28 June 2011


The idea to be able to communicate without physically plugging in is an old one.  A few years ago infrared was the primary technology used to achieve this.  The most popular standard was one specified by the Infrared Data Association (IrDA); the acronym IrDA was often used to refer to the infrared port itself.  IrDA communicated at 115kbps.  A fast infrared specification (FIR) was also published that was supposed to communicate at 4Mbps.  One did not encounter many FIR interfaces - and one never encountered a 4Mbps infrared link.

The problem with visually showing an infrared port is that they mostly look like black plastic.  A small black rectangle on the side of a (black) notebook or cellphone does not make an interesting picture.  So, here is an IrDA unit that may be installed on a personal computer to enable it to communicate with cellular phones and other portable devices with a similar interface.

As can be seen from the picture the front of this adapter indeed looks like black (reflective) plastic.

From the top it is somewhat easier to see what the unit looks like.

Normally infrared light is not visible to the human eye.  However, some cameras are able to record infrared light that makes it indirectly visible.  The next picture was taken with such a camera while the IrDA unit was communicating.  The red light on top of the unit (appearing to be orange in the picture) is a normal red light that flickers so that the user can see that the unit is operating.  This red light causes some other red reflections in the translucent plastic.  The important thing to notice is the light that looks purplish in the picture.  This is the infrared light shining behind the translucent plastic to transmit data.  As noted this light is normally invisible to humans.
To get two devices to communicate via infrared one would line them up so that their IrDA ports pointed towards one another and were in close proximity.  (Typically they had be be closer than about 1m to each other.)  And then one could request a file to be transferred or some other communication to take place.

This particular IrDA adapter was designed to plug into a DB-9 serial port.  While every personal computer in the world at some point in the history had a serial port it was surprising how difficult it was to find a computer into which the IrDA adapter could be plugged to get it to light up.  Later posts will say more about serial interfaces.

Let's also say something about the camera used to take the infrared picture.  With film cameras one can load an infrared film into any camera and take pictures (with some special considerations - for example, the point at which infrared is in focus is somewhat different from the point at which visible light is in focus, so one needs to focus at a somewhat different point than what one wants to be in focus).

In the case of digital cameras, all sensors are sensitive to infrared light.  This is nice if one wants to take infrared pictures.  However, the vast majority of people who take pictures, want to take a picture of the visible object and infrared light tends to enter into the picture at unexpected places and ruin the picture.  Therefore most digital cameras have a filter in front of the sensor that only allows visible light to pass.  Special forensic cameras are available that do not have this filter, but they are horribly expensive.

Another option is to find an old digital camera that was manufactured before these visible light filters were installed.  This is the type of camera used to take the picture above.  However, before saying a bit more about the particular camera it should be noted that two other options exist.  One is to use a video camera without such a filter; such video cameras (often with a 'night mode') are more readily available.  Finally, one may buy a modern camera and break the filter out.  (Remember not to damage the sensor!)

The camera used to take the infrared picture in this post is a Kodak DCS 315.  Launched in 1998, the 1.5 megapixel camera represented cutting edge technology aimed at the serious photographer.  At about $5 000 (in the US) or £5 000 (in the UK) or almost R100 000 (in South Africa) it was considered a budget alternative when compared to other digital cameras at the time.  As can be seen from the picture below it used a Nikon Pronea 600i body on top, with electronics added to the bottom and rear by Kodak.  Normally one would use it with a hot mirror filter screwed into the front of the lens to cut out infrared light.  The picture shows the camera without such a filter, because the filter was (obviously) removed to take the picture of the IrDA adapter in action.  In general, most current cellphones take better pictures than this professional oldie, but cellphones still can't photograph infrared light...

