# How to model a real diode 1N4148?

In my university EE lab we were working with a 1N4148 diode to rectify half a sine wave and measure the RMS value of the current in a simple circuit. To the left is the circuit we had to make and to the right it's the modeled circuit for theoretical calculations. However, from reading the datasheet (http://pdf.datasheetcatalog.com/datasheet/fairchild/1N4148.pdf) I find no reference not only to a model nor any mention of a resistance of 122 Ohm and the value of the DC voltage source of 0.48V. Is the model wrong, or am I just interpreting the datasheet wrong?

• Every model is just as complicated as it needs to be for its specific purpose. Two different models typically mean that the same thing was considered under different questions. For example, you can model a human as a heating unit doing an average of 200W heat output, or you can model it as an 80 kg mass of flesh to be accelerated, or you can model it as tough-surface bag of wet rags – each of these models fits a different purpose (building AC design, car design, scalpel design). None of them is "righter" than the other. – Marcus Müller Oct 18 at 0:22
• Unrelated, are you aware that Rd is shorted and Vd is backwards in the second schematic? – Hearth Oct 18 at 0:42
• the diode has a logarithmic change in voltage, as the current is increased; about 0.058 volts per decade of current change. Yu might use 0.66 volt at 10 mA, 0.6 volts at 1 mA, 0.54 volts at 0.1 mA, and a Rseries of 10?? ohms. – analogsystemsrf Oct 18 at 2:46

Typically datasheets concern themselves with worst-case values, and when you are doing modeling you at least want to start with nominal parameters. In the case of the Fairchild datasheet, no "typical" numbers are listed, so you're going to be looking at the graphs.

Figure 3 & 4 give you forward voltage vs. forward current, so the slope is dynamic resistance. The voltage at a given current can also be read from the graph, so subtract the voltage due to the bias current through the dynamic resistance at the operating point and you have the voltage source required.

Your voltage source + resistance model is inadequate to accurately describe the behavior with a large change in current however (it's okay as a small signal model linearized about an operating point). You can find a SPICE model for the 1N4148 and simulate the behavior or use the Shockley diode equation directly. The linked model shows a saturation current of 4.35nA and an ideality factor of 1.906.

• Would you say that for this case with a sine wave with a peak value of 10V, the model would be a good aproximation to reality? – Bidon Oct 18 at 0:32
• The error will be quite large in terms of diode voltage throughout the waveform but relatively small in terms of output voltage behaviour. Depends on what you are doing with the output voltage and what is varying. – Spehro Pefhany Oct 18 at 0:37
• It is a mere measurement to learn the basic concept of the RMS value, so maybe it would be a good approximation as the output voltage wouldn't be too affected. Thank you – Bidon Oct 18 at 0:52
• It’s slightly better than assuming an ideal diode with ~600mV drop. – Spehro Pefhany Oct 18 at 1:04

The datasheets usually don't contain models for simple generic diodes. Some manufacturers may give out spice model for simulation purposes, but they are far more advanced than simply a resistor with a voltage source.

The model you happened to create is only a model, and it is specific to the environment it is in now (source voltage and impedance, load resistance, DC bias, AC signal). If that model is given to you or it is calculated from actual measurements then it should be correct.

• Coming from a physics background that makes perfect sense, but my lack of experience in EE would you say that for this case with a sine wave with a peak value of 10V, the model would be a good aproximation to reality? – Bidon Oct 18 at 0:32
• @Bidon No. But it might be good enough, I don't know what you're using it for. It's better to use a proper diode model if you can, with the exponential V-I curve. – Hearth Oct 18 at 0:43
• In general, only worse models than this are ideal diode and ideal diode with voltage drop, so in general I would not call it a good approximation for a diode. The question is what is an acceptable quality for your case. Perhaps you can take a look at datasheet I-V curves to find out if the measured currents and voltages of the model somehow approximate with datasheet curves. – Justme Oct 18 at 0:49

The AC resistance is the ΔV/ΔI small signal changes that we call incremental resistance or Ri in Ω. This value is inversely proportion a=to all diode power ratings. e.g worst-case Ri=1/Pd 1Ω/W , best case 0.25/Pd

• 1 MOhm at 10 uA DC bias
• 50 Ohms at 1 mA DC bias
• 5 Ohms at 10mA DC bias
• 1 Ohm at 100mA DC bias
• 0.7 Ohms 800mA DC bias then flattens to 645 mOhms above this

Datasheets don't tell about the slope changes above saturation because the tolerances for VF at max current are too large and that is the load line slope above 100mA used as a switch.

But if you were modulating high current the change in voltage is actually less than the V/I load line slope. This incremental AC resistance, Ri or Rs or as effective AC series resistance or as I like to call it, ESR, like in cap ESR.

The log V vs log I is a straight line until mobility saturation is complete and the electrode semi bulk resistance then dominates the slope to an almost constant value.

Here is a set of 3 ON SEMI IV curves Here is my set of 1 curve with Ri shown below the load line for diode impedance vs current. Based on these parameters I should mention high power currents must be <1% duty cycle to achieve these voltages otherwise the NTC Vf coefficient shifts the load line to a lower voltage with rising temperature and the PN junction Ri reduces slightly and speed improves.