Parallel Communications Explained

Parallel communications are used over very short distances; typically inside the computer itself and to printers. This method, together with the connector, was developed by Centronics and used by IBM in its first PC. As a result, it has become relatively standardized. It is fast-ish, and distance-limited. The original design was one way only, but bi-directional ports have been developed, a side benefit being the ability to control a printer directly with suitable software. Enhanced parallel ports (EPP) are bi-directional high speed data buses that can transfer data at ten times the speed of standard ones. This makes them useful for attaching tape streamers or extra hard drives.

The wires are very close together, and because the strength of any signal diminishes the further it goes down the line (due to the work it has to do to get past the resistance of the wire itself), there is a chance they could be interfered with when they become weak enough, giving the possibility of crosstalk, on top of normal attenuation, where signals from one wire will be reflected in the next. This is why parallel communications are generally restricted to short distances unless boosters are used.

Just to send the data, we need a minimum of eight wires, so a complete character can be sent at a time. However, that's not all. What if the receiving computer wasn't able to take everything at once (perhaps because it ran out of memory) or detected an error and wanted to tell the sender to stop and send that last bit again?

There will be other connections between the computers with another conversation going on, telling each other when to stop and start sending, and giving themselves progress reports. As you can imagine, we now begin to collect quite a few wires, usually about 15 in all, depending on the equipment.

As the most common encounter with parallel transmission is with printers, we will take an example of how characters are sent to one as a more detailed illustration of what goes on. However, it's also possible to make computers communicate in the same way, as when transferring data between machines, or linking to a network.

The parallel port is actually controlled by two chips, one for information and the other for control signals. Both are linked to a decoder which sorts out to which pins in the connector the relevant bits are sent.

When the printer is on line, it sends a Ready signal to the computer, whose response is Initialization, sent on another wire, which clears the buffer (a bit of memory that stores data so it can be sorted out before printing).

Once the printer is initialized, the data is lined up, one bit per wire (like the start of a race) and, assuming the printer hasn't sent a Busy signal, a low voltage signal (DataStrobe) is sent to the printer. This acts like a starting gun to send the data down the lines, at the other end of which it assembles in the buffer ready for printing, after it's had a coffee.

As soon as the data arrives, the printer acknowledges receipt with another low voltage signal (Acknowledge) sent on yet another wire. Then the whole process starts again. Anytime it needs to, the printer can send a Busy signal which tells the computer to hold everything until given the all clear. There is also a fault signal which stops the process if unhealthy conditions exist, such as lack of paper or a stuck ribbon. Or cold coffee.

As you can send all the bits at once rather than one after the other, parallel communications are fast and accurate, because it's easy to identify which bit is which by knowing what wire it came in on, and when. Thus, because you're not carrying overheads for error checking, as with serial (below), it's possible to transfer data very efficiently, but only for short distances, as noise and signal deterioration increase rapidly with distance.

This is fortunate, because trying to arrange long cables of the thickness needed for parallel communications around corners is very frustrating-and more so if a separate ground wire is twisted with each signal wire to reduce interference. If you couple this with switching arrangements for each one, you can see that parallel transmission has the potential for being impractical and horrendously expensive.

This is an article from  Phil Croucher, author of  Communications and Networks.  Phil has a way of explaining in "plain" English. The information is well presented and is well above A+ standard.