12
\$\begingroup\$

I must be overlooking something but I really don't see the need to have a bulk terminal. I am studying the body effect whereby it somehow interferes with the threshold voltage. I don't fully understand it because I'm finding creating the terminal in itself counter intuitive.

I am told that it must be connected to the lowest or highest voltage in a circuit, depending on n or p channel. First of all, if this interferes with the threshold voltage, wouldn't my supply rails directly interfere with the mosfet since they could be 1V or 15V. Not to mention if there was noise in the power supply, it would directly transmit through, no?

What would happen if there was no body terminal?

Why does a BJT not need one?

Do enlighten me.

\$\endgroup\$
1

2 Answers 2

Reset to default
16
\$\begingroup\$

The body terminal is there whether we like it or not because of the construction of a MOSFET:

enter image description here

The gate voltage modifies the channel width. But the gate potential is measured relative to the substrate potential, and the channel is formed in the substrate material. Therefore the MOSFET behavior depends strongly on the substrate (body) potential.

If we did not make a contact to the body, it would be free to float, and we could not control the behavior of the transistor. We connect to the body to take control of this behavior. Normally we tie the body to either a most-positive or most-negative potential, like you say, because that gives the lowest |Vgs| relative to the appropriate supply and prevents forward biasing the source-body junction. If we did not contact the body, it would still be there, but we would not have control over its effect.

(MOSFET image by Wikimedia user CyrilB licensed under CC-BY-SA)

\$\endgroup\$
6
  • 1
    \$\begingroup\$ It may be worthwhile to note that the geometry of MOSFETs on a wafer is such that it's easiest to have the body terminals of many FETs attached to a common voltage separate from their source/drain terminals, but for larger MOSFETs the most efficient geometries require that the body terminal be attached to one of the other leads. \$\endgroup\$
    – supercat
    Commented Jan 10, 2014 at 19:09
  • \$\begingroup\$ Discrete BJTs don't have a body terminal because the body is one of the functional terminals. BJTs integrated into a wafer definitely do depend on the substrate potential for isolation. \$\endgroup\$
    – Dave Tweed
    Commented Jan 10, 2014 at 20:21
  • \$\begingroup\$ @DaveTweed, you're right, I'll remove the discussion about BJT's. Feel free to edit (or add your own answer) if you can work out how to explain what's going on in a BJT briefly. \$\endgroup\$
    – The Photon
    Commented Jan 10, 2014 at 21:23
  • \$\begingroup\$ I just want to add that some SOI (silicon-on-insulator) MOSFETs do not have a connection to the body. If I recall correctly, it is fully-depleted SOI (FDSOI) that has that characteristic. I don't know about FINFETs, they may also lack a body conneciton. \$\endgroup\$
    – Joe Hass
    Commented Jan 10, 2014 at 21:34
  • \$\begingroup\$ The reason for connecting the NMOS body to ground and the PMOS body to Vdd isn't for "best performance", it's to make sure the source-to-body junctions are never forward biased. The body effect can be used to tweak performance (either static power or speed) depending on how performance is defined. \$\endgroup\$
    – Joe Hass
    Commented Jan 10, 2014 at 21:36
0
\$\begingroup\$

Without the body we have 2 "on" junctions back to back in series so it won't conduct. The body is used to connect it to source.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.