In Lecture 1 we introduced the idea of a current source, and mentioned that these were slightly more difficult to implement than the perhaps more familiar voltage source.

We are now in a position to learn a practical way of implementing current sources. The circuit schematic below (which has been entered into the pSpice circuit simulator) shows a simple way of doing this using a JFET (see Lecture Notes page 86). Note that the symbol used for the JFET, type 2N3819, is not quite the same as the more widely accepted one we have encountered!

The negative terminal of the supply V1 (battery symbol) is connected to
Ground, also known as Node 0, and represents the reference point for any
measurements we might make. Although the battery voltage is indicated as
5V, pSpice will let us vary this experimentally over any required range.
The current developed by the circuit flows in the drain circuit, and is in fact
I_{D}. For this application a light-emitting diode LED is shown as
the load, in which the constant current flows. This is a very typical
usage for this kind of circuit; the LED works most consistently (giving a
steady light output) if the current passing through it can be maintained
constant, even if the supply voltage changes - for example, as the battery
discharges.

The circuit can be described as a self-biased JFET: the source resistor R
determines the gate source voltage through the relationship V_{GS} = -V_{S0
}= -I_{D} R

This allows a range of currents to be achieved, since the operating
conditions of the JFET can be chosen by selecting the value of R, which
determines the relative values of I_{D} and V_{GS} . The
analysis is as developed in Lecture 7.

pSpice can help us to predict many aspects of the circuit design's behaviour
before we construct it. In order to model the voltage-current
characteristic of a non-linear device like a FET, pSpice can use a number of
different mathematical models. For this application, the JFET is modelled
using an expression based on a binomial approximation for the dependence of I_{D}
on V_{DS} and V_{GS}. Different JFET types require
different coefficients to be used in the expression, and these can be
incorporated.

Here the pSpice simulator has been used to predict the performance of the circuit as the supply voltage is varied over a range. If the design is working correctly, we expect to find the current to be independent of V1. pSpice has been used to predict the circuit's behaviour with different values of R, from 0 ohms to 1000 ohms. We expect to see the current change as different values of R are introduced.

The family of curves shown above represent the pSpice output. The
expected value of I_{D} (the current in the LED) is plotted against the
supply voltage V1, for various different values of R.

We can see that provided the supply voltage V1 lies within the range 5 volts to 25 volts approximately, the current hardly varies with V1: it is effectively constant. However, when V1 falls below about 5 volts the circuit is seen to fail, as the current drops rapidly.

Had we used a resistor on its own to try and control the current through the LED, the current would have been strongly dependent on V1 for all values.

The selection of the value for R can be seen to have a considerable effect on
the constant current obtained. By suitable choice of R, a constant current
anywhere in the range of about 2 - 12 mA can be obtained. Looking at the
curve for R=0, we can see how the circuit on page 85 of the notes will
behave. Under these condition we know from the transfer charact eristic
that the JFET will operate with a drain current equal to I_{DSS} (see
Lecture Notes, page 85). The disadvantage of this very simple circuit is
that you are stuck with the value of I_{DSS} for the particular
transistor you have chosen, and this can vary considerably from transistor to
transistor. Including R gives the circuit much greater versatility.

Although the JFET can provide a simple solution for the design of a current source, it is not the best solution obtainable. Much better designs can be achieved using bipolar transistors (which you will meet in the second year), or by using operational amplifiers (which you will meet in lectures 17-18 of this course). By use of suitable mathematical models for these components, pSpice can also predict the performance of these different approaches, and is therefore a powerful ally for the electronic circuit designer. However, it is important to remember that the results obtained are only as accurate as the numerical models used to represent the devices being simulated. In a real circuit, manufacturing tolerances and other variables (e.g. temperature) may cause the real circuit to operate quite differently from the simulation. A good designer will always take these possibilities into account.

David Holburn October 2005