Even More Forward Bias

If we know the voltage and current for a diode at any point on its I-V curve we can find any other operating point: Alternatively in Log base 10: Here we can see that for a decade change in current, the voltage drop across the diode will change by 2.3nVT, for a diode with n=1 this is approximately 60mV at room temp. If the temp increases, then the voltage drop will increase for a fixed current.

I1 = e (V2 V1 ) nVT I2
I2 V2 V1 = nVT ln I1

V2 V1 = 2.3nVT log
Log base 10

I2 I1

Cut-in Voltage
With the exponential dependence of the current on the applied voltage, the I-V curve in forward bias appears to start off with zero current near zero bias and then suddenly increase substantially. Of course there is a current near zero bias, but it is extremely small (pA or nA) The cut-in voltage is typically around 0.5V. This voltage drop will have consequences for circuits using diodes as we shall see later. A diode that is fully conducting is typically operating with biases of 0.6 to 0.8 Volts.

Reverse Bias and Breakdown

The reverse bias region is much simpler than the forward bias region.
If we go back to the diode equation, we can see that negative values for the applied bias result in an effectively constant current: Perhaps you can see where the term saturation current came from, in reverse bias the current appears to saturate. In reality though, the current in reverse bias is dominated by leakage currents, due to imperfections in the diodes. These leakage currents are about 1000x larger.
v nV e T 1 i = IS

i I S

Just like the forward bias region, even larger negative biases lead to breakdown. This is not covered in this model*. The voltage that this occurs at is called the breakdown voltage VZK where the Z stands for Zener (well talk about that later) and the K stands for knee meaning the point where the slope changes dramatically on the I-V curve.

* Its worth mentioning here a general issue. Just about every equation you see is an approximation. You should always keep in mind what its limits are!

Modeling a Diode
Though accurate, the exponential model is also the most difficult to use because of the added complexity of calculating logarithms. Lets look at a simple circuit with a diode and resistor to see just how bad it can get:
Well assume the voltage across the diode VDD is greater than 0.5V so we can use the simplified exponential model. The other equation that governs the circuit is the Kirchhoff loop equation

Assuming we know the diode parameters, we have two equations and two unknowns. But solutions are made challenging by the fact that one has its unknown sitting in an exponential. The solution to this problem is either a graphical solution or using iteration.

ID = ISe

VD nVT

VDD VD ID = R

Graphical Solution
The graphical solution to the problem involves plotting the two curves, one for the diode, the other for the resistor on the same graph. We refer to the resistor curve as the load line because in most cases we consider the loads on a system to be purely resistive. The intersection of the two curves is the operating point of the circuit. Note that the voltage along the X axis is the diode voltage, not the resistor voltage. Now try to imagine solving this problem if we had multiple loops and more than one diode!

Iterative Approach
The other approach to the solution of this problem is an iterative approach. It is impossible to just do a straight solution of the problem because the current and voltage need to be solved simultaneously. The approach to take then is an iterative solution, where a starting value is assumed for one parameter and the other is solved for. Next that solution is plugged into the equations in place of the initial assumption. This process is then repeated until the solution converges on a stable answer. Note, if the initial guess is off by a lot, many iterations will be required, or the results will run off away from the final solution.

Iterative Approach II
Determine the current ID and the diode voltage VD for the circuit we just looked at. VDD=5V, R=1k. Assume the diode has 1 mA of current and a voltage of 0.7 V And its voltage drop is 0.1V/Decade. First Iteration To start with, well calculate the current using the loop equation:
ID = VDD VD 5 0.7 = = 4.3mA R 1k

Second Iteration Well plug this new voltage back into the first equation again to get a new current

ID =

Now we want to find a new value for the voltage across the diode. Since we have the voltage drop per decade, well use that:

5 0.763 = 4.237mA 1k 4.237 = 0.762 4. 3

Now recalculate the voltage:

V2 = 0.763 + 0.1 log

I2 I = 0.1 log 2 I1 I1 I 4.3mA V2 = V1 + 0.1 log 2 = 0.7 + 0.1log = 0.763V I1 1mA V2 V1 = 2.3nVT log

There is very little change here, so we can stop.

The Need for Speed

As we can see, the iterative approach does yield answers with very few iterations. But still, if we have a more complicated circuit, this approach will soon bog down too. Obviously we need something less complicated to get rapid answers for designing a circuit. The solution to this problem is to simplify the diode model.
We gain speed by choosing a model with linear equations We loose our accurate description of the diode and must keep that in mind. But for many applications this will work just fine.

There are a few things we need to preserve in any simplification:

The proper turn on voltage The average conductivity of the diode when fully on, or at least what is appropriate for the particular small signal

The first model to consider is another piecewise-linear model. Recall that the ideal diode model is also a piecewise linear model, but it did not assume non-zero resistance or turn on voltage.

