Which PCB-mounted SMPS brands are good?

Preload is going to be a problem with these supplies because the pins are tapered sharply. So they come to rest at a certain height above board. The only way to apply preload would be to flex the whole thing which is a bit scary.

But Gorilla glue does sound like good medicine. Probably better than the Chinese tar blob stuff.

I'll second Traco. We use then in portable test boxs all the time and they are bricks.

That is the stuff. After the boards are cleaned and tested, I use a small brush and "paint" water around the perimeter of the transformer. Actually it will cure with the moisture in the air but will take much longer to cure. With the tip of the glue bottle with the smallest hole that will allow the glue to flow I place a bead of glue around the edge of the transformer. The glue flows under the transformer and after 5 to 10 minutes begins to foam. In an hour or so it is done. I never tried for flexibility. The cured glue can be cut with an exacto knife.

If it doesn't crumble under an Exacto knife it might have some flexibility. How does it feel when you press the Exacto knife onto it? Like porcelain, wood, or RTV?

It's pretty tough stuff--the consistency is closest to green hardwood. It's definitely harder than the stuff they use to glue trunk panels on cars, but not an order of magnitude harder.

Cheers

Phil Hobbs

That might be just right. Too soft wouldn't work either because it needs to protect the comparatively tiny solder joints from becoming mushed. Of course the Traco is best with it's two 3.5mm screw connections in addition to soldering. But quite expensive.

I don't know glued trunk panels since mine (so far) were screwed or riveted. But that stuff is exposed to vibration so it must be able to hang on.

Everywhere I've worked 'brick' meant it was a dead piece of junk.

IBM service people used to carry a radio dispatch terminal, affectionately known as "the brick". It was just about the size and shape of one, too.

broke=20

failure.

So=20

much=20

wouldn't=20

=20

=20

Bigger thru holes so that the PSU sits closer to the board?

Then you have a non-standard mounting method, can result in liability issues. Iso can be maintained by voids on the main board but the weight of the supply module might cream those four li'l solder joints quite rapidly. They should just provide more of those pins and the problem would go away.

=20

broke=20

failure.

a=20

So=20

much=20

wouldn't=20

dummy=20

reduce=20

from=20

=20

=20

the=20

Non-standard mounting? How is that?

Because the supplies are designed with stepped pins, maintaining a defined distance to the board below. If you spec larger drill sizes to let the suppy slide in that's a non-spec mounting method.

This is not to say that it can't work. One can always make sure that the safety distances are still maintained. However, it does not solve the flexing and popping issues.

break=20

=20

broke=20

activates=20

failure.

retains a=20

with,=20

onto=20

=20

so. So=20

much=20

wouldn't=20

dummy=20

reduce=20

from=20

the=20

=20

pins=20

=20

thing=20

than the=20

=20

weight=20

=20

Sorry, i had the impression that the pins were tapered rather than stepped. Completely different situation that way.

These are also stocked by Farnell in the UK. Price (1 off) £24.50 or £30.20 for the version with screw terminal block inputs and outputs.

It's over our budget. Also, just like the MeanWell supplies the temperature derating curve starts at 50C or roughly 120F. That won't fly. Why is it that when you want the really good stuff you have to always roll your own?

None of the SMPS I ever designed begins to derate at 50C. Reminds me of my Chrysler where the alternator stopped falling out of the car only after I built my own bracket for it. Hurumph!