Maker Pro
Maker Pro

Heatsink for DPAK

P

Pasquale

Jan 1, 1970
0
Hello to everybody,
anyone of you knows how to fix an Heatsink for DPAK on PCB ?

Thank you,
Pasquale.
 
J

Jim Yanik

Jan 1, 1970
0
The DPAK is designed to be soldered to the surface of a PCB.
If you want to use a metal heat sink, you should probably
look for a TO-220 version that is made to be mounted on a
heat sink.

or maybe you could solder it to a copper HS and solder that to the
PCB....if the leads will still reach. you could rout a hole in the pcb for
the copper HS and glue it in flush with the PCB.
 
E

Eeyore

Jan 1, 1970
0
Pasquale said:
Hello to everybody,
anyone of you knows how to fix an Heatsink for DPAK on PCB ?

You mount it on a large copper area. Ideally on both side of the PCB
with some decent size vias joing the two.


Graham
 
D

David L. Jones

Jan 1, 1970
0
Pasquale said:
Thank you very much.
I'm gonna do a copper area large enough in order to fix the DPAK and the
Heatsink for DPAK. Do you think is it ok ? I have a MOSFET DPAK where the
drain should lies on the copper area. What do you think is the best way
to solder it ? Remember that I should solder it by hand.

You may have a bit of trouble soldering this by hand. They are designed for
reflow soldering.
What equipment do you have available?

BTW, have you done the thermal calculations to figure out what size heatsink
you actually need?

Dave.
 
P

Pasquale

Jan 1, 1970
0
You may have a bit of trouble soldering this by hand. They are designed
for reflow soldering.
What equipment do you have available?

Actually I don't have really good equipment (I have the stuff to solder
by hand, "regular" equipment), for sure I don't have a reflow solder
station. I know that is gonna be difficult but how can I manage to solder
it by using "regular" equipment ? I need suggestions in order to get the
best result with the equipment I have.
BTW, have you done the thermal calculations to figure out what size
heatsink you actually need?

Not at all. I've seen some heatsink and I though to take the best one. I
know that it's not good choice but I don't have any idea about the
calculation. Could you help me ?


Thank you,
Pasquale.
 
E

Eeyore

Jan 1, 1970
0
Pasquale said:
Thank you very much.
I'm gonna do a copper area large enough in order to fix the DPAK and the
Heatsink for DPAK. Do you think is it ok ? I have a MOSFET DPAK where the
drain should lies on the copper area. What do you think is the best way
to solder it ? Remember that I should solder it by hand.

Slightly tricky as they're normally soldered using paste. However if you do
it carefully you may be able to 'flow' some solder under the package but not
that won't work with leadfree solder because it has no surface tension to
talk of.

Graham
 
R

Richard Henry

Jan 1, 1970
0
Slightly tricky as they're normally soldered using paste. However if you do
it carefully you may be able to 'flow' some solder under the package but not
that won't work with leadfree solder because it has no surface tension to
talk of.

I wsa under the impression that lead-free solders have a slightly
higher surface tenson than classic tin-lead solders, which, judgeing
from the context, may have been what you meant anyway.
 
T

Tim Shoppa

Jan 1, 1970
0
You may have a bit of trouble soldering this by hand. They are designed for
reflow soldering.

True, that's what they're designed for, but for a part with a pad more
than a few mm by a few mm (for example the TO-252 has a pad 6.5 x 6
mm, quite big by most SMT standards) I find that a plated-through hole
big enough for my soldering iron tip lets me do a very decent job of
getting the part attached to the board. I'm sure there's some loss off
cooling ability because there's no copper under part of the pad but
I'm not pushing the limits.

Tim N3QE
 
P

Pasquale

Jan 1, 1970
0
True, that's what they're designed for, but for a part with a pad more
than a few mm by a few mm (for example the TO-252 has a pad 6.5 x 6 mm,
quite big by most SMT standards) I find that a plated-through hole big
enough for my soldering iron tip lets me do a very decent job of getting
the part attached to the board. I'm sure there's some loss off cooling
ability because there's no copper under part of the pad but I'm not
pushing the limits.

Tim N3QE

I'm really interested in what you have done. Could you, please, send me
some pics or something in order to understand better ? Or some links
where I can check your procedure ?


Thank you,
Pasquale.
 
T

Tim Shoppa

Jan 1, 1970
0
I'm really interested in what you have done. Could you, please, send me
some pics or something in order to understand better ? Or some links
where I can check your procedure ?

No pictures handy, but: just put a plated-through hole underneath the
thermal pad in a rectangle of copper. To solder the device to the
plated through hole, turn the board over, stick the soldering iron
through the hole to heat up the device and the board, when it gets hot
enough to melt solder all around, stick in the solder and flood the
hole. Solder will wick between the device and the copper pad.

