filling vias

Ok then, go with the Reverse Lead DPAK. The point is there is no real need to deal with the issues of getting a PCB to conduct a lot of heat. Just mount parts intended to be heat sinked with something other than

1 oz copper.
--

Rick
Reply to
rickman
Loading thread data ...

I am engineer. Hear me roar.

formatting link

Reply to
John Larkin

Of course, I want to manufacture stuff, namely pick-and-place a lot of mosfets on a PC board. No manufacturing weirdness.

Reply to
John Larkin

That's a sweet part. Too bad there isn't a wider selection.

formatting link

(Note: I assumed John wanted a FET, just as a for-instance. He never said it.)

Cheers, James Arthur

Reply to
dagmargoodboat

m?

ide

-fill, and

go for

e

ured about

a

rse.)

has

ND

CB

vias,

e-down.

He'd need a better part than that.

20x.025", solder-filled, 1.6mm board = 2K/W. 20x.025" copper-filled, .8mm board yields .25 K/W.

About .9 K/W for a PowerSOT-8 I recently used. Total Rj to pad ~=1.2K/W.

Through-hole is so "last century." Why not just use 4cx350s?

High-power thermal stuff with SMD can be very demanding. No disrespect, but if you've done it, it's not showing here.

Cheers, James Arthur

Reply to
dagmargoodboat

like

measured about

was a

e worse.)

ay has

AND

T

Suggests a Highland Tech. Halloween theme, eh?

Cheers, James Arthur

Reply to
dagmargoodboat

I'm not the one who picked DPAKs requiring the hassles of specifying special board designs. I prefer to use more conventional technology when it serves the purpose very well. As it turns out the Reverse Lead DPAK would work perfectly according to the OP. So why bother with the copper filled vias when they are an unneeded resistor to heat flow?

The idea that a technology that has been in use for decades is naturally inferior is a bit pretentious. 99% of the technology you use is "last century".

--

Rick
Reply to
rickman

hem?

d side

ld

w

er-fill, and

d go for

.

ike

asured about

as a

worse.)

ay has

AND

PCB

d vias,

ide-down.

K/W.

h

The concept is excellent. We don't know if there are suitable parts.

No, it was funny. Like using fire-bottles. We are allowed to have fun, ar en't we?

If you want to hand-assemble boards, that's certainly your privilege.

Cheers, James Arthur

Reply to
dagmargoodboat

Like a DPAK that's 0.43 K/W maybe?

Reply to
John Larkin

..and that has the distinctive advantage that a solder "pool" like i mentioned cannot happen.

Reply to
Robert Baer

A thinner PCB helps a lot; half thickness means double thermal conductivity; 14 mil is super.

Reply to
Robert Baer

them?

d side

uld

ew

der-fill, and

ld go for

d.

like

easured about

was a

worse.)

hay has

p AND

s PCB

ed vias,

side-down.

0.45 K/W was as good as I could do offhand (D2PAK)...
formatting link
df

Cheers, James Arthur

Reply to
dagmargoodboat

On Monday, October 5, 2015 at 11:21:19 AM UTC-7, John Larkin wrote: [about cooling surface mount packages through the board]

Or, go low-tech, and instead of printed wiring on plastic, use printed wiring on porcelainized steel. It's a mature technology, and tolerates a bigger temperature range, if your other components can handle it.

Reply to
whit3rd

The thing that determines theta is mostly the size of the silicon. A DPAK is about as good as a TO220, although the TO220 might achieve a little more lateral spreading, and the mounting hole limits that.

A lot of DPAKS are rated Tjc mounted on some PCB copper patch, which isn't Tjc at all. Some data sheets say SEE NOTE without providing such note.

Ricky is wrong as usual.

Reply to
John Larkin

Do people make PCBs on porcelainized steel? That would be cute. These resistors are great:

formatting link

We used them in a pulsed-load situation that was destroying wirewounds.

Reply to
John Larkin

Thought about this last night. If you use Lasse's slick upside down DPAK idea, the DPAK's contact patch is only 4.1x4.3mm. So, if 0.2K/W is correct for one square inch, the pad would add 7 K/W for a DPAK-sized contact area, yes?

Handy list of thermal resistances for MOSFET packages:

formatting link

Cheers, James Arthur

Reply to
dagmargoodboat

Yup, about 7 is right. Kinda high for my application, which might run

20 watts per device maybe.

The mechanics don't look very appealing to use the upside-down DPAKs.

Neat. They show generic D2PAKs as having less theta than TO220s.

Their SOT223 number looks pessimistic.

Reply to
John Larkin

Interesting that John is so obsessed by me. I wonder why. Maybe it's just another one of his uncontrollable urges.

--

Rick
Reply to
rickman

I have an example board or two. They look like any other PCB, but the edges are radiused, and magnets stick to 'em. They're two-sided, holding a bunch of heavy-ish relays, and don't have any fine-line features. I never did figure out how the plated-through-hole problem was handled.

Lots of 'porcelain steel printed circuit' hits on Google.

Reply to
whit3rd

...

Without reading all the replies to see if anyone else suggests such sacrilege, I'd consider reasonable-sized to moderately big ones with chunks of copper soldered into them - i.e. good old fashioned through hole-type jumper wires. They don't need to jump anywhere (one jumper could fill two holes on the same trace.) I suppose the idea will upset surface-mount centric modernists and the related production-lines.

A more artisanal approach might involve tiny sections of copper wire just the thickness of the board, precisely deformed so they would stick in the hole with a bit of space for solder to wick. But that seems like a real PITA...

--
Cats, coffee, chocolate...vices to live by 
Please don't feed the trolls. Killfile and ignore them so they will go away.
Reply to
Ecnerwal

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.