# IRF2804 Mosfet Question

Discussion in 'Electronic Design' started by Ray, Jan 7, 2004.

1. ### RayGuest

I want to build a 75 to 100 Amp PWM controller for a battery charger. I
ordered some IRF2804's from digikey, but was only able to get them in a
TO-260 case.

Question 1 What is the best way to mount them to a heat sink ?

Question 2 The leads seem awfully small to carry the 75 amps they are rated
for.... Is it only because the leads are short that you can get away with
such a small lead size ?

Ray

2. ### Robert BaerGuest

1. Direct mounting to a heatsink; smooth bright metal surface where
the device must mount to, and use a very small dab of heatsink grease
uniformily spread on the mounting surface.
Torque to specs.
2. Look carefully at the data sheet.
Note the rather large gate drive to guarantee saturation.
Then note the voltage drop across the device at that current.
Most *especially* notice the pulse curves on the datasheet with
respect to DC conditions.
And watch the temperature of the device(s) rather carefully!

3. ### RayGuest

The Problem is the TO-262 has no mounting Hole

Ray

4. ### Paul BurridgeGuest

Clamp it. Off-centre holes aren't much good for heat dissipation
anyway.

5. ### Winfield HillGuest

Ray wrote...
Oops! Caught by the old FET-current-rating trick! Power MOSFET
current ratings are generally theoretical numbers calculated from
the R_theta-JC thermal resistance and Rds(on) electrical conduction
resistance. We know that P = Id^2 * Rds and Tj = P * R_jc + Tc.

Often two current ratings are presented, one for a case temperature
Tc = 25C and the other for 100C. The 25C number is supposed to be
helpful for a quick evaluation of short-pulse capability, while the
latter is more useful for continuous-current conduction.

We can combine the above two equations to derive the theoretical
maxium drain current, Imax = sqrt [(Tj - Tc) / (Rds * R_jc)].

Let's play with some numbers for the irf2804. We alway set Tj to
the max, 175C, and the case to ambient, Tc = 25C. From the data
sheet, http://www.irf.com/product-info/datasheets/data/irf2804.pdf
we have R_jc = 0.45C/W, and Rds(on) = 0.0023 ohms at 25C, and also
0.0042 ohms at 175C (using a 1.82 multiplier taken from figure 10).

First let's calculate the FET's theoretical-maximum power-handling
capability, Pmax = (175 - 25)/0.45 = 333W. They say 330W, good.

Now let's try calculating the FET's Imax, shown on the data sheet.
Assuming the junction is 175C, we'll use the higher Rds(on) value.

Imax = sqrt [150 /(0.0042 * 0.45)] = 281 amps. They tell us 280A,
very good. Now let's try with Tc = 100C, so Tj-Tc = 75C, which is
half the difference. Taking the square-root into account, we get
Imax = 200A, just as they show. In this way we could derive the
Id(max) plot of figure 9. Except, of course, for the one piece of
hard reality in the figure, an actual 75A continuous-current limit.

So we see the simple-minded nature of FET drain-current maximum
values. They're completely theoretical, and therefore dangerous
to rely on for real design work, unless coupled with a complete
understanding. But they are useful numbers to compare with other
FETs evaluated in the same way, as is the industry practice.

Thanks,
- Win

whill_at_picovolt-dot-com

6. ### R.LeggGuest

There are two packages offered that do not provide a mounting hole; D2
and TO262. Obviously the mfr doesn't expect you to use them
interchangeably with TO220, without taking into account simple
differences in heatsink contact surface area mounting method and lead
available for manual soldering has it's advantages, regardless of the
mounting method intended in the final configuration.

Both D2 and TO262 were intended for direct solder bonding to a thermal
interface; typically a direct bonded copper on alumina or aluminum
substrate. This provides potentially lower thermal impedance to the
ancilliary heatsinking than can be provided by an isolation barrier
at the immediate package interface, no matter what isolation material
is used.

The source lead obviously needs careful attention if it's contact is
to provide reliable high-current service, regardless of the package
size.

Below 20mOhms, the 15 to 25nH typical lead inductance of these
packages can begin to dominate device performance at common conversion
frequencies. This is one reason why the drain backplate contact is a
prefered connection for higher current operation, and why some mfrs
offer modified source lead construction for similar devices.

RL

7. ### Robert BaerGuest

The IRF2804 comes in TO-220, TO-262 and the D2-PAK; i had thought that
you were speaking of the TO-220 package.
The TO-262, like the D2-PAK is meant to be soldered to a PCB or
similar substrate.
So, if you have a very good soldering iron, and some 1/16th thich (or
thicker) copper sheet, then you can use that sheet as part of a
heatsink.
According to the datasheet, the gate drive should be over 7 volts for
500mV drop at 100 amps.
Nothing i see shows DC operation; the best is a pulsewidth a bit
better than 10mSec, *single* pulse, Tcase=25C and Tjunction 175C.