question about guard ring vs body effect

Hello,

For nmos, usually guard ring is connected to a separate and quiet GND. But in this case, the source and bulk are not at the same potential. So there will be body effect that has influence on vth. So the guard ring reduces substrate noise, but increases body effect, am I right?

For analog design(like a PLL), if the current is very small, say less than 1mA. Is it okey to short guard ring and nmos source, or separate them ?


Any advice is appreciated.

Guard ring makes the body resistance where it counts
(i.e. right under the channel) lower. The potential of the
body relative to source is where the body effect lies.
You want no body current, otherwise you have base
current in the parasitic BJT and really, really care about
body-source resistance then.

If there is no body current then you don't really care
about the style or value of body tie resistance. But this
is not a good assumption to make, under some conditions
(such as pin injected substrate currents, or circuit designs
which forward bias a body node relative to source, or some
radiation environments).

I use BTS (body tied to source) devices all the time.
I prefer them to remote sub / well ties, because I can
trust them and the layout guy (if he's not me, which he
usually is). And in SOI you had better assert a body
potential explicitly because the substrate won't do it
for you.

Hello freebird,

Thanks for the explanation.

In my design, I also use BTS(body tied to source) for all the devices, and the design is almost done. But then I read threads that substrate is better to connect to a quiet ground. I don't have enough time to change the whole design, but I can add one quiet ground rail in the design.

So I am wondering if it will be better to have both approaches: devices are BTS connected and a quiet ground connected to substrate nearby. So that potential of substrate under the channel region will be stable, because the current injected by the source will drain out by the quiet ground.