A circuit I'm designing requires electrostatic discharge (ESD) protection on the I/O lines of a microcontroller, and the solution that keeps popping up in my research is to use a transient voltage suppression (TVS) diode. Although their operation is easy enough to understand, and I've had experience with similar devices like zener diodes, I still wanted to get a feel for how they worked. Sometimes you can be sure you know how something works but it doesn't become concrete until you see it on a scope screen.
TVS diodes are basically avalanche diodes designed to conduct large peak currents for a short amount of time. When forward biased, the device will start conducting at about 0.7 V, just like a standard diode, but when reversed biased the diode won't conduct significantly until the breakdown voltage is reached. The breakdown voltage is however adjustable in manufacturing from a couple of volts to hundred of volts. By placing a reversed biased TVS diode between ground and a point in a circuit, you can very quickly clamp the voltage across the device to a safe level.
What if you wanted to protect a signal line that carried an AC waveform? Depending on the orientation of the diode, the clamping action would work for either positive or negative signals, but not both. When the polarity of the signal reverses it would clamp the waveform to the 0.7V forward voltage of the diode and possibly interrupt the signal. To overcome this you can use two diodes back to back. The clamping voltage for either direction will be equal to the clamping voltage of one diode plus the forward voltage of the other. You can get this configuration in a single package called a bidirectional TVS diode, although manufactured, differently the principle is the same. It's this kind of diode that I happened to have laying around.
To improve my skills I've been practising soldering surface mount components on an old dial up modem (what else are they good for) and noticed that near the power input is what looked like a TVS diode.
|Modem with TVS diode near the power input jack|
You'll have to forgive my camera work. The photos were taken in low light while holding a jewellers loupe over the camera lens. They didn't come out too bad for a first go.
The part designation on the silk screen was GAP3 which may indicate that originally some type of spark gap was to go here. Who knows.
After removal, I was able to identify it as a P6KE27C diode. The data sheet shows it is a bidirectional TVS diode with a breakdown voltage of 27V and a clamping voltage of 37.5V. This means that conduction will start at 27V, and at peak current the voltage across the diode will be 37.5V. To verify these characteristics I put together a basic circuit to plot the current voltage curve of the device.
By probing the V1 and V2 points on a oscilloscope set to XY mode, and then sweeping the floating voltage V, you can obtain a crude IV curve for the device. The current through the TVS is equal to -V2/120. Although I'll be able to check the breakdown voltage using this method, I can't test the clamping voltage because those conditions are only valid for short durations, if I were to apply the maximum current at the clamping voltage continuously, the device would be destroyed.
|TVS current voltage curve|
The plot came out as expected, the current is shown inverted on the vertical axis, with a scaling of 100mV/120 or 830 uA per division. The breakdown voltage is obvious on the plot at approximately 27V and agrees with data sheet. The bidirectional nature of the device is also evident from the symmetry of the plot.
I now feel more comfortable using these devices in a design. ESD protection is still a bit of a mystery but I'm slowly piecing things together from what I've found online.