21 June 2011

19-inch cabinets

Network equipment is often placed in cabinets that happen to be 19" wide.  Now I first have to explain to my students that 19" is read as 19 inches, where inch is the imperial unit of length used in South Africa until the end of the 1960s.  (It is still used in a few places in the world.)  The simplest rule of thumb to covert between imperial and metric units of length is to remember that a standard ruler was 12" (that is 1', or one foot) long and those same rulers are now 30cm long.  Hence we are talking about cabinets that are 47.5cm wide, but nobody calls them 47.5cm cabinets.  Some people do call them 19" racks, though.  (A more precise conversion of 19" is 482.6mm.)

Below is a picture of such a 19" cabinet.  The glass door has been opened so one can clearly see the innards. The most important things no notice are the 'pillars' with square holes in them on both sides of the cabinet.

An old ruler has been placed at the bottom of the cabinet; it dates from the 1960s and is therefore marked in inches.  The crop below shows that the distance from hole to hole is about 18".

In a real network many units will be fastened to the cabinet.  The picture below is an example that shows a typical installation.  Careful observation reveals that each of the pieces of equipment in this cabinet is exactly three holes high - somewhat more precisely they are 44.5mm high.  The height of a cabinet is measured in units, abbreviated as U.  Counting the holes in the cabinet above (27 holes) reveals that it is a 9U cabinet.  Since the top and bottom of the cabinet in the picture below are not visible, it is not possible to determine its size from the picture.  At least eleven units are visible in the picture, with spaces between some of them.  This cabinet is obviously much bigger than our earlier example.  Note that some pieces of equipment may be larger than 1U - a rack-mountable computer may, for example, be 2U high.

Here is the earlier cabinet again, this time from the inside.  The rounded shape is distortion caused by the lens used.  This cabinet has some horizontal pieces of metal, which is unusual.  The most important piece of metal in this picture is still the pillar on the right with its holes that face the glass door.

The traditional way of fastening equipment to the cabinet is by means of bolts and cage nuts.  Note the two nuts in the picture below are indeed in their own little cages.

A closeup makes the nature of the cage even clearer.

To use the cage nuts they are clipped into the square holes of the front pillars.  The picture below shows how the caging clips into the hole and the cage then holds the nut in place.  It is now possible to fasten a piece of equipment from the front without having to try to hold the nut at the back.

In the picture below two bolts have been screwed into nuts.  In this case they hold a matchstick - to indicate scale.  These two screws have one hole open between them.  The matchstick is therefore about 1U high.

20 June 2011


In many office buildings a wire conduit runs along some of the walls that enables one to easily install a new cable to any office (and, in principle, to remove old unused cables).  The terms reticulation or cable management are used for guiding cable through such conduits.  Often the conduits are installed about 1m above the floor to make it easy to plug plugs into sockets installed in the conduits.  Such conduits are known as trunking.  (Apparently they are known as wireways in some parts of the world.)

Here is an example of trunking installed just below the window sill in an office that will not be identified.

Viewed from a more conventional angle one can see a South African electrical socket on the cover over the top duct and several sockets and cables attached to the cover on the bottom duct.
The first socket on the bottom lid is an old South African telephone socket featured in an earlier post.  It is no longer being used in this installation.  Next, just below the electrical socket, is a pair of thinnet coaxial cables.  The logic of having a pair is that one goes to the left in the duct and the other to the right.  The coax cables are   no longer being used in this installation.  The third item is a pair of RJ-45 sockets.  They are actively being used.  Finally there is a pushbutton that once upon a time could open the security door that allows access to the corridor.  The button is no longer being used in this installation.

Removing the covers reveals the innards of the trunking.
The top duct clearly carries the power lines - live, neutral and earth.

The middle duct is unused.

The bottom duct contains a large number of UTP cables - as one would expect in a modern setup.  Note the black coaxial and cream phone cables also making an appearance.  Also note the loops in some of the cables: When installing cables one quickly learns to take a cable beyond its intended termination point and then return with a loop.  This gives one some space to work when attaching the socket or other fixture to the end of the cable.

