From: Winfield Hill
Subject: Re: MOSFET "Normalized On-Resistance" question
Date: 16 Dec 2002 20:27:50 -0800
Organization: Rowland Institute
References: <firstname.lastname@example.org> <email@example.com>
X-Newsreader: Direct Read News 4.20
John Muchow wrote...
> Regarding the tempco...
> Is there anything else other than the "Normalized On-Resistance vs.
> Junction Temperature" and "Max. Id vs. Case Temperature" graphs that
> I can use to determine the tempco of a MOSFET? The linear derating
> factor certainly affects what I can do, but it's not something I
> typically associate with a device's tempco.
Well, when it comes to that, there are a number of different
"tempcos." For example, Id Vs Vg is an important tempco for
the issue of paralleling MOSFETs. In the IRF1404 data sheet
this shows in Fig 3, Typical Transfer Characteristics, where
you can see that, for Gate voltages above 5.2V, the drain
current goes down with increasing temperature. This property
means that a FET that's experienced extra heating won't start
hogging the current from the other FETs sharing the load. But
sadly this nice feature usn't useful for linear applications,
because it sets in at currents above 200A, much too high for
us to use. At lower gate voltages and lower drain currents,
Id increases with temperature, which is a property that tends
toward thermal runaway troubles.
> I'm still trying hard to understand all of the graphs in these data
> sheets. ...
Before leaving the issue of tempcos, study fig 1 and 2, the two
Typical Output Characteristics curves. Note how the resistive
part of the 175C curves are shifted to the right compared to the
25C curves, indicating lower Id at high temperatures. But then
note the higher 125C currents for the 4.5V gate-voltage curve
at drain voltages above 0.7V, this means the Id tempco goes from
negative to positive in this region. Too bad they don't include
gate voltage curves below 4.5V, but you can imagine their traces,
which would show similar properties.
> John "another virtual beer is on the way" Muchow
OK, thanks. *hic* *hic*