that keeps our connection to the public network running. Cable modems are very common as either primary or backup devices at transmitter sites, homes, offices, etc. The basic cable modem uses some type of DOCSIS (Data Over Cable Service Interface Specification) modulation scheme. This system breaks up the bandwidth on the coaxial cable into 6 MHz channels for downstream and upstream transmission. Generally, downstream transmission is 16 channels of 256-QAM signals. Upstream is 4 channels of QPSK or up to 64-QAM signals. Depending on your traffic shaping plan with the cable company, this will allow up to 608 Mbps down and 108 Mbps up. Those speeds also can change due to network congestion, which is the bane of coaxial cable based internet service.

The internet should now be considered a public utility. Especially after the COVID-19 emergency, distance learning, telecommuting at all the other changes we are experiencing. I know in the past, ISPs were reluctant to accept that role, as there are many responsibilities. That being said, when the public network goes down, many things grind to a halt.

Sometimes the problem is at the cable office or further upstream. Loss of a backbone switch, trunk fiber, or DOCSIS equipment will cause widespread outages which are beyond anything a field engineer can deal with.

Then there are the times when it is still working, but not working right. In that situation, there are several possible issues that could be creating a problem. A little information can go a long way to returning to normal operation. One thing that can be done with most newer cable modems, log into the modem itself and look at the signal strength on the downstream channels. Again, most cable modems will use 192.168.100.1 as their management IP address. The user name and password should be on the bottom of the modem. I also Googled my modem manufacture and model number and found mine that way.

Navigate around until you find a screen that looks like this:

DOCSIS 3.0 Downstream Channel Statistics

There is a lot of helpful information to look at. The first thing is the Pwr (dBmV) level. DOCSIS 3 modems are looking for -7 dBmV to +7 dBmV as the recommended signal level. They can deal with -8 to -10 dBmV / +8 to +10 dBmV as acceptable. -11 to -15 dBmV / +11 to + 15 dBmV is maximum and greater than -15/+15 dBmV is out of tolerance.

The next column to look at is the SNR (Signal to Noise Ratio). DOCSIS 3 needs to be greater more than 30 dB and preferably 33 dB or greater.

The last two columns are the codeword errors. This is a Forward Error Correction (FEC) system which verifies the received data and attempts to correct any corrupted bits. The lower the codeword error number, the better the data throughput. Codeword errors are often due to RF impairments and can be a strong indicator of cable or connector issues. Another possible cause is improper signal strength, which can be either too high or too low.

Upstream data is transmitted on 4 channels.

DOSSIS 3.0 Upstream Channels

The only statistic that is useful on the upstream channels is the Pwr, which should be between 40 and 50 dBmV.

I have found a few simple parts and tools can sometimes restore a faltering cable connection. First, I have several attenuator pads; 3dB, 6dB and 10 dB with type F connectors. This has actually cured an issue where the downstream signal was too hot causing codeword errors. Next, some good Ideal weather proof crimp on F connectors for RG-6 coax and a good tool should also be in the tool kit. I have had to replace mouse chewed RG-6 from the outside cable drop into the transmitter building. Fortunately, there was some spare RG-6 in the transmitter room.

If these attempts do not fix the issue, then of course, be prepared to waste a day waiting for the cable company to show up.

Installing a WISP on an AM broadcast tower

This is an interesting project currently underway at one of our client’s AM sites.  They have decided to go all in and create a WISP (Wireless Internet Service Provider) for the community around the AM tower.  I thought it would be interesting to explore this topic, as there are not many opportunities for AM towers to lease vertical real estate.

First a few basic ideas.  For an AM broadcaster, (aka medium wave or standard broadcast band) the entire tower is part of the transmitting antenna.  There are two types of towers; series excited and shunt excited.  A series excited tower has a base insulator, like this:

AM tower with base insulator
AM tower with base insulator

A shunt tower usually has a series of wires called a skirt, separated from the tower by standoffs, which go to the top of the tower or nearly to the top of the tower. The base of the tower is grounded, like this:

AM tower with out base insulator
AM tower with out base insulator

A shunt excited tower has distinctive advantages for co-location opportunities in that the tower itself is grounded, greatly simplifying placing additional antennas on the towers.  That is not to say that antennas can not be installed on series excited (insulated) towers, it just requires an extra step of using isolation coils.

In all cases, the tower should have a structural study done to insure that the additional antennas do not overload the tower and cause structural damage or collapse.

In this case, the tower is new and was designed for the extra load.

The plan is to create a sectorized wireless internet system using four 90 degree panels, each with three access points.  A tower mounted sixteen port switch is mounted behind the panel antennas and the switch communicates with the ground mounted router through two fiber optic cables.  A 54 volt DC supply powers the switch, access points and point to point radios mounted on the tower.  There are two fiber runs, one is for subscriber traffic and the other is for radio management.  This system is using Ubiquiti gear.

Ubiquiti 90 degree sector antennas and radios
Ubiquiti 90 degree sector antennas and radios

A word or two about Ubiquiti gear.  Ubiquiti specializes in cheap equipment manufactured in China.  That is a double edged sword.  On the plus side, if anything breaks or gets damaged by lightning or whatever; throw it out and install a new one.  On the negative side, I  have seen Ubiquiti gear do some strange things, particularly after a firmware upgrade.  The newer stuff seems to be better than the older stuff.  All that being said, as this is a brand new operation and seems to be a proof of concept, then the Ubiquiti gear will be fine to start with.

Going up
Going up

The tower crew made quick work of installing the sectorized access points.

