[1and1.com] [KO4BB Home Page] [Manuals Home Page] [KO4BB Wiki]

Lesson 5 - A Better Voltage Regulator (Discussion of Closed Loop Systems)

In the previous lesson, we designed a simple voltage regulator, simulated it and demonstrated its performance using SPICE. The circuit we studied is an actual circuit that could have been found in the schematic of a commercial piece of electronic gear.

We also found that while simple and effective, the load regulation left room for improvement. The output voltage could vary by about 200 mV when the load current changed from 5 mA to 100 mA. The temperature stability was good but not excellent, with a variation of about 80 mV between 25 and 75 degree C.

There are cases where the output voltage must be more tightly controlled.

In this lesson, we are going to design a better voltage regulator, with much improved temperature and load regulation characteristics.

Closed Loop Systems

Just like we improved the operation of the amplifier through feedback, we can improve the operation of the voltage regulator through feedback too.

However, this time it will be a little more complicated than simply adding a resistor.

The general idea about feedback is to feed a portion of the output signal back into the input, with the right phase and amplitude.

For instance, you can compare an open loop system to driving a car blindfolded (not recommended, but just imagine....) You might know by experience that a certain pressure on the accelerator will move the car at a certain speed.

However, if the car tows a boat, or if the car goes uphill, the same pressure will not yield the same speed. If you remove the blindfold, look at the speedometer and see the speed dropping, you increase the pressure on the accelerator. That's feedback.

The general idea behind feedback is to try and keep something constant in spite of changing conditions.

For our voltage regulator, we are going to use a technique called "closed loop" to more acurately control the output voltage. This will have a number of advantages.

A closed loop system compares the output of a system (in a general sense) to a reference and drives the control input of the system up or down when the two don't match.

A well known example of a closed loop system is the thermostat. It compares the actual temperature to a set point, and if the temperature is too low, it turns on the heater, and if it's too high, it turns on the air conditioning (or simply turns off the heater if the outside temperature is not above the set point.)

Another example of a closed loop system is the mother who watches her child explore new territory and only intervenes when the child goes outside of what the mother considers safe.

The opposite of a closed loop system would be an open loop system, which is more often used with teenagers: parents gives them the key to the car and say "don't speed" and hope for the best, because for better or for worse, teenagers need to have some freedom to learn self control and it would be counterproductive in the long run to not give them this opportunity.

However, going back to regulators, I prefer my regulators to be tightly controlled, and we are going to see how we can do that with a closed loop system.

Here is what a basic closed loop system consists of:

<ClosedLoop.png>>

This diagram shows the main elements that are in all closed loop systems, even though sometimes several functions can be regrouped within one physical device:

A Power Source
A Reference
An Error Amplifier which compares the reference to a feedback signal elaborated from the output
A Power Controller
a Feedback Circuit (which samples the output and feeds it to the error amplifier)

Please note that while I used an electrical schematic capture software to generate the Closed Loop System Diagram above, the elements of the system can but do not have to be blocks of electronic circuitry.

If we compare the closed loop system diagram to the schematic of our regulator circuit, we can easily identify the power source ( that would be voltage source V1), the power controller (that would be transistor Q1) and the reference (D1), but it is not so clear what are the error amplifier (it's the base=emitter junction of the transistor), the feedback circuit (it is a simple connection because one input of the error amplifier, the emitter, is tied to the output, and the other, the base, is tied to the reference).

Comparing to another closed loop system, the cruise control on your car, here are the elements:

Power Source: the engine
Reference: the speed you entered in the cruise control by pressing the "cruise" button
Error Amplifier: the cruise control hardware
Power Controller: a small servo-motor (sometimes electrical motor, sometimes operated by the vacuum from the engine) that drives the accelerator
Feedback Circuit: the speedometer that converts the speed into an electrical signal going to the cruise control hardware.

Here is how the simple voltage regulator works, in closed loop system terms:
When the load current increases, it tends to pull the emitter of transistor Q1 down (closer to ground).
Since the base is maintained at a fixed voltage by virtue of the Zener diode, that increases the voltage between base and emitter, which draws more current into the base, which causes the transistor to deliver more current to its emitter until it delivers the right amount.

The reason why this circuit has what we call "loose" regulation, is that the element that compares the output to the reference is also the control element, the one and only transistor in the circuit.

The lone transistor does not have enough gain to perform both tasks.

