[TriEmbed] Power FETs Vol 25, Issue 27

[Shane Trent]

Brian,

1) I expect the length of wire to the gate would NOT be a limiting factor
in your application. It might slow down the switching a little but it
sounds like it would not be an issue. I would use a resistor (100 ohm or
so) between the I/O pin and gate and add a large pull down resistor,
100k-1M ohm, between the gate and source to ensure capacitive coupling
cannot start turning on the FET.

2) You are correct that speed is a big reason to use a gate driver chip. If
you are happy with the speed and Rds that you are getting with your
existing logic drive you would not gain much by adding a gate driver.

3) The MOSFET gate is actually insulated from the Source and Drain by a
thin layer of oxide insulation (hence the Metal Oxide in MOSFET) that
insulates the gate and connected pin. So until failure of the gate oxide
(over-voltage to static discharge), you are rather safe connecting an I/O
pin directly to a MOSFET gate. The minimum protect required is typically a
small resistor between the I/O pin and gate to limit the pulse current to
the gate (the resistor can also reduce the tenancy of the MOSFET to ring or
oscillate during operation). I am not sure that you gain a layer of
protection by adding an expander chip, other than shorting out the wire to
the gate would overload the expander chip rather than the Pi.

Shane

On Thu, Jun 18, 2015 at 9:11 AM, Grawburg <grawburg at myglnc.com> wrote:

> Nice technical explanation that I will need to ruminate on, but I do
> understand it well enough to ask these questions:
>
> 1. If the length of wire between the Pi and the FET is much longer than a
> foot I'm guessing that using a driver is mandatory, especially with a FET
> like the one I'm using.
> 2. If fast response is not essential, since we're only talking about
> milliseconds, the driver may be unnecessary.
> 3. Suppose I use an externally powered (5VDC) I/O port expander (MCP23008)
> between the Pi and the FET? I gain a second layer of protection between the
> Pi and the load and now supply the gate with 5V.  Once again, I'm making a
> presumption that lightning-fast response is not essential. I really do like
> using these I2C devices.
>
> Brian
>
>


-- 
A blog about some of my projects.  http://fettricks.blogspot.com/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.triembed.org/pipermail/triembed_triembed.org/attachments/20150618/af0b31e0/attachment.htm>