Friday, 2 January 2015

Clarke's Third Law.

Happy New Year All!

Having been enjoying the long Christmas shutdown of the company I work for, I've been getting out and about, a few days ago I took a couple of busses, had a good walk, and got myself up to Winter Hill. It coalesced some thoughts I've had for years.

In the North of England north of Manchester on the western fringe of the Pennines, Winter hill has a TV transmitter mast and a variety of other telecommunications masts. It has long been a hub for radio communications due to its height overlooking Manchester, Liverpool and Lancashire, all major regions with high populations.

One of the oldest masts, perhaps the oldest, is the BT mast (formerly General Post Office), which is seen above partially obscured by the massive guyed TV mast. This was part of Backbone, a 1950s microwave relay system designed to survive nuclear war by avoiding major cities and military targets. Close examination of the above photo shows that the BT mast looks odd, a more pyramidal form than the adjacent more modern communications masts. This can be seen in a shot of the BT mast.

One reason for this shape might be greater resilience to blast. However when I visited Winter Hill back in the 1990s this mast had horn antennae with right angled parabolic reflectors, similar to the antenna used by Bell Labs engineers to find the cosmic microwave background. In this image the horn antenna is facing the ground. The large, heavy, horn antennae would have needed a very robust structure to mount them.

Also shown in the photo below is a mast that used to be owned by Mercury Communications, former Home Office masts are to the centre and right, the former Mercury mast is to the left. the Home Office being the UK's government department that dealt with internal national affairs. They had their own engineers and maintained the radio networks for Civil Defence and the Emergency Services. One of those former Home Office sites is still used by the Emergency Services (it has a fixed GPS receiver used in differential GPS).

What is apparent from these masts is the concentration of cylindrical antennae on the masts, these are called drums and are used for microwave point to point fixed links. In general, compared to masts at other locations I've seen, there isn't much in the way of antennae designed for wide area broadcast. Including the TV mast there are 8 masts, of which seven (including the TV mast) are massive nodes of point to point microwave radio links, I guess just under 50, at one site in one part of the UK.

Before I continue, I should put some frequencies into perspective. Medium Wave Broadcast is about 1MHz, FM broadcast radio is around 100MHz, Cellphones work in bands around 900MHz, 1800MHz, 1900MHz and 2100MHz, WiFi and microwave ovens operate around 2400MHz. The link frequencies used by these drums are often of the order of 20,000MHz and 30,000MHz. However since I was working on such systems I understand that bands over 60,000MHz have come into play.

In the following shot of another mast it can be appreciated why the point to point antennae are called drums.

Bottom centre is a drum antenna that has lost its cover, looking closely it is possible to see the feed horn antenna inside the drum. The construction is a drum housing a parabolic dish with a plastic sheet stretched over the front to protect the feed horn from the elements.

I've also pointed out two other types of antenna. There is a dipole (elongated loop), with a reflector behind it, this is a wide area antenna designed to broadcast a VHF (very high frequency, in this case between 100 and 200MHz) signal to mobile units in the region towards which the dipole/reflector set points. Without the reflector a dipole would send the signal all around (360 degrees) in the horizontal plane, the addition of a reflector restricts the area of coverage to around 180 degrees, in this case pointing away from the camera taking the shot. Strictly speaking this is a 2 element yagi. Note, I am being deliberately vague about frequency here!

The UHF (ultra high frequency) colinear array uses dipoles, but they're much smaller than in the dipole reflector set up described above. That's because a dipole antenna is half a wavelength from tip to tip. With the speed of light about 300M m/s at 100MHz a half wave dipole is 1.5m from tip to tip, at 300MHz the dipole would be just 50cm from tip to tip. In this case the UHF colinear array operates between 400 and 500MHz. It's termed a colinear because the dipoles are stacked on top of each other, in a similar way to the dipole and reflector described above this is intended to change the radiation/reception pattern. In this case it is desired to keep full 360 degrees in the horizontal, but to make the vertical radiation pattern flatter in the vertical axis. This improves the reception/transmission around the antenna while reducing it above and below the antenna.

