SMT vs thru-hole connector mechanical strength

I just about always use connectors with through-hole connections to the shell. It's cheap insurance against field failures, especially when they're user-accessible.

Cheers

Phil Hobbs

Through hole in my book is always better but you should have solder on both sides with via's in-case the header does not sit totally flat on the top side of the board.

Jamie

Well, FR-4 says peel strength 11 lbs/in. So, however much stress you expect on a connector of so-and-so width..?

We're talking peel strength, so if you can guarantee the pulling force will never be more than so-and-so (and evenly distributed, not twisting/flexing), you'll be pretty good. And probably as good as anything through hole (you contend with solder mushing out of pins, and laminate strength around smaller pins, versus laminate strength around big pads).

Connectors on the edge of the board, with a lot of leverage, will peel off in no time. Weren't the BNCs in the early TDS1000s notoriously s***ty SMTs?

I would consider SMT connects sufficient for internal connectors (which have to withstand shock&vibe, assembly and infrequent service, but not users), and not for external.

Tim

I understood, I just think it's generally foolish. YMMV.

Cheers

Phil Hobbs

Ideally, no human-caused stress is carried to the printed circuit board (and no foreseeable mechanical stress on wiring is carried by any electrical solder connection).

So, on a protoboard, I'd fasten a cable with two wire-ties (one for pull, the second for torque), then tack-solder the cable wires. A complex of controls and displays can be connected with flexible printed wiring. Switches and knobs and some coax connectors get fixed to the chassis in various ways. There's solder/rivet/screw provisions for D connectors, builtin plastic snaps on modular sockets, etc.

The rule is, stress is transmitted to the chassis, or mechanically to the board, not the wiring solder joints. Heavy wire, goes to a rivet instead of to the foil.

Seed the topside solder pads with vias to prevent ripping them off the board.

We use an SMT-only micro-B USB connector, no problems.

Dome connectors don't *come* with shells or protrusions

"... expect on the CONNECTIONS of a connector"! I.e., if the only anchored parts of the connector are its connections to the PCB, then those are the only elements that will handle the forces.

That's the problem. You can't expect users to mate/unmate connectors the way they *should*. E.g., folks who unplug power cords by tugging on the cord... not even "in-line" with the plug (i.e., off to one side)

Yeah, I'm leary of anything on an edge. Or, anything that effectively presents a long "lever".

You're screwed then. ;)

Cheers

Phil Hobbs

The operative word there is "ideally". I can arrange for any "abuse" at assembly to be minimized. But, once in the hands of the user, I

And, the typical user attitude towards that: it's a faulty PRODUCT, not ABUSE!

Unfortunately, I have none of the above, available.

I will probably have to do some empirical studies: mount different varieties and see how they fare in "simulated real world" environments.

I've encountered LOTS of "dangling" micro-B's. Too much leverage available to the user (the connector fails *normal* to the insertion force, not inline -- the connectors don't "grip" each other tight enough for there to be any real insertion/withdrawal force "inline")

Years ago we did that with a connector, pads only. It was a disaster. The only temp fix we could do at the time, and were lucky with trace routing, was to drill 2 holes in the pcb and tie-wrap the connector down. I think it was a RJ45 but might have been a audio or video connector.

If it's completely internal you can get by with it but if it is something user will be accessing, in my book dont do it.

I can ensure there are no stresses from cable flex, etc. (e.g., anchor the cable to the board so the board bears those stresses). But, eventually, I'll have to deal with the user unmating the connection.

E.g., an RJ45 has very little (electrical) "contact area" with a board. I wouldn't hesitate to mount a "binding post" directly to a board (aside from the high profile it presents -- lever arm)

(sigh) I'm going to have to look harder at my options and run some tests to see how bad/notbad the problem really will be. I just don't think of "finesse" when I think of users -- regardless of the target audience! :-/

"Dome" s.b., "Some"

Of course, point being -- a generalization isn't very useful, but if you're looking at connectors, you can get a rough idea of how much force or shear or torque it can handle for a typical footprint.

You could adjust for stress raisers by guessing how much the area the force is distributed over is reduced by. Or adjust for vias (and other hacks) by adding the tensile strength (hopefully) to the figure.

Further, if you have limits on where the connector can move (maybe it's not bonded to the enclosure, but captive in a hole?), you can guess the force for a certain deflection, assuming stiffness of the PCB stock, and mounting, and all that. And you can modify that at the PCB level by cutting slots or moving mounting points around, assuming you have the layout area and other freedoms to do something like that.

Tim

I don't know that shells are so needed, but my experience has been that SMT connectors are not very reliable with repeated insertions. I can point to any number of connectors that have failed because the SMT solder joint gave up the ghost.

I expect it would take a lot to damage a through hole connector even if the soldered pins are not attached to the body or a frame directly. I think the pins take most of the abuse anyway. We had a hard drive 3.5 to Compact Flash adapter and the durn connector fell off the board when all 44 solder joints failed. The body was just a small piece of plastic that aligned all the pins for assembly really.

I've also had USB connectors in a laptop fail when their connections popped off the board, some removing some of the PCB trace with it.

Really well-made through hole connectors soldered into well-made plated through hole boards will often fatigue the leads off before the solder connection fails. Connectors soldered by surface mount without some kind of rigid fastening for the connector body are much more fragile. I've seen LOTS of these develop cracked solder joints, and in some cases the connector falls entirely off the board, occasionally ripping off a few traces.

Now, this is for things that are externally accessible, and routinely having the plug inserted and removed. Internal connectors that are only inserted when the device is assembled shouldn't cause much trouble.

Jon