circuit-fantasia > circuit stories > building circuits > current source (philosophy)

# Circuit Stories on the Whiteboard...

## (philosophy)

### Circuit stories about current source:

Problem

Passive current source

Bipolar transistor current source

Field-effect transistor current source

Bootstrapped transistor current source with shifting capacitor

Bootstrapped transistor current source with shifting diode

Widlar op-amp current source for grounded load

Op-amp inverting current source for floating load

How to build Howlend op-amp current source (version 1)

How to build Howlend op-amp current source (version II)

Negative feedback transistor current source

Op-amp non-inverting current source for floating load

Voltage regulator acting as a current source  Problem. Have you ever thought about the fact that almost all the natural electrical sources are constant voltage sources? Actually, there are not constant current sources in the nature (excepting inductor and Van der Graaf generator). So, if we need a current source, we have just to build it!

In this story on the whiteboard, we will discuss the philosophy of the current source creating. Here we will attempt to answer the questions: What is inside the constant current source? How do we build it? How many ways of current source creating exist? What is the best of them?

In order to reveal the truth behind these famous circuits, we will discover all the possible ways of current source creating. If you want to know more about the concrete implementations, follow the links pointing to the according concrete pages.

Ideal current source in ideal load conditions. Every, even the most elementary electrical circuit, consists at least of a source and a load. Let's consider the most elementary current supply circuit.

A current source likes a load with zero resistance and voltage (see below to know why). So, let's first short the current source output with a piece of wire (it will behave as an ideal current load LI with zero resistance).

Passive compound current source in ideal load conditions. Now, let's first build a simple current source... Well, we have a voltage source V but we need a current one. So, in order to build it, we have to connect a voltage-to-current converter (a humble resistor Ri) after the voltage source. Here is the building "formula":

V-source + V-to-I converter = I-source

Paradox. Note, that this is the simplest current source. Only, as it works at ideal load conditions (shortened output), it behaves as a perfect current source. Do you agree?

If you want to know more about the simple current source, visit the page below:

What is the idea behind a simple current source?

Analog electronics 2004 - Class 1

Passive compound current source at real load conditions. But we have created this compound current source with purpose to use its current. So, we have to break the circuit and connect a real load. For concreteness, let's consider the case when the load has an ohmic resistance RL (e.g. an imperfect ammeter with internal resistance).

Problem. Only, one problem arises here - the load resistance RL affects the current (the voltage drop VRL across the load is harmful as it enervates the excitation voltage V). Now the voltage difference V - VRL determines the current instead the voltage V. As a result, the current decreases. What do we do, in order to decrease the error?

Remedy 1: keeping a comparatively constant current by depreciating the load. This is the well known electrician's recipe for current source creating. See the page below to know more:

How to build a simple current source

Analog electronics 2004 - Class 1

Remedy 2: keeping a constant current by varying the internal resistance. Obviously, we may change Rvar (Ri) to the opposite direction thus keeping a constant total circuit resistance (Rvar + RL = const). In this arrangement Rvar will act as a resistor with dynamic resistance, which complements RL to the full constant circuit resistance.

Experiment: Well, let's go to the laboratory and connect two rheostats acting as RL and Ri. Now move randomly the RL slider; I will watch your actions and will move the RL slider to the same but opposite position. What about the current? Is it constant?

Only, I am a human being; so, I prefer to assign this donkey work:) to a BJ transistor or to a FE transistor (is it better, why?)

How to build a simple bipolar transistor current source

How to build a simple bipolar transistor current mirror

How to build a simple field-effect transistor current source

and continue thinking about the philosophy of current source creating... Remedy 3: keeping a constant current by varying the voltage. Then, a new idea flashes on us - to change the voltage Vvar (V), in order to compensate the voltage drop VRL across the load. Now, Vvar will act as a following voltage source keeping constant voltage difference (Vvar - VRL = const), which produces constant current.

Experiment: Move again the RL slider. I will replace the steady voltage source V with a varying (dynamic) source Vvar and will connect a voltmeter across the load. Then, looking at the voltmeter, I will move the Vvar knob to the appropriate direction and position. What about the current? Is it constant?

Finally, let's assign this boring task to a transistor or to an op-amp

Bootstrapped transistor current source with shifting capacitor

Bootstrapped transistor current source with shifting diode

Widlar op-amp current source for grounded load

while we think about the next current source creating idea...

Remedy 4: keeping a constant current by adding a voltage. The two ideas above are good enough but in many cases we can change neither resistance nor voltage (e.g. imagine that V and Ri are located at long distance). What do we do then, in order to keep a constant current? Well, let's reason again...

When the current flows through the load a voltage drop VRL appears across it. This voltage is harmful voltage as it decreases the excitation voltage V. Remember what we do in real life when an obstacle stands in our way - we remove it by an equivalent useful "antidisturbance". Then, why don't we compensate the harmful voltage VRL with an equivalent "antivoltage" -VRL? We have just to add so much voltage to the excitation voltage source V, as much as it loses across the load!

Experiment: Move, as usual, the RL slider. I will add an additional supplementary voltage source Bs and will adjust its voltage VBs = -VRL by looking at a zero indictor. What about the current?

Again, an inverting op-amp can accurately do this routine work:

How to build an op-amp inverting current source Remedy 5: keeping a constant current by adding a current. As we can see, the current I decreases with VL/Ri because of the load voltage drop VL. Above, we aided the exciting voltage source with an additional supplementary voltage source. As you probably already guest (my students usually guest about this clever idea), we may aid the voltage source injecting an additional current Is = VL/Ri by another "helping" current source. We may build it by connecting a "helping" voltage source Bs through another resistor Ri to the load. But this is the great Howland's idea about current source creating!

Experiment: Move again the RL slider. I will connect a voltmeter across the load and will move the Vs knob so that Vs = 2VRL. What about the current? Is it constant, as we have supposed?

Only, how to make an op-amp do this routine work? If you want to know, visit the page below:

How to build Howlend op-amp current source (version 1)

How to build Howlend op-amp current source (version II) Remedy 6: keeping a constant current by negative feedback. So far, we were keeping constant current without using negative feedback. To be more precise, we have implicitly used negative feedback e.g. to create a copy of disturbance (an "antidisturbance"). But we were just compensating the disturbing quantity (voltage, resistance etc.) without interesting in current magnitude. Do you agree?

Of course, the best solution is to monitor the current I while changing the resistance Ri or voltage Vvar. As we prefer to measure voltage than current, let's break the circuit and connect a constant resistor RI acting as a current-to-voltage converter. Then, using a reference voltage VREF and a zero indicator, let's keep a constant voltage VRI across the resistor RI. As a result, we get a stable current I = VREF/RI.

Experiment: Move for the last time the RL slider. I will connect a zero indicator between the VREF and RI; then I will move the Vvar (Ri) knob so that VRI = VREF. What about the current? Is it constant as before?

Assign this work to a transistor, an op-amp, voltage regulator etc.

Negative feedback transistor current source

How to build an op-amp non-inverting current source?

Voltage regulator acting as a current source #### Internal links to html pages and flash movies:

Analog electronics 2004 - Class 1

Analog electronics 2004 - Class 9

How I revealed the secret of parallel negative feedback circuits

Voltage causes current

Voltage-to-current converter

Passive resistive summer

Op-amp inverting summer