Geoff asks a very good question that can easily elude the reader.
On page 4 there is this statement: “Observe the intercept point where the ideal curve extrapolation crosses zero volts. The corresponding current represents the diode’s saturation current and is marked in Figure 1 by the solid wavy line.”
For some reason I don’t follow this as the intercept point and the wavy line do not share any points.Geoff Peters (AB6BT)
For the QEX article, I made a decision to avoid a discussion of solid-state physics since the application is for computational purposes. For very low forward currents, there exists a region of recombination where the ideal diode IV curve departs from the ideal as illustrated in the figure by the drooping blue line as it leaves the green (the ideal). Manufacturers never, EVER publish this region of operation since it is first of all irrelevant but also because it has somewhat of an aspect of unpredictability.
Again, the actual region of recombination is irrelevant even for this discussion. But what is relevant is that the ideal region of operation for computational purposes was premised by Berkeley (originally, professor Ron Rhor and his students) upon a concept that a recombination region does NOT exist. Additionally, I can tell you, as a professional in the analog simulation industry, that Professor Rhor’s premise was a good one. We will never, in one mission years, be interested in simulation the recombination region of operation.
What I think you are asking about, Geoff, relates to a discussion of computational curve fitting. What is critical to the diode model performance is that it duplicates the published device performance for the ideal region and with reasonable success for the high-injection region. On a semi-log plot, the ideal region appears as a straight line until about a little bit better than 0.6 Volts of forward drop. Since the published IV curve will never show recombination current performance, we pick a data-pair for that lowest point and a second data-pair as high as we dare without getting into high-injection effects. We then write an equation representing a straight-line for a semi-log plot and extrapolate to the zero volts intercept. That intercept serves as the characterization saturation current parameter for the model.