12 June 2011

Phone plugs

Many South Africans will remember the phone plugs used in the old days - the plug shown in the picture below.  In fact, many probably still use these old plugs.  (I remember when these old plugs were the new ones that replaced an even older variety...)
Of course, early modems sold in South Africa used the exact same plug.  However, in later years modems had the now ubiquitous RJ-11 plugs and came with a separate adapter.
The RJ-11 socket should look familiar to most readers of this post.
It's the other end that looks strange to a new generation of students.
Of course, it's not only South Africa who had strange phone plugs.  Many kits were available that allowed one to convert local phone sockets to RJ-11 sockets wherever one travelled in the world.  The picture below shows one such kit manufactured by Targus that enabled one to travel through most of Europe and remain connected.
All these plugs split the local socket into an RJ-11 socket, as well as a local socket.  One could therefore unplug the phone in one's hotel room, plug in the adapter, and then plug the phone back into the adapter.  In addition one could plug one's modem into the RJ-11 socket of the adapter.

Here are the 'business ends' of the adapters.
Unfortunately these plugs did not solve all one's problems.  One also needed a local ISP to call.  Alternatively one had to make an incredibly expensive call to one's home country that typically suffered from really bad interference.  The best option was to use an ISP with an international presence.or one that had a roaming agreement with various national ISPs.  Unfortunately such ISPs were scarce - especially in the early years of mobile computing.

11 June 2011


Originally LANs were built using a rather thick variety of coaxial cable.  Later, when a thinner version was adopted, the original was colloquially referred to as thicknet and the later variety as thinnet.

A cross-section through thicknet looks more or less as one would expect.

Nobody can deny the inherent beauty in such cables.

As always a sense of scale might be useful.

  A more realistic comparison is one with related cables.
The coax on the left is thicknet.  Officially it is known as RG-8/U or, in Ethernet applications, as 10Base5.  The 10Base5 designation stems from the fact that it could be used for baseband transmissions at 10Mbps over distances of up to 500m.

Next to the thicknet is an example that looks like thinnet.  This example is RG-59B/U.  Real thinnet looks very similar, but has an impedance of 50Ω, compared to the 75Ω of RG59B/U.

The white coax is RG-6/U - commonly used to connect antennas and satellite dishes to TV sets.

The final cable is Cat 5e UTP - commonly used in current network installations.

10 June 2011

Air-blown fibre - Part 3

Air-blown fibre, as noted earlier, is blown through a microtube that is installed between the two endpoints.  Such microtubes may be connected using various connectors.  A simple example is one that provides a straight-through connection.  Here is an example.
To connect a tube it is simply pushed into the connector, which locks it in place.  The tube it should be connected to is inserted at the opposite side.
To release the tube again, the white ring at the edge of the connector is pushed inwards.

09 June 2011

Air-blown fibre - Part 2

The air-blown fibre shown below has yet more secrets to reveal.
Now one can clearly see the eight cores (or eight fibres) in the single 'fibre'

Zoomed in a bit more matters become even more obvious.

Yet again we need our matchstick to get a sense of scale.

Perhaps the setup to take these pictures may be interesting to some readers.
In this case a Pentax K-5 (silver!) fitted with a Pentax Auto Bellows K and a Pentax DFA 100 f/2.8 manually stopped down to f/8 at an exposure time of about 15s was used.  Strange how the auto functions of years gone by become manual (if they are old enough).  The Auto Bellows can step down the lens, but no longer fire the shutter, because the shutter mechanism is triggered by an electrical contact.  I cannot use my latest DA lenses because they have no aperture ring to limit the degree of stopping down that will occur.  But I am digressing.  The next post will focus on networks again.

07 June 2011

Air-blown fibre

Installing optical fibre can be a costly and difficult process.  One relatively recent option to ease the task is known as air-blown fibre.  Rather than installing expensive fibre, one installs tubes, which look like straws that are often bundled together.

Fibre is then blown through these tubes as required.  The tubes may be installed over hundreds of meters - even a few kilometers.  And fibre can be blown through the tubes at tens to hundreds of meters per second.  Initially more tubes can be installed than required and fibre could be blown through them when required.  And an obsolete fibre can be removed and replaced with a new one.

