4
\$\begingroup\$

I am working on a bit of a strange design: all components, SMD and Through-Hole, are mounted on the top side of the PCB, except for some really beefy MOSFETs, which are mounted to the bottom side of the board. The idea is to have a flat heatsink (like the ones meant for BGAs, with push pins) cover the whole bottom of the board, making as direct a contact to the MOSFETS as possible. The MOSFETs use IR's DirectFET packaging, which is about 0.5mm thick, so the heatsink will sit quite close to the surface of the board.

This leads to my problem: For obvious reasons I do not want leads from my through-hole components shorting to the metal surface of the heatsink, and I don't want to be drilling clearance holes in it. Is there anything preventing me from just cutting the leads flush before soldering?

I notice NASA's guidelines require at least 0.5mm of protrusion, and from what I've heard IPC requires leads to be "visible". I imagine this makes wave soldering more reliable, but assuming I am hand soldering the components and am careful to allow the solder to wick all the way through the PTH, are there any other disadvantages to cutting the leads flush (things like reduced mechanical strength, etc)?

EDIT: the thru-hole components are "can"-style caps, and terminal blocks, so they'll need to be soldered from the bottom.

(Probably unimportant) details: The MOSFETs should only put out 15-20 watts of heat combined, at absolute most, plus a watt or two from some large thermal pads. So I want to leave it an option for the user to remove the heatsink and mount the board directly to a metal chassis, with some sort of thin thermal pad to avoid shorting. So that's another reason why I don't want to drill clearance holes in the heatsink, for user convenience.

EDIT: board pics for clarity: top bottom

(If you're wondering about missing traces, it's a 4-layer board)

\$\endgroup\$
  • \$\begingroup\$ SMD terminal blocks seem hard to find, and through hole caps look to be cheaper (10uF polymer). As well, I want this to be as hand-solderable as possible (except for the directFETs) (weird! someone asked "why through hole", and then deleted the comment) \$\endgroup\$ – dn3s May 12 '15 at 18:17
2
\$\begingroup\$

If you don't want the leads to protrude on the bottom side you essentially want to mount your TH components as SMDs. So I would suggest to use SMD pads, push the leads onto the pads, and solder them there.

For large/heavy components this might give a mechanical reliability problem: a round wire soldered at both sides (or soldered at the bottom only but with a component immediately at the top) can't easily rip the copper from the PCB, but a lead soldered to an SMD pad can!

PCB material is not optimized for conducting heat, so you might want to put lots of via's in the appropriate places to conduct the heat.

I read in your edit that you want to do this for trough-hole screw connector blocks. My gut feeling is that this will seriously impact the mechanical reliability. You could consider using those two-level connectors for screw terminals, which would free room so you could put your heatsink under only a part of your PCB, leaving the area of the screw terminal terminals to be soldered at the bottom.

\$\endgroup\$
  • \$\begingroup\$ thanks for the advice. I ended up redoing the design to avoid this. \$\endgroup\$ – dn3s May 15 '15 at 20:20
2
\$\begingroup\$

Other than the difficulty in manufacturing I don't see any problem with this. From an electrical point of view, the lack of a conical section on the bottom side of the board won't have any appreciable effect on the connection resistance, and from a mechanical point of view, well there are so many other factors that determine if the lead/pad combination is strong enough that this is unlikely to be an issue. If in doubt you can do some tests on it anyway or add some adhesive to the component body.

I would suggest you oversize the holes though so that you can visually verify that the solder has wicked all the way through and right around the lead.

\$\endgroup\$
  • \$\begingroup\$ This is good to know. I ended up redoing my design to avoid my concerns, but I like the idea of flush-mounting a PCB to a (non-conductive) chassis, doing away with standoffs. \$\endgroup\$ – dn3s May 15 '15 at 20:20
1
\$\begingroup\$

Sounds a bit risky, component leads pushed into holes which you don't want to protrude through the other side of the board in case they short out with a metal heatsink on the otherside.

Don't forget you may have thermal expansion issues to cope with, may not be a problem, but certainly think about it.

I guess you're wanting to solder the through-hole components on the top side of the board, assuming there's copper traces there?

You could do that for through-hole resistors and some types of capacitors, but some capacitors are cans with both leads protruding from one circular face, these are typically mounted right down on to the board so it's going to be hard to solder these, unless the capacitor is raised above the board and leaving some lead showing, then solder, which would look a bit odd.

Then you can visually inspect the underside of the board to ensure the solder hasn't flowed entirely through the hole. I recall doing something like this years ago.

One the hole has been plugged by soldering on the top side of the board and the lead is too short such that its end is firmly within the hole and doesn't protrude, then you might make it all work.

But seriously, think about some heat conducting, electrically isolating silcone washers over the holes which are at risk where a short can occur between top and bottom side of the PCB, similar to that used on power transistors between the tranny and its heatsink, to provide electrical isolation so that you don't end up with a short. These washers are very thin, so shouldn't add to the thickness of the entire assembly, unlike using 5 mm long steel spacers and extra holes required for them.

\$\endgroup\$
  • \$\begingroup\$ I'm convinced. I ended up re-doing my design and moved the thru hole components outside the heatsink's footprint. \$\endgroup\$ – dn3s May 15 '15 at 20:17
1
\$\begingroup\$

There are few solutions for such problem,

  1. A super/ultra flush cutter can do wonders with TH leads.
  2. A thermal pad will provide a flexible, soft thermal interface at the cost of an increased thermal resistance.
  3. Thermally conductive epoxy can act as a filler, but as far as I know, it is also conductive.
\$\endgroup\$
1
\$\begingroup\$

Although I have not personally used one of these machines, some of the facilities that I have visited over the years have a rotary cutting system that trims the leads protruding from the bottom of the PCB to a very short length.

I saw one in use and it seemed pretty straight forward: there was a bed that the PCB mounts into (leads to be cut off pointing UP), then the protective shield was closed. The rotary saw looked something like a pneumatic die grinder mounted in a sliding X-Y table - this was moved manually over the surface of the PCB.

The whole process took very little time - tens of seconds between successive boards.

\$\endgroup\$
  • \$\begingroup\$ that is awesome! I am working on a really small scale and don't have the clout to talk directly to manufacturers, but if I ever make it big, I'll definitely keep a process like that in mind! \$\endgroup\$ – dn3s May 15 '15 at 20:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.