These colinear arrays are part of  system called scan telemetry. The outsite installations provide telemetry and receive commands, they have directional antennae like a UHF TV aerial. Here at the central node requests for data are sent, with installation address in the message, commands given, and replies received. This form, with a central node talking to many fixed outstations is called point to multi-point, as opposed to the point to point which the microwave drums described above use. Note that such colinear arrays are not just used for scan telemetry, they're also used to increase the area over which vehicles can radio a base station, and they're also sometimes seen at VHF. The image above and above detail is for a system that keeps the lights on and the gas supply running in the UK.

There are a couple of points that may have occurred to the reader. Firstly, why are these sites generally on top of hills? Perhaps more to the point: Why go to the bother of erecting large masts just to hold some relatively small antennae? This is because at VHF, UHF and above, the radio waves tend to be easily blocked by obstacles, so the best signal is obtained when one is in line of sight with the mast. So higher is better.

The second point is: Why are the drums different sizes? A major issue here is path length. The transmitter for the link is not of unlimited power, and the longer the link between stations, the greater the loss of signal between those two stations. The bigger the drum the more focussed the beam of radio waves it produces, so a longer hop means a bigger drum is needed. However this is not the only factor.

When data is modulated onto a signal it spreads the signal in frequency. Consider a 'carrier' transmission, i.e. a pure unmodulated radio signal at for example 1000MHz. All the power in that signal is contained in the spot frequency 1000.0000MHz, the moment the signal is modulated the power spreads out with frequency. So the spot frequency of 1000.0000MHz might become 1000.1000 to 999.9000MHz.

In the above graphic the red block indicates the power spreading due to 1 cellphone basestation. That basestation is linked by a microwave link to a node site (often but not always on a hill top), one site link is highlighted in red. There are three other outstation cellular sites, and the node site also carries its own cellular basestation site. So in its link to the switching centre (blue) it has to carry five times the amount of information as does one cellular base station. If the same power were used then it would drop the spread power very close to the receiver threshold for an acceptable bit error rate. Using a bigger drum/dish gives greater antenna gain and can bring the signal level back up. In reality the receiver threshold would be far lower than 0.2 Watt.

So because of the tendency to longer path lengths, and the carrying of more information, the larger drums tend to be trunk links, the smaller drums tend to be local links serving one or two remote sites. These are often cellular sites because microwave links are cheaper than laying cables, however some customers prefer microwave links as a robust alternative to cable, in some cases augmenting cable as a fallback in case of landline damage for critical systems (e.g. hospitals). In general engineers will always go for the shortest path and the smallest drum because shorter hops are less prone to weather disruption (heavy rain mainly), and big drums cost more.

Much of the links discussed above serve the mobile phone network. When you make a phone call, or use data services on your smartphone, your phone is communicating directly with a local gateway. The gateways we know as cellular phone sites.

The above photo is from another location, not Winter Hill. It shows a typical cellular phone mast serving a heavily populated semi-rural location with a motorway running through the cell.

There are two companies sharing the one mast, separation is maintained even to the point that looking at the two legs of the mast facing us, on the right leg goes the cables for the lower cellular antenna array, on the left leg goes the cables for the top array. Even the link antennae are so separated, with the lower two link antennae being for the lower cellular array, and all the link transceivers and the large link drum for the top array being amongst that. This minimises the angry "what the f**k has your engineer done to our kit" calls between site sharing operators.

Bearing in mind the principle I outlined above, that larger drums tend to be trunk links and smaller drums tend to be local links, look at the dishes and drums associated with the top tier of cellular antennae on that mast. Again we see a cluster of smaller dishes/drums, with one large trunk antenna, which connects all the sites linking to this site to a main node (on top of an office block in a nearby city, not much further away than some of the other cellular sites linking to those smaller antennae).

