What is acceptable ways of soldering wires directly to PCB?

[eV1Te]

I am designing a PCB which will have wires going to a front panel. It will only be produced in small numbers. The device is designed to last 10+ years, hence I want to decrease risk of wire failing at the solder joint (fatigue, lifting pads, etc.)

What is the best way of soldering wires directly to a board except using turrets? Is there some industry guideline for this?

In IPC-A-610 (standard for final electronic assemblies, IPC-A-600 is for circuit board manufacturing) I can not find any description of how one shall solder a wire directly to a PCB, it only covers jumper-wires or wires soldered to turrets (see attached image for turret) as far as I can see. 

Perhaps I am missing something?

[schmitt trigger]

There are eyelets also which are pressed into a PCB hole.

Having said this, if your wires are smaller than #25 AWG and/or will not see any pulling stresses, you could even get away with soldering the wires directly to the board.

[schmitt trigger]

This is something that you could also use:

https://www.digikey.com/product-detail/en/te-connectivity-amp-connectors/172781-2/A111048TR-ND/2328795

[Yansi]

If you do not want the wire to not break at the soldering point, then do not solder! It is that easy.

Use crimped connections! Either connector, or special wire terminations designed to be soldered to PCB.

//EDIT: Yep, that thing that Schmitt trigga posted.

[IanB]

In many items I have disassembled the wires have been looped once or twice through holes in the PCB for strain relief and then soldered in the normal way to a through hole pad. I have never seen a turret or special fixture used on a commercial board (other than detachable connectors).

[voltsandjolts]

I use the loop method a lot and it works fine.
It's not good for volume manufacturing though; takes too much time feeding wires through the holes.

[Arjunan M R]

If you do not want the wire to not break at the soldering point, then do not solder! It is that easy

//EDIT: Yep, that thing that Schmitt trigga posted.

[Arjunan M R]

I use the loop method a lot and it works fine.
It's not good for volume manufacturing though; takes too much time feeding wires through the holes.

Why don't make a loop maker machine
Or looper

[jmelson]

If there is minimal flexing, then soldering wires into plated-through holes in boards is quite reliable.  If the wire or cable will be flexed, then some kind of support that prevents the end where it is soldered from flexing will be OK.  The picture of the turret terminal is from NASA sort of practices, where life-critical systems need to survive insane levels of vibration, thermal stress and other abuse.

Jon

[Gyro]

Agreed, the way to stop wires falling off is to avoid flex anywhere near the soldering point. Loom them and use a cable tie through 1 or 2 holes in the PCB to mechanically restrain them against movement.

BTW, the wire shown in the OP photo appear to be silver plated, PTFE insulated. This is often used in aerospace but is the worst for solder wicking and fracture near the ends. It's an absolute requirement that these are physically restrained away from the solder connection.

[JPlocher]

You might also ask yourself if soldering the wires is the best solution, or if there are ways to control or eliminate wire movement:

• In low quantities, could the cost of a connector buy you peace of mind?  Maybe a crimp on ribbon cable with a connector on each end plugged into header strips and locked with a squirt of hot glue?
• Could you mount the PCB to the front panel in such a way as to eliminate any flexing?
• if there is repeating flexure between the board and front panel, in addition to supporting the wires at their termination points, you will want to investigate good wire management (wire guides, looms and tying, vibration damping...)

[Psi]

The device is designed to last 10+ years

Some issues to consider for 10year lifespan.


- Environment: You will need to make sure its at least somewhat sealed from the environment and moisture. This maybe done for you depending on where the product will be used.  Inside vs outside etc..
You probably also want conformal coating on your PCB.
Temp cycling also needs to be considered.
Check if all your parts will be ok running over the temp range they will be used in. As well as their lifespan in said temp ranges.

- Vibration: This is what you were saying about wires snapping.
If your device is never going to get vibrations then you don't have to worry about soldered wires snapping as long as they are generally well retained and not flopping around.
If your device does get vibrations then you have to consider how bad they will be. For automotive products inside the cars cabin isn't that bad. Vibrations aren't very strong and are low freq so as long as wires are restrained well it's fine.
If your device is going on some industrial equipment that's function makes it vibrates 24/7 then you really do need to choose parts and interconnects very carefully.

