I have been working with enhancement MOSFET for a long time. But I have never seen any circuit using a depletion-MOSFET.
What are some typical use-cases of the depletion-MOSFET?
Indeed, they are not very widely used, but still have a few reason to be available.
if you put a resistor between source and gate, then you create a constant current source:
If the current increases, it increases the voltage drop across the resistor and therefore lowers the gate voltage, which will turn the mosfet off a little. If the current decreases, the mosfet turns on a little. This will always find equilibrium and you therefore have a current source with only two components, whose current solely depends on the resistor and gate threshold (not very accurate, though).
These supplies use a controller chip on the primary side (220V or 110V). The chip needs some low voltage to run (usually 10V), and this voltage can be provided by an auxiliary winding on the transformer in order to be efficient (if you power the chip by dropping the high voltage on primary with a zener, you'll waste some power which becomes significant at low load). This is fine, but when the supply starts, there is no voltage on the auxiliary winding yet, so the controller cannot be powered and it never starts.
So, somehow, you have to power the controller by dropping the high voltage, at least during startup. But, once it has started up, and the controller can be powered with the aux winding, you'd like to cut this current path which wastes power. If you do it with a depletion fet, it is very easy: basically, you just have to set its source to the supply pin of the controller, the gate to the ground of the controller, and the drain to the high voltage (this is a simplified view):
This way, when the controller is unpowered, the high voltage powers the controller (no voltage at the gate), and once the controller is powered, the high voltage path is interrupted (negative voltage at the gate). Every other way to do it with an enhacement mode fet would be less efficient (more components, more complex, more wasted power). This is why most standard depletion mode fets you can find are actually high voltage parts.
This application is limited to the protection of signals, or low-current supplies, because the depletion fets usually have very high RDSon. This is the typical circuit:
Even if the signal voltage goes too high, the gate will be kept at the zener voltage. The output will therefore not be able to go above Vz+VGSthreshold, because the mosfet would then stop conducting. It actually works like a regulator and clamps the signal. You can protect IC inputs with this, the only consequence in the nominal case being the RDSon of the mosfet (lower impedance than just a resistor and a zener).
Notice how the above circuit looks like a simple NPN regulator. There is one big difference, though: with the NPN regulator, the output voltage is at Vz-0.6V. With the depletion FET, the output voltage is Vz+VGSth. The clamped output is above the reference.
Another example of overvoltage protection usage, with a regulator:
The principle is the same as above, except we are using the regulator output directly as the reference fed to the gate (the zener can be avoided). This is where the fact the output of the FET is above the reference is useful: the reference being the regulated 5V, you know you'll have VGSth allowed for the regulator dropout.
So, since depletion FETS can be easily obtained for high voltage ratings, you can make a regulator able to withstand several hundreds volts easily (useful for mains voltage). Once again, just keep it mind this can be done for low currents only (a few tens of mA).
They have been used in logic ICs at a time, too (early 80's).
Basically, they were used as the high-level pass element, instead of the P-type FET now used in CMOS ICs. It acted mostly as a pull-up resistor whose value became higher when the output was low, to reduce power consumption and still having low impedance at the high-level state. Example with an inverter gate:
There are several app notes available for more information:
Depletion mode fets are useful in energy harvesting where very low voltage operation is wanted.A typical depletion mode fet will outperform a Si BJT and even do better than a Ge BJT .Sure depletion mode mosfets are a bit rare but from a production viewpoint they are less undesirable than Ge BJTs .Another use is Valve replacement when restoring Vintage radios .Audio valves can be easily found but Radio Valves are sometimes unobtanium .Small high voltage depletion mode mosfets have low gate capacitance making them potential candidates for RF/IF/Mixer.
How about a device that is neither enhancement nor depletion mode? Or is ambiguously one or the other?
A lot of CMOS processes have "native" transistors in them. These are transistors in which certain implants have not been applied and because of that have very low threshold voltages. In some processes this threshold goes negative (for NMOS) and thus is a depletion device.
These are present so that they can be used in bias circuits, pull ups/downs that go to the rails and in op-amps for Rail to Rail (R-R)operation. Although it is not necessary to have native transistors to get R-R operation.
In a bias circuit they are very handy so you can have active control during power up (these circuits come alive first) and also so you can increase operational range, for example a classical current mirror does not operate close to the rails (below Vth). You can use a ative device to control a normal device in it's subthreshold region of operation.
So even in todays world these devices are far more common than one would suspect.
As a note, the Wikipedia entry on these devices is wrong in stating that there are additional implants. While it is might be true in some cases , in about 5 different foundries that I am aware of, these devices have process steps removed.