Tower crew waiting for equipment lift
Tower crew waiting for equipment lift

Going up the face of the tower, there are the aforementioned fiber cables, the 54 VDC power cable and one backup Ethernet cable.  All of the Ethernet jumper cables used to connect the access points to the switch are UV rated, shielded Cat 5e and use shielded connectors.  This is very important on a hot AM tower.  Due to the skin effect, the shield on the shielded cable protects the interior twisted pair conductors from the high AM RF fields present on the tower.

Transtector LPU 1101-1158
Transtector LPU 1101-1158 Ethernet cable protection unit

At the base of the tower, the DC power cable and the Ethernet cable go though high quality lightning protection units.  These are Transtector 1101-1158 Ethernet and 1101-1025 48 volt outdoor DC power units.  Even though the DC power supply is 54 volts, the 48 volt LPU’s will function adequately.  The TVSS devices used in the LPU circuit are rated for 88 volts maximum continuous voltage.

Transtector 1101-1025 48 VDC lightning protection unit
Transtector 1101-1025 48 VDC lightning protection unit

In addition, I made a service loop on the DC cable with also creates an RF choke.  Several (12-14) turns of cable 18-20 inches (45 to 50 cm) in diameter act to keep the induced RF at the input terminals of the LPU low so the protection devices do not fire on high modulation peaks.  This also helps to keep the AM RF out of the 54 VDC power supply in the rack.

Making ethernet jumper cables, TIA/EIA-568B
Making ethernet jumper cables, TIA/EIA-568B

The backup Ethernet cable has a similar setup.  Regarding the Ethernet cable and induced RF, this station runs 1 KW.  As long as the shielded RJ-45 connectors are applied properly and the tower mounted switch is grounded along with the LPU, then all of the RF should be on the very outside of the cable shield (due to the skin effect).

Base of AM tower with WISP equipment installed
Base of AM tower with WISP equipment installed

This principal also applies to lightning strikes.  Although lightning is DC voltage, it has a very fast rise time, which makes it behave like AC on the initial impulse of the strike. The voltage induced on the shield of the cable will not effect the twisted pairs found deeper within the Ethernet cable.  Of course, all bets are off if there is a direct strike on a piece of equipment.

AM stations running powers more than 1 KW, Superior Essex makes armored shielded cable called BBDG (the new trade name is EnduraGain OSP).  This cable comes with a heliax like copper shield with an optional aluminum spiral armor.  This cable looks very robust.

Enduragain OSP armored shielded Category cable
Enduragain OSP armored shielded Category cable

On series excited towers (those with an insulated base) fiber optic cable can be used to cross the base insulator without any problems, as long as there is not any metal in the cable (armor or aerial messenger).

LBA Group TC-300 tower lighting choke
LBA Group TC-300 tower lighting choke. 180 turns #12 AWG enamel wire on 6 inch coil form.

DC power can cross the base insulator using something called a “Tower Lighting Choke.”  This device is a set of coils wound around a form which passes the DC power but keeps the AM RF from following the DC power cable to ground.  These work relatively well, however, lightning protection units still need to be installed before the DC power supply.

BE AM tuning network

Occasional reader Scott asked for a picture of the inside of a BE AM output tuning network.  I figured it might be helpful to make a short post about it.

These things are pretty simple; a T network with a capacitive leg to ground.

BE AM Output tuning network
BE AM Output tuning network

This particular unit is for 1230 KHz.  I believe the capacitor is frequency determined and they may also use larger inductors for lower frequencies.

BE AM output tuning network schematic
BE AM output tuning network schematic

The inductors are Kintronic LV-15-20 (15uH 20 amp) and the capacitor is 0.0018 uF CDE 6KV 5.6 amp.

The issue with this particular unit is dirt.  The inductors have round metal plates that roll along the inductor coil to make the variable inductor tap.  Dirt has accumulated on the coil turns and on the inside of the plates.  This, in turn, causes arcing anytime the Tune or Load controls are moved.  A through cleaning should take care of the problem.

More AM retuning work

Working on another old AM station, this one is a simple Class C one tower on 1230 KHz.

Broadcast Electronics AM Output Tuning Network
Broadcast Electronics AM Output Tuning Network

The main problem today was this BE AM output network unit between the BE AM1A and the ATU.  This site has had some dirt difficulties over the years and the internal parts of this tuning unit arc at full power.  I attempted to drive the ATU directly with the transmitter, which was a no-go.

Gates Radio 1 KW AM ATU, circa 1947
Gates Radio 1 KW AM ATU, circa 1947

I took a look at the ATU, which is a pretty standard Gates 1 KW ATU from the late forties or early fifties.  I have seen perhaps dozens of these things.

My first thought was that over the years, likely due to changes in the ground system, the base impedance has shifted away from its licensed values.  However, a quick measurement of the base impedance shows it to be exactly at the licensed value, 17.3 ohms.  The tower is 67 degrees tall, so that impedance value is right in the theoretical norm.

I  measured the input to the ATU, which showed 38 ohms with about 7 ohms capacitive reactance.  I can only surmise that it has always been this way.  The transmitter in use before the BE AM1A was a Harris/Gates Radio BC-1G.  That model transmitter will drive anything including an open transmission line.

Retuned ATU input; 49 ohms resistive, 0 ohms reactance
Retuned ATU input; 49 ohms resistive, 0 ohms reactance

Having the bridge on hand, I decided to retune the ATU for a better match. I put the bridge on the input terminals of the ATU and set it to 50 j0.  Using the remote control, I turned the transmitter off and on while making small adjustments to the output strap on the coil until the resistance was 49 ohms with zero reactance.  I would have gotten it to 50 ohms, but the strap on the output side of the coil would not stretch far enough to reach the proper spot on the coil.

Now the transmitter will run into the ATU directly at full power with about three watts reflected.   The BE AM output matching network unit has been removed for cleaning and repairs.  I will reinstall it once those repairs are completed.