I remember when I was very young, peering outside the window and seeing a bunch of workmen and trucks pull up in front of the house. In a short time, they rigged a steel wire between the power poles, then pulled up a thick insulated cable, and used some rotating machinery to wrap a wire to bind the cable to the support cable. Not before long, I would hear on the news that this was the roll-out of the hybrid fibre-coax network, known as HFC for short, that was to carry pay TV to consumers. This was around the mid-90s, and there was a frantic competition between Telstra and Optus to build the largest HFC network.
Not before long, I remember there were consumer complaints about the large number of unsightly cables in the air and damage to trees which had holes cut through them to let the wires pass through. In the end, I believe it was agreed that Optus and Telstra would not overbuild each others’ coverage footprint. With that, it seemed that HFC technology was to be relegated as a technology for the privileged few. In a way, just like how Galaxy TV (a pay TV microwave broadcaster) was for the privileged few as well.
I never had any pay TV service myself, so things were simple and unchanged. The phone ran from copper lines from Telstra (although resold as Optus). Internet, which came later, was via dial-up access. Nothing complicated.
But when I passed by some of my friend’s houses, I noticed something high up on their wall near the eaves. We didn’t have one of these – it looks a bit like a kick-board from swim school but had cables running in and out of the bottom. I took the time to ask and the answers I got ranged from “I don’t know”, to a less-than-satisfactory answer of “that’s our telephone line”. For years, I just assumed it was an oversized demarcation box, but then I noticed some units had a second square grey box next to it … that could also have been a demarcation box. In which case, why two boxes?
This mystery has been in the back of my mind since I entered primary school, and was not resolved for me until just recently. Just the other week, our local area council had its final scheduled clean-up, which meant one last chance to go salvaging. To my luck, I found one of these units sitting kerbside. It was quite covered in dust, grout, spiderwebs and even came complete with some live beetles inside. I took it home, cleaned it up, and that’s what is pictured above.
Up close we could see the Motorola branding, so it was probably going to be a piece of radio equipment. Aha, HFC is RF based. On the rear we see it’s an SII/CAU 1LN 65MHZ with a model number of MM1012A. This unit appears to be refurbished September 2003, which isn’t that long ago. The unit was secured by security Torx which took a lot of work to remove, which I did, to remove the scraps of cut wire and clean out the wildlife.
With that information alone, I didn’t find too much information on it. I couldn’t even find a picture of it online, even when every third house down some streets around here still have them hanging on their wall. I was still left somewhat in doubt as to what it was – so only once I opened it up did I fully realize what this unit was.
HFC Cable Telephony
After a bit of looking around, I resolved the mystery box to be a Motorola CableComm Series II Cable Access Unit. The product code seems to indicate it is a one-line telephony interface, with an operating frequency of 65Mhz (likely a 6Mhz wide cable channel). But lets take a step back and recognize this for what it is.
In the early 1990s, competition in the telephony market was difficult. Telecom Australia (later Telstra) owned the copper line infrastructure which every reseller used. It was hard to offer more competitive prices when your minimum costs are set by your rival. Optus pinned its hope on HFC to break this deadlock.
Over in the USA, where HFC is the primary telecommunications media in some areas, the desire to perform “triple-play” services over HFC resulted in the production of cable modems of various proprietary protocols. Systems designed to provide a voice port over HFC included Motorola’s CableComm, Tellabs CableSpan 2300, Unisys DCSS, ADC HomeWorx, General Instrument’s Mediaspan, Scientific-Atlanta’s CoAxiom and Arris Cornerstone.
Motorola was a late entrant into this rather crowded market, but using their expertise in trunked radio systems, they announced the CableComm system in 1994. Early trials were named with TCI and Teleport Communications Group in Arlington Heights, Illinois with some tests between the houses of 25 selected employees. Full-scale commercial roll-out was planned for the first quarter of 1996 with a per-line cost of US$350 to US$550. The system is claimed to use TDMA, allowing between 500-1000 calls in a 6Mhz channel.
In November 1996, Coherent Communications Systems provided an echo canceller solution for Motorola CableComm, which appears to operate in the digital domain. This article seems to suggest that aggressive expansion would commence “during the remainder of 1996.” This suggests that things took a little longer to get off the ground than expected.
But Motorola definitely did bet big on it at chipping away the grip that phone companies had on telephony – in the article by Chicago Business, the boxes are described as “looking like paper towel dispensers”. Apparently, Motorola rented the Lake Zurch factory for four years to produce these devices.
In early 1996, Optus chose the CableComm system to provide its competing telephone service. Even in 2002, they proudly claimed to have secured a long term supply of this equipment. Note that they misspelled CableComm in the press release! A search for CableComm online shows that some people were locally employed to keep these units going past their discontinuance by Motorola.
