Author = Frank V. Hughes
Field effect 
transistors can be classified between those which operate in the enhancement 
mode and the depletion mode. 
Enhancement mode means that the FET must be 
biased on to set the operating point for use as an amplifier (much like biasing 
a bipolar transistor).
Depletion mode means that it must be reverse biased, 
or biased off, to set the operating point (as in a valve amplifier).
The 
junction FET or JFET operates in the depletion mode and to reverse bias a JFET 
stage, it is only necessary to insert an appropriate value of resistance in 
series with the source electrode (much like cathode bias in a valve 
stage).
Figure 
1 shows a P channel JFET and an N channel JFET, each with a source bias resistor 
Rs. 
The only difference between the two circuits is the polarity in 
connecting to the supply rail.
Voltage is developed across Rs and applied the 
gate-to-source junction in reverse or depletion polarity.
Due to the reverse 
feedback, the drain current is stabilised at a value determined by the value of 
Rs.
Now instead of Rs in the P channel JFET, lets replace it with the source 
and drain of the N channel JFET, and instead of Rs in the N channel JFET lets 
replace it with the source and drain of the P channel JFET.
Now we get the circuit of 
Fig 2 and this is our Lambda circuit.
Connected in this way the two 
transistors interact with each other an produce the "negative-resistance" 
characteristic or backward bending curve. 
This is well known in the Tunnel 
diode and the Tetrode vacuum tube.
Because of the similarity of the Greek 
character LAMBDA to the shape of the "neg-resistance curve" the term 
Lambda-Diode was coined to describe the circuit element under discussion. 

To make a 
negative resistance oscillator we simply connect a tuned circuit in series with 
the Lambda circuit and the drain to source supply and set the supply voltage to 
a suitable level for oscillation to occur. 
The advantage of this 
type of oscillator is that two-terminal (i.e. no taps) inductance coils are all 
thats needed!
Since P-channel JFETS have a relativley shallow transfer curve, 
a bipolar transistor can be used in its place.
The circuit shown is that 
of a "gate-dip" oscillator, used to determine the resonant frequency of parallel 
LC circuits.
FET T1 and the bipolar transistor T2 form the Lambda diode. 
The "Lambda-diode" so formed is biased to the negative resistance region by 
means of P1 .
Experimentation would suggest that point is about 4 volt.
 (In a tunnel diode this point is about 1 volt).
The Lambda diode curve can be 
plotted at your test bench and by considering the neg-resistance region A-C: a 
resistance value of approx 600 ohm is obtained by taking the ratio of voltage 
change to current change for "A-C" curve section. 
From the point of view of 
RF currents: the base of T2 is connected to the drain of T1 and the gate of T1 
is connected to the collector of T2 (via 47pF capacitor).
With the wiper 
of P1 most +ve oscillator is inoperative and meter reading will be at a 
minimum.
The oscillator will work provided that the shunt resistance of the 
resonant circuit is at least greater the the -ve resistance value of the lambda 
diode.
This we established earlier as being about 600 ohm.
Even low-Q 
tuned circuits can have a high shunt resistance which means that the circuit 
oscillates so freely it can be hard to pick up the dip even when the test LC 
network is being "dipped".
Therefore advance P1 wiper till deflection just 
increases (oscillator is now running).
Also the shunt resistance of the LC 
circuit may be artifically lowered (by paralleling with a fixed resistor of 
approx 1.6K to 4.7K - Higher freq' least resistance - see table).
This allows 
this (very efficient) oscillator to operate with just enough feedback to produce 
a reliable output, and hence when used as a Dip-meter,
the "dip" phenomina 
will be more evident and not masked by the ability of the Lambda oscillator to 
produce full output even with a Low Q tuned circuit and the proximity of a 
nearby test LC network trying to absorb some RF 
energy!
REFERENCES:
Wireless World: Jan & Feb 1957 M G 
Scroggie (Cathoderay)
NEGATIVE RESISTANCE - MYSTERY OF THE BACKWARD BENDING 
CURVE
Elektor magazine October 1979. A modern grid-dip meter page 
10-22
Amateur Radio (Australia) January 1997 A dip meter using the Lambda 
diode page15