Piecewise-Linear
In modeling the diode, we want to maintain as much similarity to the diode equation as we can. If we look at the exponential curve for the I-V characteristics, we can see a decent approximation:
No current flow for biases up to a turn on voltage Ohmic behavior after that.

We can represent this diode with an equivalent circuit

Here is where we Would start worrying About deviations from The exponential model

Equivalent Circuit
To produce the output of the piecewiselinear model, we need certain features:
Zero conductivity until some turn-on voltage. The turn-on voltage must be above zero. Above the turn on voltage, the behavior is ohmic, with a specific resistance

So how do we construct an equivalent circuit? What components will produce this and in what combination?
First, we clearly need a non-linear component, otherwise well never get the switch in behavior at the turn-on voltage. A voltage source will shift the ideal diodes turn on point above zero volts A resistor will produce the ohmic behavior. Note that below the diodes turn on point, the ideal diode has infinite resistance ensuring zero current will flow.

Example
We can re-examine the simple circuit of the battery, diode and resistor using this piecewiselinear model. Here we need to specify the turn on voltage and the resistance of the diode Using the parameters from before, we get RD=20, and VDO=0.65V Now the current is described as:
ID = VDD VDO 5 0.65 = = 4.26mA R + rD 100 + 20

In forward bias, with the voltage across the diode being more than the turn on voltage (well check that later), the circuit reduces to two resistors and two opposed batteries Now that we know the current, we can find the diode voltage
VD = VDO I D rD = 0.65 + 4.26mA 20 = 0.735V

This is greater than the 0.65V turn-on, so our assumption of conductance was correct.

Constant-Voltage-Drop
In a further simplification, we can just look at the fast-rising part of the exponential curve and take that to be a vertical line. This probably seems too simplistic, but it is an effective model to use in the initial design phase, where the first point of concern is how much the voltage will be reduced by adding a diode. If we applied this to the previous exercise, we would end up with VD=0.7 Volts. The equivalent circuit looks like the piecewise-linear model, except there is no internal resistance.

Now Thats Just Crazy-Talk!

The ultimate simplification of the diode is to revert back to the ideal diode model, where the diode is a perfect conductor for forward bias and a perfect insulator otherwise. Why would anyone in their right mind go back to something so vastly different from a realistic diode? Well, if the applied biases are so much larger than the turn-on voltage we can ignore the voltage drop in the circuit due to the diode. But, with almost no extra effort, we can use the constantvoltage-drop model and get a more accurate answer. Its not totally useless though. We used it before to determine which diodes were conducting in a circuit, something that is good to know.

Small signal
There are applications where a diode is biased to operate at a point on the forward bias I-V curve with a small AC signal superimposed on it. Here we can have a linear model, if the AC signal is sufficiently small. The idea of the small signal model is we choose a resistance that is tangent to the slope of the diode curve at the DC Bias point. From the figure, you can see why it has to be a small AC signal: Any large deviations from the bias point voltage and the model will not be an accurate representation. Whats nice is we can replace the diode with a resistor!

Diode Small-Signal resistance

The effective resistance is: This is typically valid if the signal amplitude is smaller than
10 mV for n=2 5 mV if n=1 Note this comes from the fact that VT=25mV at room temp (yielding a fraction of 1/5 or less)

nVT rd = ID
iD rd = 1 vD i = I D D

We can also see that this resistance depends on what the bias point current is. You might notice that we didnt calculate a turn voltage for this model.
This is because we are only really interested in the AC signal. We trust the diode is in the forward bias region, otherwise our effective resistance will be very large. Note that the equation we have even accounts for this.

When we use this model, we eliminate all DC sources in the circuit and replace the diode with the effective resistance. This requires the DC analysis to be done first, but that makes sense because how else would we know the bias point?

Example
Lets try an example of this:
The power supplied to this circuit is 10V DC with a 1V AC signal. The resistor is 10K The diode has a 0.7 V drop at 1mA of current. (assume n=2)

We want to find the DC current and AC voltage across the diode By first assuming the diode to have zero resistance, we can see approximately what current will be flowing.
We can do this because we expect the resistor to be much larger than the effective resistance of the diode We get 1mA, so assume the voltage drop is 0.7V

Thus

ID =

10 0.7 = 0.93mA 10k

Example II
We see the current is very nearly 1mA, so our assumptions so far are OK. We can then get the diode resistance from the DC current we just calculated
rd = nVT 2 25mV = = 53.8 ID 0.93mA

To do the AC analysis, we can reduce the diode to the effective resistance that was just worked out. The AC voltage drop across the diode is found by the voltage divider rule. We end up with 5.35mV This is less than the 10mV, so we were safe in using the small-signal approximation

vd ( peak ) = vs peak

rd R + rd