MosFet: Source-Bulk-Diode Problem...how to insure Bulk is at highest Voltage?

Hi all,
Lets say I intend to design a simple circuit that shoud connect a
(large) capacitor with a pulsed voltage-source, if and only if the
voltage of the pulsed source is at a higher level than the voltage of
the capacitor (the cap is charged). To do this I want to use a
(certain) MOS-Fet Transistor.

Now I've got the problem, that there is no signal where the potential
is allways at the highest level in the circuit. E.g. If the pulsed
source is charging the capacitor with a large current, there will be
quite a bit of voltage-drop across the MOS-Fet, and while the source
is at level 0 the transistor will be turned off an there will be a
voltage drop the other way.

So how schould I connect the bulk of the mosfet in a way that the
source-to-bulk-diode never becomes foreward biased (to high)?

I thought maybe a kind of "bulk-switch-circuit" could help? Maybe also
just a simple diode (with a foreward voltage lower than that of the
transistors-diodes)?

Any ideas, experience?

So far, Thanks for reading,
Regards,

Diego

 

You have a "MOS-Fet" with a separate bulk connection?

Post a schematic on a.b.s.e

Your description is clear as mud.

...Jim Thompson

 

Diego Stutzer wrote...

Which one?

Most MOSFETs we use these days are vertical or VMOS types, which
have three terminals. Generally the drain is the substrate, with
an intrinsic diode to the source. The pre-connected substrate
is also present for the popular lateral MOSFETs used in audio
amplifiers. Many of the old-style "ordinary" MOSFETs with four
terminals, like the 2n4351, etc., had a fourth terminal connected
to the case, not to the substrate. So in all these cases as a
designer you get to (have to) simply ignore the substrate.

If you're using a 4-terminal MOSFET with a substrate pin, in many
cases you can still ignore the substrate, and let it seek its own
voltage through occasional substrate-diode conductance and low
leakage-discharge currents. In rare cases you'll take charge of
this pin to set the voltage, to reduce capacitance, etc.

Yes, this is what some IC designers did, especially with CMOS
analog switches, e.g., as first shown in RCA's CD4066 datasheet.
Here they wanted to improve Ron as a function of signal voltage
and used a second analog switch to connect the n-channel signal
FET's substrate to the signal when the switch was on and to Vss
when the switch was off. This trick is still commonly used in
linear CMOS switches.

As explained, the necessary diode is already present.

BTW, Diego, it's not good to crosspost to so many groups.

 

Hi Diego,

Wouldn't a diode alone do that? Anyway, if you are concerned about the
substrate diode shorting your cap back to the source the only way I see
is to place a diode in series.

Win is right about too many cross posts. Also, I wouldn't go across
language barriers, at least not a lot. Except if you worked at my
favorite radio station when I was a student, where they happily jumped
between four languages ;-)

Regards, Joerg