Self Cascode and single transistor

self-cascode composite transistor

I hope the question is in the right place.

Can anyone explain the difference between using self cascode structure and using a single transistor that has larger L.

What I see is that the self cascode can have same equivalent parameters (W/L & β ) but at lower output capacitance, I am not sure?

The self cascode, or composite transistor, for DC looks very similar to a single transistor. There will be some AC differences, associated with the parasitics that exist on the composite device that do not exist on the single device, which can degrade the speed of the structure.

If the W is the same, then the transistor will look (for DC) like a transistor that has the sum of the lengths.

If the W is wider on the "cascode" device, then there is some DC output resistance improvement. Using a relatively narrow and long device, along with a wide and short “cascode” device can provide an output resistance enhancement in a smaller size than with a single device. For example, a 1u/10u device with a composite 10u/1u transistor might respond (again, to DC) similar to a 10u/100u device, but with a considerably smaller area. While this is not a huge difference in output resistance and stage gain, it might be sufficient for changing a design that does not quite meet gain specifications into one that does meet the specifications.

An additional (and probably more critical) use for the composite transistor is that the intermediate node between the two devices is available for connection. This node can be used as a moderately low impedance current summing node, like in a folded cascode structure. It can also be used as a point for miller feedback that helps to remove the feed-forward zero.

So , is this mean if it is only used as an ordinary amplifier the only advantage is area ?

The savings is in the area of the device for the same output resistance and W/L ratio.

There are things that come along with this:
Gate capacitance is smaller (faster current mirror)
Offset will be worse than equivalent single transistor

I would consider the composite transistor as something between a single transistor and a full cascode, but closer in performance to the single transistor. I have seen that simulation can be inaccurate with a composite transistor (usually simulating higher output resistance/gain than reality).

The composite transistor can be very useful in the following situations for an "ordinary" amplifier:
- Output voltage range near the supply rail, where a true cascode transistor will go linear and stop benefiting. The composite transistor takes nearly the same headroom as a single transistor.
- You need a few dB of additional gain or output resistance from of a circuit. A properly designed composite transistor will boost output resistance slightly.

Thanks a lot JPR.

Are there other combinations (but formed from ordinary transistors; not bulk driven or floating gate) other than self cascode that works fine in low supply circuits ?