Figure 6.4 Using a clock pulse to control timing of a circuit
A register is just a collection of flip-flops. A flip-flop can only store one bit so to handle 32 bits at a time we would need 32 flip-flops and would refer to this as a 32-bit register. To save space, Figure 6.5 shows an 8-bit register.
Figure 6.5 An 8-bit register
The register has two distinct groups of connections: the data bits 0 to 7 and the control signals. The data connections or data lines carry the binary levels in or out of the register. The number of data lines determines the size of the register so a 64-bit register would have 64 data connections.
The three control signals include two new ones
1 Enable. This is a simple on/off switch for the register. We met this in Chapter 5 with the tri-state buffer. The line over the top of the word indicates that it is ‘on’ when this line is ‘low’ or at logic zero. We tend to say the line is ‘active low’ in this situation. Therefore, it follows that the register is disabled or switched ‘off’ when the enable line is at logic 1 or ‘high’. Nearly all control lines are active low. The benefit of having the enable line is that we are able to disconnect a register without doing any physical uncoupling of links etc.
2 Read/write. The terms ‘read’ and ‘write’ are used to describe the direction of data movement. We ‘write’ data into a register then ‘read’ the data to recover it.
You may remember that the flip-flop in Figure 6.1 included a separate line for reading the data and another for writing. Now, while this was OK with a single flip-flop, a 64-bit register would require 128 lines just to carry the data in and out. By using the tri-state buffer from Chapter 5, we can use each line to read and write data as required. The tri-state buffers are all controlled by the logic level applied to the read/write line. The normal convention applies – the line over the ‘write’ means that this line is taken low to write data and, of course, high to read data.
It may be interesting to look inside the register to see how the tristate buffer is used to achieve this two-way traffic on a single wire. Have a look at Figure 6.6. Two tri-state buffers are connected back-to-back. In the first example, logic 1 input will enable the top buffer. The control voltage is inverted to a logic 0, which then disables the lower buffer. Data can now flow from left to right. When the control signal changes to a logic 0, the top buffer is disabled and the lower one is enabled and the reverse direction of data flow is possible. Note how two buffers and an inverter are used for each line to be controlled. A single control line is used to switch all the data lines at the same time.
Figure 6.6 Two-way data flow
What are registers for?
Registers are storage areas inside the microprocessor. Almost the whole of the microprocessor is made of registers. They store the data that is going to be used, they store the instructions that are to be used and they store the results obtained. Nearly all registers involve tri-state buffers to control the direction of data flow.
In most cases, the data to be stored is applied to the inputs of the register and, after a short pause to let the voltages stabilize, the register is enabled by the voltage on the enable control. The information is then safely stored until it is next required.
The sequence is:
1 The read/write line is taken to logic 0 to allow the register to receive data from an external source.
2 The enable control switches ON the tri-state buffers at the input to each flip-flop.
3 The data is written to each flip-flop and then the enable control puts the register to sleep until the next time it is needed.
How long can it be stored?
It will be stored until the power supplies are removed – either by an equipment fault or, more usually, by the system being switched off. The data does not deteriorate in storage.
These are a variation on the register theme. They still consist of group of flip-flops but the interconnections have been changed. Have a look at the arrangement in Figure 6.7 and see if you can guess the likely outcome.
Figure 6.7 A shift-left register
This is called a shift register because the data is shifted from one flipflop to the next each time the clock pulse occurs. Specifically, the one shown is a shift left register because each bit moves one place to the left on each clock pulse. All the bits move at the same time. The last one in bit 7 drops off the end and is lost while at the other end, a new bit is entered into bit 0.
In Figure 6.8, the register has been loaded with the binary equivalent of 3610 or 24H and a series of zeros has been chosen to be loaded at the bit 0 end.
Figure 6.8 A shift-left register in action
Follow the sequence through and in particular note what happens to the numbers stored:
1 After one clock pulse, all the bits will have moved one place to the left. A new ‘0’ will have entered bit 0 and the last, which was in bit 7, will have fallen off the end of the world. The bits stored at this time are 01001000 and the numerical value is 48H or, in denary, 7210. Notice how shifting the bits to the left has multiplied the value by 2.
2 After eight clock pulses, all the existing data in the register will have been flushed out and refilled with zeros. The register will hold the number zero so there is a limit to how many times we can multiply by shifting the register.
3 After 5000 clock pulses, it is still full of zeros. Admittedly, they will be new zeros that have replaced the others but that will not make any difference.
What happens if we don’t apply any input data to enter bit 0?
If the input connection is simply left unconnected, there will be no voltage information coming in to the first flip-flop. The input is said to be ‘floating’ and will assume some voltage which may be low or high. As the clock pulses are applied this may well result in random data entering the register. Random data is of no help to anyone so we normally overcome this problem by building in a bias in the design of the register to make it have a tendency to move towards one logic level rather than the other. It is up to the manufacturer but most floating inputs will float high and enter ones.
The shift register considered has been a shift-left register, which means that the information is fed in at the right-hand end and moves progressively towards the left along the register until it drops off the end.
By re-arranging the register, it is easy to produce a shift-right register as in Figure 6.9. This has all the same properties except it shifts data towards the right and divides the number by two each time the clock pulse is entered. Compare Figures 6.7 and 6.9.
Figure 6.9 A shift-right register
A real world use for a shift register
It is interesting that a shift register can perform simple multiplication and division but it can do many jobs that are more interesting.
One example would be automatic checking of inputs. In Figure 6.10 it is controlling an automatic ticket dispenser. The customer inserts some money and presses any button of the eight available to obtain the ticket required – but which button was pressed?
