Tunnel diodes

Tunnel diodes are used in various circuits in Tek gear made from the early 1960's until the 1980's.

Applications

Tunnel diodes were used where it was desirable to have fast and clean switching between two states. They were used in

trigger circuits as Schmitt triggers,
sweep and timing circuits as flip-flops,
pulse generators for converting a slow-rise signals to fast-rise pulses,
countdown/sync circuits

Issues of Drift and Failure

Tunnel diode characteristics (peak and valley voltages and currents) tend to drift. Usually this can be handled by adjusting the surrounding circuit. Sometimes tunnel diodes completely fail. Replacement usually involves scavenging a similar tunnel diode from some other device. There are some people in the Tek community who may have some tunnel diodes they can sell. Germanium tunnel diodes are extremely sensitive to overheat, especially at soldering work. Be aware and use low melting solder and appropriate tool to protect the body from overheating!

Relevant Distinguishing Parameters

Many different types of tunnel diodes were made. The primary parameter that describes one is the peak current, which is the current at the top of the hill in the I-V curve. The other two relevant electrical parameters are the capacitance of the diode and whether it is made of GaAs or Ge. In some circuits, another model of tunnel diode can be substituted with only minor modifications to the surrounding circuit. Stan Griffiths describes such a modification here:

http://www.reprise.com/host/tektronix/reference/tunnel_diode.asp

Emulation using Common Parts

A common question is whether an electrical equivalent to a tunnel diode can be made out of modern, available parts. It is not hard to emulate the I-V curve, but there is no known circuit that can be made from available parts that has right I-V curve and the high switching speed of the real diode.

Testing a Tunnel Diode

Before concluding that a tunnel diode is bad, it is important to be sure that it has been measured correctly. A high resistance reading on a DMM indicates that the diode is bad. A low resistance on a DMM and a low voltage on a diode tester are both normal when measuring a tunnel diode. A more thorough test of a tunnel diode is to drive it through a resistor with a ramp voltage source while observing the voltage across the tunnel diode. The resistor should be calculated so that the peak current just exceeds the peak current that the tunnel diode is rated for. Of course if a curve tracer is available, it is great for measuring the I-V curve of the diode.

Tektronix 571 curve tracer run of 10mA tunnel diode.

10mA tunnel diode mounted in Tektronix 571 curve tracer fixture.

Modeling

The fast switching action of the tunnel diode can be understood by modeling it as a nonlinear voltage controlled current source (VCCS) in parallel with a small parasitic capacitor. The nonlinear VCCS is controlled by the voltage at the terminals of the diode and is responsible for the S-shaped I-V curve. (Alternatively and equivalently, it can be modeled as nonlinear resistance. However, the nonlinear VCCS model might be preferable because it avoids the confusing notion of negative resistance.) Consider a tunnel diode biased by a DC current source that is slowly brought up from zero to a current just a few microamperes less than the diode's peak current. The quiescent voltage will be just less than the peak voltage. Note that the I-V curve is nearly horizontal at this point, and therefore the incremental resistance of the diode is very high at this point. For simplicity, we can assume that the incremental resistance is infinite at this quiescent point.

(The following section will be clarified in the coming edits.)

Estimating Switching Speed

Now that we have established the initial bias conditions, let's look at the event when the tunnel diode switches state. Assume that the triggering signal is coupled to the tunnel diode through a resistor. The current through the resistor adds to the current from the DC current source. Since we are assuming that the incremental resistance of the diode is infinite at the initial bias point, all of the current due to the trigger signal flows into and out of the diode's capacitance. If enough charge is added, the instantaneous voltage across the diode will be in the second region, where the slope of the VCCS function is negative.

Once the diode enters the second region, increases in diode voltage cause decreases in diode current. Applying Kirchhoff's current law at the node where the diode meets the DC current source, we can see that the current entering the parasitic capacitor at any instant is the difference between the DC current source and the nonlinear VCCS current at the this instantaneous voltage. We can use this fact to estimate the switching time of the tunnel diode. (The shape of the transition can also be estimated.) As an example, let's take the case of a tunnel diode with 10mA peak current and 5pF capacitance. A first-order estimate of the switching time can be made by assuming that to make the transition from <math>V_1</math> to <math>V_2</math>, a certain amount of charge needs to be added to the parasitic capacitance of the diode.

From <math>Q = C*V</math>, we know that

<math>\Delta Q = C * \Delta V</math>, which is

<math>\Delta Q = C * (V2-V1).</math>

We can assume that

<math>V1 = 65mV</math>, and

<math>V2 = 465mV</math>. So,

<math>\Delta Q = (5*10^{-12})*0.4</math>, which is 2 picocoulombs.

Now we bravely assume that the charging current during the transition is constant, and is half of the peak current. 5mA is 5 millicoulombs per second.

<math>\frac{2 * 10 ^ {-12} coulombs}{5*10^{-3}amperes} = 0.4 nanoseconds.</math>