Maker Pro
Maker Pro

MOSFET packaging

E

ErikBaluba

Jan 1, 1970
0
Hi,

I'am putting together a H-bridge and I have found some interesting SO-8
MOSFETs with combined P/N channels on the same chip.
The maximum power-rating for such packages seems to be 2.5W. But many of
these mosfets have a rating of several amperes continous current.

If I have a MOSFET like this with e.g. Ids continous = 4A, and Pmax=2.5W,
does this mean that Vds cannot be larger than 2.5/4 = 0.62V ?

Am I right to say that the ampare rating for such MOSFETs is not relevant
for driving motors, except perhaps small lov-voltage motors with low torque
and high speed?

Now, for TO-92 or I-Pak power MOSFET the watt-rating is suddenly very high.
What MOSFET packaging would cover the usage between these extremes? For
example, I have some very small Lithium-Polymer batteries with several
amperes discharge current. If I want to drive a 3V motor with 1A current
using such a battery I would need a mosfets that can handle about 4W to be
safe. Most of the TO-92 power mosfets I see seem to be completely overkill
for this, and I also want things to be as small as possible.


erik
 
D

DJ Delorie

Jan 1, 1970
0
Power dissipated by the MOSFET depends on its own Vds, not the drop
across the load. A 4A mosfet can pass 4A as long as the drop *across
the mosfet* doesn't exceed 0.6v when it's conducting. The drop across
the load can be whatever you want, as long as you don't exceed the
other parameters of the mosfet (like max off-state Vds).

If you have a 3v motor with 1a current, the *motor* dissipates 3
watts, but that doesn't mean the *mosfet* does. Let's say the on
resistance of the mosfet is 1.5 ohms. At 1A, it's a 1.5 volt drop.
That's 1.5 watts of power dissipated by the mosfet itself.
 
W

Walter Harley

Jan 1, 1970
0
ErikBaluba said:
Hi,

I'am putting together a H-bridge and I have found some interesting SO-8
MOSFETs with combined P/N channels on the same chip.
The maximum power-rating for such packages seems to be 2.5W. But many of
these mosfets have a rating of several amperes continous current. [...]

To add to what DJ said:

The basic idea of switching control is that the MOSFET is either turned on
all the way or turned off all the way. Power = current * voltage. So when
it's turned off, no current is flowing, so power is zero; when it's turned
on, current can flow but the voltage drop across the MOSFET is small (= Rds
* I) so power is still low.

So average dissipation in the MOSFET will be (Rds(on) * I * duty-cycle),
where duty-cycle is the fraction of time that the MOSFET is turned on for.

When you look at power ratings, keep in mind that the power has to get
dissipated as heat, and that heat has to go somewhere. So you need to
consider "thermal resistance", which is expressed in degrees C per watt.
The datasheet will tell you the junction-to-ambient resistance. For
instance, a TO92 case might have 200 C/W of junction-to-ambient thermal
resistance. That means that if the MOSFET is dissipating 0.5W, the junction
temperature (which is what matters) will be 100C hotter than the air around
the transistor. If the air around the transistor is 50C (and remember, if
you have your electronics inside a case, it will be hotter than room
temperature), that means the junction is 150C, which is probably its rated
maximum.

So you can see that it is usually thermal resistance that is the real power
limitation - the rated power maximum of the device is not the first thing
you run into, unless it is mounted to a very good heat sink.
 
E

ErikBaluba

Jan 1, 1970
0
....
To add to what DJ said:

The basic idea of switching control is that the MOSFET is either turned on
all the way or turned off all the way. Power = current * voltage. So when
it's turned off, no current is flowing, so power is zero; when it's turned
on, current can flow but the voltage drop across the MOSFET is small (= Rds
* I) so power is still low.

So average dissipation in the MOSFET will be (Rds(on) * I * duty-cycle),
where duty-cycle is the fraction of time that the MOSFET is turned on for.

Ok, but I assume you mean power dissipation is Rds(on)*I*I*duty-cycle?

....
So you can see that it is usually thermal resistance that is the real power
limitation - the rated power maximum of the device is not the first thing
you run into, unless it is mounted to a very good heat sink.

Thanks a lot, that was a very useful. I tend to skip pass those C/W
parameters in the datasheet :) I would think those ratings will not vary
much for different SO-8 MOSFETs, I will take a look around again.

As for Rds(on) ratings, it seems this follows the same principle as
resistance in a copper-wire, where tick wires have less resistance and thus
lower voltage drop and power dissipation? I noticed that only physically
large mosfets in TO-92 casing etc provide really small Rds(on) in the
milli-ohm range. The smallest Rds(on) ratings I found for MOSFETs in SO-8
casing was several Ohms, and with a much higher power dissipation as a
result.

erik
 
W

Walter Harley

Jan 1, 1970
0
ErikBaluba said:
[...]
Ok, but I assume you mean power dissipation is Rds(on)*I*I*duty-cycle?

Correct. Sorry, my mistake.
[...]
Thanks a lot, that was a very useful. I tend to skip pass those C/W
parameters in the datasheet :) I would think those ratings will not vary
much for different SO-8 MOSFETs, I will take a look around again.

You're right; for a given package type you won't find a whole lot of
variation. Laws of physics, and all that.

Note that for surface-mount components, free-air thermal resistance is
usually very high; but in practice, they dissipate their heat through the
leads, to the circuit board traces. So you will typically see a thermal
resistance spec for that, too; and it will often specify how big and thick
the traces need to be.
As for Rds(on) ratings, it seems this follows the same principle as
resistance in a copper-wire, where tick wires have less resistance and
thus
lower voltage drop and power dissipation? I noticed that only physically
large mosfets in TO-92 casing etc provide really small Rds(on) in the
milli-ohm range. The smallest Rds(on) ratings I found for MOSFETs in SO-8
casing was several Ohms, and with a much higher power dissipation as a
result.

Roughly speaking, yes, it has to do with the size of the channel. Of
course, they play all kinds of interesting tricks to improve Rds(on) while
keeping the die small. But the more channel area, the higher the
gate-to-source capacitance is (for a given breakdown voltage, anyway), which
introduces other problems; so there are a bunch of tradeoffs.
 
Top