However, as with all technology, it has a lifetime. The time was up for Motorola CableComm even before 2008 had hit, as an expert report from Michael G. Harris of Harris Communications Consulting LLC points out. In this report, Optus is criticised for severe underinvestment in their HFC network with oversubscription. They are also criticised for relying on Motorola CableComm – a discontinued proprietary system for which a multi-dwelling unit compatible telephony adapter was never developed. The older circuit switched nature of the system was also not contemporary with the technology used in the US which had or was in the process of being replaced with packet switched solutions. The report seems to point to Optus’ own lack of confidence in their own network – preferring not to hook up customers where other providers would have done so, and avoiding investment in the belief that HFC is an inferior technology. This apparently stemmed from Telstra’s disagreement with Optus in regards to how customers were not connected to their own networks even when it was available, instead opting to use Telstra’s copper network instead.
In fact, as CableComm itself is a proprietary standard used by their early internet cable modems, it was soon supplanted by the Data Over Cable Service Interface Specification (DOCSIS) industry standard which involved a large number of industry players. As IP-converged technologies became more popular, services were run over the “internet” data connection rather than occupying its own discrete channel spectrum, hence the move from circuit switched to packet switched networking.
In 2011, NBN Co cut a deal with Optus to shutter its HFC network and noted it may use sections of it to run the NBN network. By 2012, Optus had publicly announced it will be turning off its HFC network in 2014 with complete dismantling by 2018. Customers would be migrated to NBN. Unfortunately, and not unexpectedly in light of the previous critiques, the network was not up to scratch, with underinvestment meaning having equipment reaching end-of-life and a rather complicated network architecture. To ensure quality service would have required overbuilding. In the end, it appears they abandoned the cable instead, so the whole network had a lifetime of about 20 years. Despite the NBN choosing a mixed technology mix which should have accelerated the deployment of the NBN and reduced the cost, the use of HFC seems to have caused its own headaches.
In researching this, I have now come to understand a lot more about HFC telephony. What I discovered was somewhat fascinating to me, since it seems that some customers may not have had cable TV, and the HFC was solely used to provide telephony service to customers.
Unfortunately, as I was not ever served by the CableComm system, I don’t know what it’s characteristics were. Did it really provide carrier-grade service indistinguishable from a POTS line? Did data services work reliably? Even V.90? I have no idea. But it’s important to recognize that in some way, these “early” HFC telephony services were a first step towards competing against traditional telecommunications companies and were the “step” in-between POTS and full-blown packet-switched VoIP (whether through a softphone or ATA) or even VoLTE.
That being said, as cable is a shared medium, security would have been a potential issue. While it is already presumed to be a fully digital codec system, did it utilize encryption? Was there the potential for calls to be “eavesdropped on” by others served by the same node? Was there any encryption on system metadata? Were there actual vulnerabilities in the system which might allow others to place calls on other subscriber’s accounts? In modern SIP VoIP telephony, we made a big step backward, with most calls going out in the clear using G.711/729a. Now that we have a security focus, maybe such old systems would not mass muster anymore.
Even on the RF side, it would be fascinating to know – did it really operate at 65Mhz on the cable x 6Mhz bandwidth? How was the TDMA and return channel achieved? What was the physical layer coding? Was FEC involved? Is it still operating for the last few people that might need it before NBN comes in?
To be honest, circuit switched technology may be outmoded, but it does have a more consistent guaranteed quality as compared to packet switched technology which can be affected by latency, jitter, packet loss and contention which established circuit switched calls do not have to worry about. When it comes to running voice-band modem data services, circuit switched technology generally works better.
Unfortunately, information seems to be scant. It’s probably moot anyway, but it’s always intriguing to know. Unfortunately as I don’t have any equipment to use this unit with, it’s not something which I can determine.
However, one disadvantage is already apparent – as it’s not a traditional copper line, you can’t get ADSL2+ unless you reconnect a traditional copper line. With a “CableComm” provided line, you can only get cable internet at the price Optus charges you. Cable internet competition is pretty low compared to ADSL, thus it can be more expensive as a result. Maybe this was “built in inertia” to stop bundled Optus customers from going back to a copper-line. At least their newer Optus Cable services use modems with integrated VoIP packet-switched service instead.
We saw the unit front and back earlier on, so continuing on, we see there is a mounting hole at the top.
At the bottom, there are three cable entry positions, with two spring-loaded security Torx screws securing the covers. The covers can be slid down to reveal the connection terminal area. There is a recessed area next to the “Manufactured by Motorola” text which may have been used by service providers to apply their branding – but I’ve never seen it used around here.