I used this for a low-scale production AD9954 DDS a few years back.

Having the board pretinned helps a bit if you aren't used to it.

Having the hole of a good size for your soldering iron tip really
helps. If not, you won't directly heat the devices thermal pad, just
the board.

Tim.
 
H

Hammy

Jan 1, 1970
0
Actually I don't have really good equipment (I have the stuff to solder
by hand, "regular" equipment), for sure I don't have a reflow solder
station. I know that is gonna be difficult but how can I manage to solder
it by using "regular" equipment ? I need suggestions in order to get the
best result with the equipment I have.

I had a similiar question a while ago heres what I did. It's not ideal
but it worked.The poor mans reflow method.

All I do is put a dab of solder on the draintab of the mosfet,and some
solder on the drain portion of the footprint.Press the mosfet down and
heat the drain tab at the same time.Keep going around heating the
drain tab that extends above the case,and the drain part of the
footprint.I used a small insulated screwdriver to keep pressure on the
FET,while heating.

It took a few tries to get it down, but once you get the feel for
it,it does the job in a resonable time.Just dont overdo the solder.

Good luck
 
P

Pasquale

Jan 1, 1970
0
No pictures handy, but: just put a plated-through hole underneath the
thermal pad in a rectangle of copper. To solder the device to the plated
through hole, turn the board over, stick the soldering iron through the
hole to heat up the device and the board, when it gets hot enough to
melt solder all around, stick in the solder and flood the hole. Solder
will wick between the device and the copper pad.

I used this for a low-scale production AD9954 DDS a few years back.

Having the board pretinned helps a bit if you aren't used to it.

Having the hole of a good size for your soldering iron tip really helps.
If not, you won't directly heat the devices thermal pad, just the board.

Tim.

Thank you very much. I'll try your method. What about the heat ? I'm
trying to do some calculation if I need a heatsink or not. How did you
manage that ? I've see that I should use the Rthjc, Rthja and so on but
maybe a bigger copper pad will work. What do you think ?


Thank you,
Pasquale.
 
D

David L. Jones

Jan 1, 1970
0
Thank you very much. I'll try your method. What about the heat ? I'm
trying to do some calculation if I need a heatsink or not. How did you
manage that ? I've see that I should use the Rthjc, Rthja and so on but
maybe a bigger copper pad will work. What do you think ?

Thank you,
Pasquale.

First you need to know how much power your device is going to
dissipate. This value will depend on how it's used in your circuit.
If you don't know how to calculate this far then I suspect you may
have more to worry about than whether or not you can solder a
heatsink!

Dave.
 
P

Pasquale

Jan 1, 1970
0
First you need to know how much power your device is going to dissipate.
This value will depend on how it's used in your circuit. If you don't
know how to calculate this far then I suspect you may have more to worry
about than whether or not you can solder a heatsink!

Dave.

I know, more or less, how much power my device will dissipate. For the
MOSFET should be something like Pd = Rds_on*I^2. For this reason I'm
trying to understand how to choose the right heatsink,


Thank you,
Pasquale.
 
E

Eeyore

Jan 1, 1970
0
Pasquale said:
I know, more or less, how much power my device will dissipate. For the
MOSFET should be something like Pd = Rds_on*I^2. For this reason I'm
trying to understand how to choose the right heatsink,

If it's switching at high speed you also need to add the switching losses.

Graham
 
D

David L. Jones

Jan 1, 1970
0
Pasquale said:
I know, more or less, how much power my device will dissipate. For the
MOSFET should be something like Pd = Rds_on*I^2. For this reason I'm
trying to understand how to choose the right heatsink,

The device should have a degC/W figure for dissipation into ambient (no
heatsink).
If that temperature rise is acceptable in your application then no extra
heatsink is needed. In practice because the device will be soldered to the
board, it will have some extar heatsink already. If that temperature rise is
not acceptable then you have to start looking at thermal resistance figures
of the package and heatsink.
Generally, you can get a ballpark figure by simply taking the heatsink
degC/W figure and ignoring the thermal resistance losses (junction to case)
in the package etc. The heatsink I quoteed is 25degC/W, not marvelous, but
it helps.

Often with DPAK packages, a PCB heatsink is enough, and there are various
ways to improve this with layout. But PCB heatsinks can be harder to
calculate a figure for.

Dave.
 
P

Pasquale

Jan 1, 1970
0
If it's switching at high speed you also need to add the switching
losses.

Graham

Then I think it's gonna be something like that :

fs = PWM freq.
tc_on = switching time off-on;
tc_off = switching time on-off;
T = PWM period;
t_on = time when the PWM in high;

P = 1/2*(Rds_on*I^2)*fs*(tc_on+tc_off)+Rds_on*I^2*(t_on/T);

Correct ?


Thank you,
Pasquale.
 
Top