A Better Regulator

The first step to making a better voltage regulator is by separating the error amplifier and power controller functions.

<Regulator-5-1.png>>

In this circuit, Q1 is the error amplifier and Q2 is the power controller.

Please note the use of a new component: I1 is a Load (select Components->Load). It is a different type of current source and for all practical purposes is interchangeable with a current source.

This circuit has another feature due to the fact that the error amplifier and the power controller are separate devices: we can implement an actual feedback circuit, which is composed of R1 and R2. The real value of the feedback circuit here is that it allows to set the output voltage to something that may be different (always higher in this case) than the Zener voltage.

Here is how the circuit works:

At power up, the output voltage is 0 V and the source voltage is ramping up. The voltage at the feedback point is the output voltage divided by two (because R1 and R2 are equal values), so it would be 0 V also. At that time, the voltage on Q1's emitter does not matter because with 0 V on the base, Q1 will be in cut-off (turned off), so it will not draw any collector current, so all the current available from R3 will go to Q2's base, turning it on.

As Q2 turns on, the output voltage will rise. When the output voltage reaches 13.6 V, the voltage at the feedback point will be 13.6 / 2 = 6.8 V, so Q1 will have 0.6 V between base and emitter and it will start to turn on.

As Q1 turns on, it will start drawing current through its collector, reducing the current available to drive Q2's base.

Eventually, the circuit will stabilize with about 0.7 V across Q1's base-emitter junction, which should correspond to 13.8 V output voltage.

Let's run a simulation to see how far off we are. The plot below shows the output voltage as a function of load current:

<Regulator-5-2.png>>

Where the first circuit had about 200 mV of voltage drop when the load current changed from 5 to 100 mA, this circuit has only about 60 mV, a significant improvement.

Let's plot the base voltage along with the output voltage:

<Regulator-5-3.png>>

We can clearly see that the base voltage of Q2 is going up as the load current increases, showing the effect of the gain in Q1 trying to compensate for the drop in output voltage by driving Q2 harder. In the previous circuit, the base of the pass transistor was driven directly by the Zener and therefore was operating at a fixed voltage.

Let's see what the temperature stability is. Add a Spice directive as follows:


        .STEP TEMP LIST 25 50 75

and run the simulation again.

Here is the result:

<Regulator-5-4.png>>

The temperature variation is now 120 mV over the temperature range of 25 to 75 degrees C. The previous circuit had about 80 mV for an output voltage of 5 V, or 1.6%. This circuit has about 120 mV for an output voltage of about 14 V, or about 0.85%, so in relative value it's about twice as good.

IMPORTANT NOTE: the circuit modifications required by this example makes the comparison with the previous regulator circuit more like apples to oranges, since this circuit generates a 14 V nominal output voltage, and therefore has to run from a higher voltage than the previous circuit.

Ripple Rejection

Right click on V1, in the Small Signal AC Analysis box, type "1" in the AC Amplitude box. This tells SPICE to use this source for the AC analysis. Click OK then click on the Simulate->Edit Simulation Cmd and click on the AC Analysis tab.

Enter the following values:

Type of Sweep: Octave
Number of points per octave: 50
Start Frequency: 10k
Stop Frequency: 100Meg

Click "OK".

Click on the "Run" icon and select V(output).

<Regulator-5-6.png>>

This plot shows the ripple rejection directly in dB as a function of the ripple frequency. In this example, the ripple rejection is 24 dB up to about 1 MHz, then it drops down to 9 dB at 100 MHz.

Please note that the dotted line on the AC Analysis plot is the phase response of the circuit. We will study this later.

Exercises

In discussing this circuit, I noted that the output voltage is now a function of not only the Zener voltage but also the feedback circuit.
Try to modify the feedback circuit to obtain a nominal 9 V output voltage without changing the Zener diode.
Tip: you only need to change a single resistor value.
Resistor R5 was the pre-load resistor left over from the previous circuit. Is it necessary in this circuit?
Try to remove it and see if it makes a measurable difference on the output regulation. Why?
Compare the results of the ripple analysis with the ripple rejection of the circuit of lesson 4.

Click here to see the answers.

Conclusions of this lesson

By adding a gain stage, we improved the load regulation.
By using a separate error amplifier, we can adjust the output voltage by changing resistor values.

In the next lessons, we will learn how to improve temperature stability.