The larger outer cover uncovers the parts which are “customer responsibility” – namely their own internal telephone wiring pair and an F-connector for the cable “loop-out”. This is terminated with a 75-ohm terminator, suggesting this subscriber may not have had a pay TV or cable internet service. The cable entry boot was inverted in the photo after I cleaned it.
A test jack is supplied to allow testing of the line output from the CAU without interrupting the wiring.
The wiring attaches into a plastic block which is snapped down into IDC fingers.
Removing the service provider cover reveals the F-connector for the incoming HFC feeder line. This is where power is derived, up to 100V. A grounding bar is also provided. The connection on the left appears to provide an alternative power connection, a covered diagnostic port and a socket which may allow for a failover service.
The port is similar to a LAN 8p8c connector.
The IDC blocks and test jacks themselves are a module that can be removed. The part number is 0104080X10, with this unit being Revision B, made Week 44 of 1999. Vendor is 912, Made in the USA.
On the inside, the CAU is further identified with its MSN/ESN. Apparently it was Assembled in Singapore. Given the numbers, I wonder if it is somehow related to GSM just running over coax, in which case encryption is part of the standard. Maybe it also runs voice compression?
The top cover is surrounded by a rubber gasket and is supposed to stay nice and “dry” as that’s where the brains are. This board showed only very slight signs of corrosion at some points. We’ll take a more detailed look at the board once it’s freed from the base.
The line interface module, however, once removed was found to be potted entirely, and thus not serviceable. Is it just a bunch of wires? Or are there some surge protective elements inside?
In the interests of complete disassembly, I removed the screw holding the can over the radio section of the CAU. Internally, part numbers of 2604550X03/2604550X04 are printed, along with Issue A Vendor 912. Date is coded as 2228 and 1758 respectively. The cans were Made in Mexico.
The top side of the board is where most of the action happens. We will break it down into sections, but overall, the board has a code of 8404629X01 Issue-0. Printed in white is 21399-03-8 with a date of 12th March 1998. It clearly identifies as Made in USA, Motorola CableComm and CAU Series 2. The barcoded label in the centre has the text “MMLN5061B 606VLW 01/11/99 MM1012A” which seems to suggest a factory alignment date of 11th January 1999 along with the product code of MM1012A. The board is a fairly complex six layer PCB. It definitely looks like Motorola quality at a glance.
Starting in the lower-left side, we can see two symmetrical line interfaces. The left side is Line 1 and is populated, with Line 2 unpopulated as this is a single line unit. Lines are protected by fuses on both tip and ring. The line interface is handled by an Ericsson PBL 387 10/1 SLIC. This takes the place of a traditional hybrid transformer. The inductors may be responsible for ring voltage generation. The resultant audio is handled by a Motorola MC14LC5480DW PCM CODEC-Filter. A row of test pads is bought out as well. The fingers at the bottom connect to the potted module to make connections to outside lines.
The right half of the image is the power conditioning portion of the unit consisting of an assortment of solid electrolytic capacitors (good choice), inductors and transformers. A 2A fuse appears to protect the unit. The power is controlled by a TOP224G with feedback via an MOC8106 optoisolator.
The top left corner has a large Motorola 5104545X01 IC, which may be an ASIC, along with a 4.608Mhz crystal. The firmware is stored on a pair of flash chips.
The left one has a Motorola label with 5170806C07 120398 printed on it. Assuming they’re both identical, the chips are AMD AM29F200AB 2Mbit Flash chips, thus the unit contains at most 512kiB of firmware. Complementing this is two Samsung Electronic Corporation KM62256DLGI-7 32k x 8 SRAM chips for a total of 64kiB RAM. A Motorola XC68LC302CPU20B 50Mhz 68k CPU is seen in the top corner. Interestingly, the CPU is pretty beefy – a relative to the 68k CPUs that powered whole Macintosh LC desktop computers. The top corner has some edge contact fingers probably used for factory alignment use.
The radio section has some Motorola custom ICs, a filter, a few capacitors, an adjustable oscillator and a bypass relay (RK1-L2-4.5V-H33). The filter networks are pretty elegant to behold, but way above my ability to design or comprehend.
The underside of the board has fewer components in comparison.
The radio section predominantly features small surface mount components. U203 is an RF MicroDevices RF2317 Linear CATV Amplifier (DC to 3Ghz). Aside from that, the other components are not very notable, with the exception of C441 which appears to be a discoloured tantalum surface mount capacitor. This unit may have suffered some damage at some point and may not be functional (e.g. surge? reverse polarity?).