The next picture shows a fibre in one of the tubes.
When one shines a light through the fibre it becomes clear that it actually consists of several fibres (or cores).  The next picture contains the same fibre as above, but this time lighted from the other end.
Of course we need an image with a matchstick in it for a sense of scale.  The following picture not only includes the matchstick, but also shows the fibre above with its outer coating stripped away.

To take matters to the real world, the next picture was taken at the place in a network room where all the various parts of some big network come together.  Various pieces of optical fibre are visible.  In particular, note the microtubes (making loops!) through which fibre has been blown to other buildings.

03 June 2011

Building the netbook - circa 1998

A small, highly portable laptop that can be used for typing and sending emails seemed necessary to me for as long as I can remember.  Whether such a machine was fast or had a big screen did not matter.  In other words, I have been craving the netbook long before its inception.  However, even though it feels like an eternity, the option to build such machines is a relatively recent one.  The laptop with the monochrome screen mentioned in an earlier post was my first such machine.  I now found the receipt: 1997!  It is less than 15 years old, has 20MB of memory and a 400MB disc...

One thing missing from many (most?) laptops of the time was any real mechanism to network it.  Dial-up was one option using some external modem.  A later post will talk about this.  But to have a truly networkable laptop, one needed a real network interface.  So I got a 3Com Etherlink III 3C589C LAN PC card for my 1997 laptop.
As can be see from the photo, the interface consisted of two parts: a dongle and a PCMCIA card (or, as some people insisted to call it: a PC card, which was technically correct).

The dongle is obviously the part that connected to the network.  Most networks I wanted to connect to used coax, but it was obvious that UTP was coming.  So, I tried to future-proof my investment by buying a dual adapter - one that could connect to coax using a BNC connector, as well as to UTP, using an RJ-45 plug.  Alas, the UTP connection was only a 10BaseT, and 10Mbps soon turned out to be too slow.

Here one sees the card and dongle assembled - ready to be inserted into the laptop's PCMCIA slot.
But the need for speed had to be satisfied.  And satisfaction came in the form of the SMC8040TX PCMCIA 10/100Mbps Fixed-Port Adapter.
The SMC still could work at 10Mbps.  But at 100Mbps it was ten times as fast as its predecessor.
Obviously the SMC dates from the year 2000.  For those who have not 'lived' the years 2000 to about 2004: Apple's iMac G3 was produced from 1998 onwards in a range of bright translucent colours.  Following Apple's lead, everything else in those years was also produced in bright translucent colours.  Here is our kettle circa 2000.

There can't be too many fields where one can reminisce about the good old days a mere decade ago...

Satellite phones

Below is a rather poor picture of a satellite phone (and another in its box).  I recently took the picture in a shop display in Johannesburg in rather poor light using my cell phone - hence the poor quality.  The visible phone is intended for use with the Iridium constellation of low earth orbiting (LEO) satellites.  Since the entire earth is covered by the constellation it is really true that one can make calls "from anywhere [on earth] to everywhere" on earth - unless one is somewhere where there is too much of an obstruction between the phone and the sky.

While the picture is poor it does illustrate the relative size of the phone's antenna when compared to the phone.  The phone itself is rather bulky too.
I did wonder whether it was an option to buy one of these phones to have a physical example in my museum..  The answer is a resounding no.  This cost of the phone is about R15 000.  Calls currently typically cost between $1 and $2 (US) per minute (depending on where on earth you are).  Data calls currently cost about $1.50 (US) per kilobyte.  The phone weighs 266g.  Its dimensions are 143x55x30mm.  The battery lasts about 30 hours while on standby and about four hours when in use.

The phone (or, at least box) on the left uses the Inmarsat geostationary satellites (GEO).  It is slightly heavier than the Iridium phone (279g vs 266g) and slightly bulkier (at 170x54x39mm).  Interestingly, despite the higher altitude of the Inmarsat satellites, talk time is double that of the Iridium phone and claimed standby time is 100 hours.  The claimed data rate is up to 20kbps.