Green solder mask and heatsinking

Loss of radiation will be miniscule. I always cover my gound planes/ heat blocks with solder mask.

Heat *radiation* will be greatly improved, because the organic solder mask is highly absorbing in the thermal IR (say 5-15 microns wavelength) and therefore is an efficient emitter. (The second law of thermodynamics requires that the absorptivity and emissivity of any surface be the same at all wavelengths, because otherwise you could have heat spontaneously flowing from cold to hot, which can't happen.) Shiny metal is a very poor infrared radiator.

Heat transfer by convection, i.e. conduction to the air followed by mass motion of the heated air, won't be changed much, just as pcw1.cad says. (Puke one cad? What kind of name is that?) The reason is that natural convection is so lousy that a mil or two of plastic won't hurt it much.

Cheers,

Phil Hobbs

Thanks for the info and for the quick reply.

Phil ... I wasn't aware that solder mask was an organic. I always thought it was some sort of enamel without any pigment. I'm not doubting you, just curious. Do you have a reference that I could read up on solder mask material? Or even better, a url pointer?

Shiny

Second the motion on that.

Do you have a SEWAG, or even a rectorandom estimate of how much of the heat is radiated and how much is convected on a device like a TO-220 that is, say, 70°C above ambient? Thermo was NOT my strong undergraduate suit.

Jim

Well, it goes on with solvent, softens just above soldering temperature, turns black when it chars, and otherwise behaves just like paint, so I assume it's organic. Some sort of polyurethane would be my guess. I don't know anything else about its chemistry.

You can compute the radiation using the Stefan-Boltzmann law, which says that the total thermal radiation per unit area from a perfectly black body is

dQdot/dA(T) = sigma * T**4,

where sigma is Stefan's constant, 5.67×10**-8 W/m**2/K**4. Of course it also receives energy from its environment, which is given by the same formula.

So a TO-220, which has a total surface area of 4 cm**2 or thereabouts, at 370 K in a 300K environment would have a net thermal radiation of

Qdot = (4*10**-4 m**2)(5.67×10**-8 W/m**2/K**4)(370**4 - 300**4) = 0.24 W.

That's an optimistic figure, because only the plastic part will have an emissivity near 1 (emissivity e = (actual radiation)/(Stefan-Boltzmann result) The convection is whatever's left.n

Cheers,

Phil Hobbs

So using my observed rule of thumb that in room temperature environments or thereabouts that the case temperature of a TO-220 (and most small cases) goes up SOMEWHERE around 20C/watt, then to go up 70C would mean that the device was pumping out about 3½ watts. If radiation gets rid of a quarter of a watt, then the convection gets to take care of the other 3¼ watts. DId I get that right?

Jim

Something like that, if you gave the tab a coat of paint. Otherwise it'd be more like 3-3/8 watts.

Cheers,

Phil Hobbs

Whoops, forgot about the lead conduction. That probably dominates.

Cheers,

Phil Hobbs

Thanks for the confirmation. In that case, I wonder why many manufacturers leave solder mask off ground planes that double as heatsinks even when there's neither a clutter of solder points nor a need to reinforce conductor cross-section with solder.

(The HS is for two 7812 regulators that together will dissipate about 10W max. The copper plane will be thermally coupled to the ICs via their ground leads and to a blackened extruded Al heatsink via mounting bolts. I considered using a switched-mode PSU but decided against it).

If you're solder-plating the copper anyway, there's no reason not to open the solder mask--the cost is no different, and the heat transfer will be very slightly better.

I'd want to calculate the expected thermal drop before I used lateral conduction in a thin foil to connect with a heat sink. You might get an unpleasant surprise.

Cheers,

Phil Hobbs

The regulators are mounted on the Al heatsink. The ground plane is also thermally coupled to the heatsink by the bolts simply because 1) it is mechanically convenient and 2) it supplements the IC leads as a heat conduction path to the copper.

Large areas of unetched copper is considered bad practice, especially with Mil-Spec, because of the differing thermal expansion between the copper laminate and the (typically fiberglass/epoxy) substrate. When large areas are needed because of shielding needs or a ground plane for a stripline, the etching pattern normally used is a grid or parallel traces which allow for thermal expansion without the copper lifting from the board during reflow. Unfortunately this means the loss of surface area for heat convection.

Dr. Barry L. Ornitz WA4VZQ

Interesting. Large pours are "normal" in some neighborhoods (I've never done it) and certainly solid internal planes are the norm. How do they get away with the differential thermal expansion problem?

--
Keith

CONFORMAL COATING WORKS FINE

It depends on what you consider "large" is! I am generally talking about areas around 35 x 35 mm, or about 2 square inches and up. It also depends on just how reliable you want the board to be and whether rework or repair is ever necessary. I suspect the modern designers are used to very fine lines and more modern copper bonding technologies allowing them to ignore differential thermal expansion. I do remember that most of the PCB CAD programs I once used had the option of "cross hatching" large pours.

Microscopically, the copper foil contacting the substrate today is covered with little "mushrooms" of copper grown electrolytically on the foil. When used with a partially cured substrate (pre-preg), the high pressure curing process locks the foil to the substrate quite tenaciously such that you rarely see the copper lift from the substrate like you did 50 years ago.

I still have some old printed circuit boards in the junque box where every through-hole is a small pad surrounded with a ring with only two small traces to connect the center pad and the ring. These are quite old and I suspect they were hand soldered.

Dr. Barry L. Ornitz

So what about ground planes?

Cheers,

Phil Hobbs

Get gold plating for your entire board. Many PCB fabs don't even charge you extra for that. Green solder mask is passe, there are many other colours that are much better looking. Gloss black is nice. A good cheap place to get your boards made is

Dave.

Un bel giorno pimpom digitò:

For the typical PCB temperatures, heat radiation gives a very little contribution anyway, it's almost all about transmission and convection. For the heatsinks in free convection, there is a difference extimation in thermal resistance less than 10% between the best radiation case (heatsink painted in black) and the worst case (bare metal). In forced convection, the difference is practically zero.

Probably if you use a dark solder mask you will get a very small, almost unnoticeable improvement.

--
emboliaschizoide.splinder.com

In the case of planes, the size of a panel is fairly normal.

That's tiny to be a hard restriction.

How is "cross hatching" any different? I'd think you still be pulling at the corners of the "hatch". In any case, planes aren't "hatched". I *have* seen boards that are somewhat balanced on top/bottom (equal number of planes, etc) to reduce warping.

50 years? No, I didn't see much of that 50 years ago either. ;-) A *lot* of things have changed in 50 years.

Wasn't that to prevent delamination?

--
Keith