The remainder is not particularly notable. But at least, now I know what the unit is for, and what’s inside.
Optus’ HFC Network
Prior to all of this, on a previous salvage walk, I decided to look at the HFC cables in my neighbourhood just out of interest. I had an expectation that Optus’ HFC network would not be usable at the time just based on observations of the equipment in the air – but since we’re discussing HFC, I might as well put up some images. I’ll first say that I’m not an expert in HFC technology by any means, nor am I familiar with the equipment used, but I present my observations even if incorrect.
At this time, I’m living in a townhouse complex with something like 28 units. Our HFC feed is this cable that crosses the road. It’s not well bound to its support wire, thus the snaky appearance, but it has just a two-way splitter at the end. No way will it serve the “population” especially if all users are connected – such equipment has to be re-planned/replaced.
The splitter itself is Scientific Atlanta branded. This brand had been acquired by Cisco in 2005. As this the pre-acquisition logo, the equipment is at least 12 years old now. At a good estimation, such equipment has useful lives of about 15 years (roughly speaking), so I suspect equipment replacement is not unexpected.
The nearest trunk amplifier seems to be this unit, with the incoming trunk passing through a Regal combiner with surge protection. It definitely looks dated, but the cabling itself is slightly confusing. It seems there are two trunks coming in on the left, with the combiner used to split the signal so the trunk continues through to the right along with other trunk, and one output from the amplifier making the three cables on the right. But two of the trunk cables have an end-to-end join under shrink-wrap. Maybe it’s the opposite direction, but I’m not familiar enough with the equipment to know.
That being said, the physical condition of the aerial cabling “needs attention” – the fine wrapping wire that secures the lines to the support wire has come free and “unwrapped” along a section.
The support wire is held to the poles with simple hardware – interestingly, in our area, there is support wiring with no HFC cables bound to it. It may have been initially planned to be rolled out, but then due to a lack of potential customers, abandoned.
Not knowing much about the specification of the cables used, I noticed that some of the cables in the area seem to be a darker colour, whereas others are more “grey”. Maybe it’s related to sun exposure, or different batches of cables, but here’s another set of joints with some “cable tie” bundling.
Sometimes, along the road, we will see such boxes and sometimes they even hum loudly to signal their presence. I don’t think this is a HFC node, but more likely a power injector as it has a mains input, and its output goes up the pole into a coupler.
The couplers/splitters in question all seem to be Regal branded in my area. The marked frequency response is 5-1000Mhz, and the model is RDLC10-12S290V.
The splitters in the area also come in some other varieties – in the case of larger splits, it seems that it is a powered splitter, but where the two core power comes from is not obvious to me. It’s not just mains is it?
We can see the varying penetration levels, but rarely are all existing ports taken, as the popularity of pay TV is relatively limited. The second splitter seems to have one port where a former customer’s drop was merely “cut off”. The improperly terminated F-connector port may well result in signal reflections and distortions, or noise ingress into the cable network as well as potential damage to the splitter from having water enter through the stub of cabling, helped by capillary action of the coax braid.
There are un-powered variants as well, which seem more “expected”.
This is an anchor point that’s attached to the support wire to secure the drop to the customer.
As for amplifiers, around our area, there’s the occasional “flat” one that looks like this. From what I can gather, this appears to be a line extension amplifier, to help push the signal further down the line. I’m not sure the customers connected at the end of this would enjoy the best signal-to-noise ratios.
The other amplifiers look as above, with a varying port position configurations and number of cables leading in and out. Just looking at the PDF about related units from Cisco’s website seems to show that they’re pretty complex.
While it does seem somewhat complicated, this is relatively anemic compared to the complexity of the networks I witnessed overseas in South Korea and Taiwan where HFC is one of their core technologies for delivering high speed broadband. I might comment more on this once I get some holiday photos up – but needless to say, HFC outages and service trucks were constantly attending to the network in South Korea, and customers were being migrated to FTTH services, so the lifetime of HFC is limited at best.
At long last, a question I had from childhood has been resolved – namely, what’s that box on the wall? In examining this, I learned a lot about the history of telephony over HFC in terms of the alternative technologies available, the hopes of the companies, and its ultimate fate.
Looking at the Motorola CableComm CAU, I’ve managed to get some good shots of its internals which I couldn’t find online, so that’s a pretty unique opportunity in my eyes. While I’m not able to ascertain the operational status of the unit, or run it, it was still good to know what was used internally.
It also gave me a chance to share some of the photos of the Optus Vision Cable HFC network equipment in my area. In the end, it seems I’m still left waiting for NBN, but it looks like it will be VDSL2 when it arrives … late and underwhelming.