SMT vs thru-hole connector mechanical strength

[Don Y]

Hi,

"Guidance" (rules-of-thumb) for relative mechanical strength
of connectors fastened to PCB *solely* by their electrical
contacts (no metal shells, swaged protrusions, etc.) in terms
of stresses from the user trying to mate/unmate the connector
in SMT vs thru-hole forms?

Are connectors in which insertion is normal to the PCB worse
(in either technology) than those where the forces are *along*
the plane of the PCB?

Any rule of thumb re: (total) foil areas vs. forces (without
risking lifting foils or compromising mechanical strength?

Thx,
--don

[Phil Hobbs]

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

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics

160 North State Road #203
Briarcliff Manor NY 10510

hobbs at electrooptical dot net
http://electrooptical.net

[Maynard A. Philbrook Jr.]

In article <m075tn$kc0$1...@speranza.aioe.org>, th...@is.not.me.com says...

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

[Don Y]

On 9/27/2014 1:22 PM, Phil Hobbs wrote:
> On 9/27/2014 4:12 PM, Don Y wrote:
>> Hi,
>>
>> "Guidance" (rules-of-thumb) for relative mechanical strength
>> of connectors fastened to PCB *solely* by their electrical
>> contacts (no metal shells, swaged protrusions, etc.) in terms
>> of stresses from the user trying to mate/unmate the connector
>> in SMT vs thru-hole forms?
>>
>> Are connectors in which insertion is normal to the PCB worse
>> (in either technology) than those where the forces are *along*
>> the plane of the PCB?
>>
>> Any rule of thumb re: (total) foil areas vs. forces (without
>> risking lifting foils or compromising mechanical strength?
>>
>> Thx,
>> --don
>
> 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.

*No* shell(s):

"fastened to PCB *solely* by their electrical contacts (no metal shells,
swaged protrusions, etc.)"

I.e., relying solely on the strength of the solder connections and foil
adhesion to the board.

[Tim Williams]

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 shitty
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

--
Seven Transistor Labs
Electrical Engineering Consultation
Website: http://seventransistorlabs.com

"Don Y" <th...@is.not.me.com> wrote in message
news:m075tn$kc0$1...@speranza.aioe.org...

[Phil Hobbs]

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

[whit3rd]

On Saturday, September 27, 2014 1:12:26 PM UTC-7, Don Y wrote:

> "Guidance" (rules-of-thumb) for relative mechanical strength
> of connectors fastened to PCB *solely* by their electrical
> contacts

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.

[John Larkin]

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.


--

John Larkin Highland Technology, Inc

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

[Don Y]

On 9/27/2014 2:29 PM, Phil Hobbs wrote:
> On 9/27/2014 4:49 PM, Don Y wrote:
>> On 9/27/2014 1:22 PM, Phil Hobbs wrote:

>>> 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.
>>
>> *No* shell(s):
>> "fastened to PCB *solely* by their electrical contacts (no metal
>> shells,
>> swaged protrusions, etc.)"
>> I.e., relying solely on the strength of the solder connections and foil
>> adhesion to the board.
>>
> I understood, I just think it's generally foolish. YMMV.

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

[Phil Hobbs]

You're screwed then. ;)

[Don Y]

On 9/27/2014 2:20 PM, Tim Williams wrote:
> Well, FR-4 says peel strength 11 lbs/in. So, however much stress you
> expect on a connector of so-and-so width..?

"... 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.

> 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).

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)

> 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 shitty
> SMTs?

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

[Joe Chisolm]

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.

--
Chisolm
Republic of Texas

[Don Y]

On 9/27/2014 3:07 PM, whit3rd wrote:
> On Saturday, September 27, 2014 1:12:26 PM UTC-7, Don Y wrote:
>
>> "Guidance" (rules-of-thumb) for relative mechanical strength
>> of connectors fastened to PCB *solely* by their electrical
>> contacts
>
> 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).

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
*expect* abuse.

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

> 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.

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.

[Don Y]

On 9/27/2014 3:39 PM, John Larkin wrote:
> On Sat, 27 Sep 2014 13:12:26 -0700, Don Y <th...@is.not.me.com> wrote:
>
>> "Guidance" (rules-of-thumb) for relative mechanical strength
>> of connectors fastened to PCB *solely* by their electrical
>> contacts (no metal shells, swaged protrusions, etc.) in terms
>> of stresses from the user trying to mate/unmate the connector
>> in SMT vs thru-hole forms?
>>
>> Are connectors in which insertion is normal to the PCB worse
>> (in either technology) than those where the forces are *along*
>> the plane of the PCB?

> 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.

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")

[Don Y]

Hi Joe,

On 9/27/2014 4:16 PM, Joe Chisolm wrote:
> On Sat, 27 Sep 2014 13:12:26 -0700, Don Y wrote:
>
>> "Guidance" (rules-of-thumb) for relative mechanical strength of
>> connectors fastened to PCB *solely* by their electrical contacts (no
>> metal shells, swaged protrusions, etc.) in terms of stresses from the
>> user trying to mate/unmate the connector in SMT vs thru-hole forms?
>>
>> Are connectors in which insertion is normal to the PCB worse (in either
>> technology) than those where the forces are *along* the plane of the
>> PCB?
>>
>> Any rule of thumb re: (total) foil areas vs. forces (without risking
>> lifting foils or compromising mechanical strength?
>

> 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.

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)

> 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.

(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! :-/

[Don Y]

<grrrr> "Dome" s.b., "Some"

[Tim Williams]

"Don Y" <th...@is.not.me.com> wrote in message

news:m07gdu$d25$1...@speranza.aioe.org...

> On 9/27/2014 2:20 PM, Tim Williams wrote:
>> Well, FR-4 says peel strength 11 lbs/in. So, however much stress you
>> expect on a connector of so-and-so width..?
>
> "... 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.

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.

[rickman]

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.

--

Rick

[Jon Elson]

Don Y wrote:

> Hi,
>
> "Guidance" (rules-of-thumb) for relative mechanical strength
> of connectors fastened to PCB *solely* by their electrical
> contacts (no metal shells, swaged protrusions, etc.) in terms
> of stresses from the user trying to mate/unmate the connector
> in SMT vs thru-hole forms?
>
> Are connectors in which insertion is normal to the PCB worse
> (in either technology) than those where the forces are *along*
> the plane of the PCB?

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