MOSFET 1962 Qualify Switching Speed Against Bipolar
# Qualify Switching Speed Against the Epitaxial Bipolar Transistor Just Completed: Losing One Comparison, Winning a Different One
## 1. Why the Same Test That Proved 1961's Improvement Now Proves This Device's Limitation
This step runs the identical pulse-test methodology the 1961 series' own Step 9 used to validate its collector-resistance improvement — a pulse generator, a device under test, an oscilloscope reading rise, fall, and storage time directly off the waveform — but points it, for the first time in this project's history, at two devices from two different series built in the same year, and the honest result does not favor the newer device. This device's switching time is set by how quickly the gate capacitance — the channel area's own $C_{\text{ox}}$ plus the overlap capacitance Step 4's alignment margin deliberately added — can be charged and discharged through whatever drive circuit controls the gate, and that capacitance, built from a structure with no self-aligned gate to minimize it, charges measurably slower than the epitaxial bipolar transistor's own collector RC time constant this project just finished qualifying. A direct head-to-head comparison, using the same instruments and the same methodology that proved an improvement every previous time this project has run it, this time documents a real limitation instead.
where $R_{\text{drive}}$ is the effective resistance of whatever circuit drives the gate, $C_{\text{ox}}WL$ the channel's own gate capacitance set by Step 2's oxide and Step 3 and 4's geometry, and $2C_{\text{overlap}}$ the parasitic capacitance Step 4 showed was unavoidable given this process's lack of a self-aligned gate — every term in this equation is a capacitance that must be charged before the device can switch, in direct contrast to the epitaxial bipolar transistor's own $\tau_{\text{switch}} \approx R_{\text{collector}} \cdot C_{\text{collector-base}}$, a comparison this step runs explicitly rather than merely citing.
## 2. Real Diagram: The Same Pulse Test, Two Different Devices Under It
The apparatus below is the identical setup 1961's own Step 9 used, run now on both device families side by side, with both measured waveforms recorded on the same time axis so the difference is a direct, visual comparison rather than a claim about two separately reported numbers.
## 3. The First Comparison in This Project's History That Doesn't Favor the Newer Device
Every earlier series this project has documented, when it ran a direct comparison against an earlier structure, demonstrated an improvement: the 1959 planar process improved on 1958's mesa reliability, the 1960 Micrologic series improved on discrete assembly's cost per function, and the 1961 epitaxial series' own Step 6 and Step 9 both demonstrated measured gains over the 1959 baseline they replaced. This step, using the same comparative discipline every one of those steps relied on, produces a different kind of result: this device, measured honestly against the bipolar transistor this project finished characterizing in the same year, switches slower, because every term in its own switching-time equation is a capacitance this structure cannot avoid charging. But the second measurement this step performs — the steady current each device needs merely to remain in its on state — inverts the comparison entirely, because this device's insulated gate draws no sustained current at all, where a bipolar transistor's base must be continuously supplied. The two devices are not simply better or worse than each other; they are optimized along different axes, and this step is the first in this project's history honest enough to measure both axes and report that neither device wins outright.
Step 9 does not conclude that this device is an inferior replacement for the bipolar transistor this project just finished; it measures exactly where this device loses and exactly where it wins, leaving the classification in Step 10 to decide what that split actually means for where this device belongs.