# predictablity of change in Vgs vs change in Ids?

Discussion in 'Electronic Design' started by [email protected], May 5, 2007.

1. ### Guest

I am using a P-channel mosfet to act as a voltage controlled resistor
to soft-start a small motor, which draws a couple amps while running
with load. My circuit quickly (<1ms) gets the mosfet gate voltage to
the point where the mosfet is passing a milliamp or so, and then it
very slowly increases Vgs to turn the mosfet on over the next 500ms.
The circuit works okay, but it expects a certain relationship of a
given change in voltage resulting in a given change in current through
the mosfet, once the mosfet is "on" (1ma or so). It doesnt really
matter if the "beginning to turn on" voltage of the mosfet changes
(i.e. via temperature or mfg batch), my circuit handles that, but it
does matter if when the mosfet is on, (a few mA passing through it),
that its "voltage to current gain" changes. For instance, if it
tripled, then my soft start would effectively ramp up three times
faster, which could cause problems, and if it went down by a factor of
three, then my soft start would take three times longer, which is also
undesirable.

So my question is, how predictable is the change in voltage to change
in current relationship for a given p channel mosfet? Temperature and
manufacturing spread are the biggest concerns here I think, if at all
(hopefully not!).

The mosfet in question is the SUB65P04 (datasheet on mouser)

2. ### Phil AllisonGuest

** Makes you a complete idiot.

Drive the same mosfet with a PWM wave at about 5 kHz - bring the motor
voltage up to full from zero that way.

........ Phil

3. ### Guest

**** you asshole.

4. ### Phil AllisonGuest

** Drop dead - you trolling cunthead.

........ Phil

5. ### MooseFETGuest

The value you want is called the transconductance of the device. In a
given MOSFET, it will vary by a factor of about 1.5:1 over
temperature.

Although the maker doesn't give you a maximum spec for this, they do
give you the minimum of 20 and a typical value on the graph called
"transfer characteristics". Picking points off the graph, I see a
typical value of about 40 or 50. Based on this I'd expect the maximum
to be something like 100.

You may want to change your circuit so that it controls the voltage
ramp on the motor instead of the current.

6. ### MooseFETGuest

Ramping the voltage is a little better. The voltage more directly
controls the speed of the motor. The back EMF of a motor is generated
by its spinning. The back EMF is proportional to the RPM. The
motor's current is more closely tied to its torque.

If you don't need anything too exact, a simple circuit may do what you
need. Consider this simplified circuit:

Vcc
!
\
/
\
!
+---- Vout
----------+
! !
--- !
--- !!--
In -\/\/--+------!!
!!--- Gnd

I've drawn this as an N channel and with a resistor in place of the
motor. When the In signal goes high, the capacitor slows the change
in gate voltage. When the transistor starts to conduct, the change in
gate voltage slows down greatly. The negitive feedback via the
capacitor prevents the fet from swithing more quickly.

I real life, you would want a resistor and zener to protect the gate
of the MOSFET.