What do we mean by 'cell'? Each base station, such as above has a limited range, and has a region around it within which it handles the calls made or received within that region (also data transfers such as internet and app-data). In GSM the maximum cell size is 35km, this limitation is because of the propagation of radio waves at the speed of light, and the impact of the time taken for those waves to travel on system timing. That alone suggests the awesome accuracy of the timings involved. While the cell region does cover a geographical area the boundaries are determined by relative signal strengths and signal quality between adjacent cellular base stations.

Consider the case for a GSM phone where one is driving along the motorway engaged in a call, at 70mph you will change cells around every 10 minutes or less. To get a better grasp on what is happening, consider a single radio channel transmitted from one of the tall oblong antennae on the above site, let's say that base antenna transmits on 950.400MHz, then your handset will transmit 45MHz below that on 905.400MHz. The frequencies are allocated by the operator according to a set plan or database with frequencies being reused by different sites separated by a long distance. More on this here.

However you share that radio frequency with up to 8 other users, in GSM this is done by timeslotting. Although your speech comes out as continuous through the phone your phone packages the compressed speech, and other data, and sends it in bursts. Looking at the following graphic, I'll just use the bottom two rows.

The slot (bottom row) shows the structure of 1 burst of data from your GSM phone, the second row up shows that in this case the user is in timeslot 4 of 8.

So for 1 slot out of 8 your phone is sending data, then for 1 slot out of eight it is receiving from the base station, your mobile never recieves and transmits at the same time. That means that for 6 slots out of eight the transmitter/receiver is idle, plenty of time (~0.028 seconds) for it to scan the band looking for better prospective base stations. Incidentally, across the network the TDMA frame and time slots are exactly aligned between base stations. In other words, hook up two communications receivers to channels 1 and 2 of an oscilloscope and tune to different base stations and the training bits, line up pretty much exactly (allowing for the speed of light). This makes the searching and eventual hand over processes much easier for the phone and base stations. I should mention that the training bits are a set pattern of 1s and 0s that are used to determine radio link quality, errors in these indicate a poor link, and because it is a set pattern such errors are easy to spot.

Considering the case of driving along the motorway, when your mobile determines that there is a consistently better signal from another cell, it asks the base station controller, through the base station, if it can switch to that cell. The base station controller being a regional computer centre. The BSC controls the handover, and without you even noticing it you move to another cellular base station and continue your call.

So getting back to the subject of Winter Hill, while some of the point to point radio links seen on those masts (drum antennae) are for individual data users, most of it is backhaul to administer the massive network of cellular phone masts that has sprung up in the last few years.

This is another long post from me, but here I've not really scratched the surface of the technical complexity involved, I've not mentioned the augmented GSM system EDGE under which instead of the variant of frequency modulation (like broadcast FM) termed Gaussian Minimum Shift Keying which normal GSM uses, a timeslot can switch to 8PSK for an EDGE enabled handset. And I've not touched upon the W-CDMA used by 3G, or even 4G systems. Furthermore I've not addressed Quadrature Phase Shift Keying or Quadrature Amplitude Modulation as used for modulation in the back haul microwave radio links. And as I'm focussing on the radio masts aside from the TV transmitter on Winter Hill, I also haven't mentioned the awesomely cunning Orthogonal Frequency Division Multiplexing system that brings multiple digital terrestrial Freeview TV channels into your home.

Even for myself as an engineer it is too easy to become blasé about this technology, and to overlook just what a highly advanced technical civilisation we have become. Indeed an engineer colleague of mine parks every day in front of a cellular site, and never realised what it was until I pointed it out, another one thought his mobile phone worked via satellite, not an uncommon misconception.

