Low (threshold)Vt vs High Vt for analog designs

Hello All,


What exactly is the difference between the low Vt and high Vt ?
which one is most suitable for analog applications at high frequencies (2-5GHz)

Thank you!!

At those frequencies you will probably prefer the low-VT
devices for their greater drive per width, since width is
driving Cdg. But I would not hold out much hope for
"classical analog" at those frequencies. What do you mean
by "analog applications"?

RF always wants low-ish impedances so a bit more leakage
will not be too much of a problem. But low power LF analog
would probably prefer the high-VT devices so as to avoid
operating in an area where subthreshold conduction's
variability (P, T) might make circuit performance variable
as well.

I am designing an ADC at high frequencies


As low vt has higher Power Dissipation compared to high vt so i thought hight vt is a good option.
But i would like to hear pros and cons who worked on them before

Your power dissipation depends only on drain current and
voltage. You can throttle a high-VT device to lower DC
current than a low-VT device. But neither, in any amplifier
that's trying to run that fast, will be at so low a current
density. Only static CMOS logic. And even there, in the
core / pipeline, very little sits still.

I have worked in technology with not one or two, but four
MOS VTs per species. Slow logic used the highest VT.
Even in some CMOS logic gates, when pressed for speed
I have used the "mid-VT" devices which are more akin to
a normal "low VT" in things like the internal clocked legs
of DFFs, tolerating the worse IDD in places where that
trade was worth it (400MHz, 3.3V CMOS - not an easy
play).

Low VT will give you more saturated operating range in
low voltage analog, somewhat. High VT might let you use
larger, better matching devices at low currents and low
frequencies (like your bias racks, if bias current is low).

You might choose to begin with standard (high) VT in
everything, and substitute low-VT where you run into
speed or headroom problems, reactively. I wouldn't try
to predict point-by-point, which to use ahead of any
analysis except for things where you know that either
max speed or min current is the overarching value. In
a mix, probably best to just pick a default and deal with
the exceptions as they are revealed.

I will go ahead with the high vt as per your input and will change the devices at critical path.