Arthur C Clarke's third law states:
Any sufficiently advanced technology is indistinguishable from magic.
We are such a sufficiently advance culture, and if we take the time to step back and consider what we often take for granted, we produce works that are indistinguishable from magic. Some claim that technical development is to the detriment of our humanity, that it is cold and impersonal. Yet the advanced, almost magical, technology I have outlined above has been driven by our most basic human needs, the need for community and communication. Even in our technological magic we display our humanity.


Anonymous said...


I've left a couple of harsh posts here in the past, but am compelled to say that this one is superb.

I find it surprising that the person who wrote the above post can "root" for the disappearance of the Arctic ice.

Humans are complex creatures.

Best in 2015

Mitch said...

You should learn the difference between monitoring and 'rooting for'. Do you understand that Arctic sea ice is significantly reduced since satellite monitoring began in around 1979? How has that happened?

The argument that climate has changed in the past ignores the fact that climate is a physical system that depends on the energy absorbed by the ground.

Chris Reynolds said...


I'm not sure who Anonymous is. But I have said here and elsewhere that I would find a rapid crash of sea ice far more exciting than a slow reduction. As it is I'm now pretty much convinced it will be the late 2020s at the earliest before we see regular virtually ice free state (late summer <1M km^2 extent).

Kevin O'Neill said...

Happy New Year, Chris.

I've spent quite a bit of time the past decade inside cell sites - usually main switches within a few hundred yards of a tower.

I even did some work in Hickory Hills, the switch through which the first cell phone call was made. When I was there a few years ago they were completing the transition to all digital. So you had the original building where the switch was row upon row of racks with green wire everywhere and then the new switch which was room with a few isolated test racks with computer servers in them connected by a few strands of CAT5.

Of course I also spent more than a decade working for a military defense communications contractor - so I learned to love my Vector Signal Analyzer. Between a Network Analyzer (I grew up on the HP 8510) and the VSA (like the HP/Agilent 89441A) if I can't measure it, it ain't worth measuring :)

Chris Reynolds said...

Happy New Year Kevin,

I was mainly working on the roll out of GSM during the transition from TACS (the UK version of AMPS). I started off antenna rigging during my degree. Since moving out of that field I've kept an interest, but don't know as much about 3 and 4G as I could (something else gets in the way ;) ).

HP(/Agilent/Keysight) spectrum/vector analysers are awesome. The main reason I like them is the manuals with verification procedures spelled out for simple people like me. ;)

Kevin O'Neill said...

"The main reason I like them is the manuals with verification procedures spelled out ..."

One of the drawbacks of calibrating equipment for the communications industry (similar to the bio-med field) is that they don't generally produce nice, clear verification manuals.

Even when the manufacturer (Tektronix, Fluke, Agilent/Keysight) otherwise produces good manuals for their general purpose test equipment, they produce little for their telcomm equipment. The cell techs over here often use an Agilent E7495A, commonly known as an 'El Gato' - good luck finding much on it. And protocols are constantly changing/emerging. So what's in firmware can widely vary from instrument to instrument.

The whole TTC/Acterna/JDSU line of equipment is another nightmare of options, versions, installed modules, etc. with very little manufacturer's support.

Chris Reynolds said...

I didn't realise that. Thankfully we're mainly a general industry calibration outfit. We do spectrum analysers, but mainly the general purpose ones. I get asked about doing things like comms test sets but all I see for that sort of kit is hours of messing about that go over what the customer is quoted for. Then you never see it until next year, if then. So I politely turn down such stuff.

If people have that sort of kit to calibrate let them go to the specialist firms and stomach what they charge.

Unknown said...

Are these ground based systems for every day folks 'sat nav'

deletedauto developer said...

Are these ground based systems for every day folks 'sat nav'

deletedauto developer said...

Are these ground based systems for every day folks 'sat nav'

deletedauto developer said...

Are these ground based systems for every day folks 'sat nav'

Chris Reynolds said...

deletedauto developer,

No. Sat Nav (GPS) is from satellites in orbit.