- Part selection
Some parts just wont last 10 years, like electrolytic caps or big switches passing lots of current.
So select parts that you know will last the duration. That might be ceramic caps or polymer electrolytic and maybe a SSR instead of a relay or power switch.  Note that ceramic caps are not so good for vibration, they crack, so there is some crossover between categories. Something good for one maybe bad for the other.

[David Hess]

In many items I have disassembled the wires have been looped once or twice through holes in the PCB for strain relief and then soldered in the normal way to a through hole pad.

If there is minimal flexing, then soldering wires into plated-through holes in boards is quite reliable.  If the wire or cable will be flexed, then some kind of support that prevents the end where it is soldered from flexing will be OK.

Exactly, soldering the wire through a plated through hole and then adding strain relief works fine.

The common alternative is to have the wire go into a connector housing which than attaches to a header on the board.  This is more convenient if the board ever needs to be disconnected.

[Smallsmt]

I like this type of SMT cable connectors:
unfortunally I forgot the manufacturer maybe WAGO has

[TERRA Operative]

I have often seen connectors that look like JST type connectors but instead of the plug and socket being seperate parts, it is only one part that is soldered down.

So pins get crimped onto the wires and pushed into a plastic housing like normal, but then that is soldered directly to the PCB.

Makes for fast assembly and a very reliable permanent connection.

[eV1Te]

I like this type of SMT cable connectors:
unfortunally I forgot the manufacturer maybe WAGO has

Thanks for the tip, it seems that Molex is making several sizes of this type of "wire to board" connector where the wire is removable with the push of a button. They even make vertical variants for less board space and perhaps easier insertion inside an enclosure (but the wire is not removable it seems).
https://www.molex.com/molex/products/family?key=litetrap&channel=products&chanName=family&pageTitle=Introduction

[Siwastaja]

Direct soldering is OK. Very often used. The risk is often exaggerated.

They fail if:
1) the factory worker doesn't know how to solder - can be seen in absolute worst offenders. This happens with absolutely no flux in the solder, or total failure to provide a contact between the iron tip and both the PCB and the wire at the same time - a cold joint, easily detected when inspected by eye.

2) they failed to provide any restraint

3) the environment is specifically tricky with a lot of strong vibration, combined to a partial edge case of 1) or 2).

Often a bit of glue solves the restraint problem. Any type which can form a large blob (so that the glue connects the PCB to the wire about 1cm away from the solder joint; it doesn't help if the glue is only over the joint) is OK. Looping through PCB holes is OK; or using a cable tie which is looped through PCB holes, or put around the PCB, kept in place with slots.

Extruded plastic fingers in the case work as well. Anything that prevents the wires from moving.

[mzzj]

Agree with Siwastaja.
I have seen lot more problems with various connectors than wires soldered directly to pcb so I wouldn't worry about it too much unless its some aerospace nuclear power plant vibrator. 

[bd139]

I've got stuff that is 50 years old with wires soldered to the board. It's fine.

The trick is to use decent quality wire. The stiff stuff like they use inside HP power supplies is good, not the cheap floppy PVC crap from China.

[T3sl4co1l]

1. Direct wire in board.  It's fine.  Strain relief preferred, especially if you're using single layer NPTH boards (helloooo cold solder joints) or solid wire (less flexible = more prone to fatigue failure).

Really, strain relief is paramount, regardless.  Crimped connectors can be better on strain (the strands are able to slide past each other right beside the joint) but you still need to keep things stable otherwise they'll shake apart eventually.

This is more of a problem where flex and vibration are common.  Finer stranded wire and more flexible insulation help a lot, and there are ways to handle repetitive flex (like the chain belts or spirals used to bundle cables to moving platforms).

2. Wire-to-board clips.  These crimp on, and have a retention latch that clips into the PCB, holding it secure for wave soldering.  Solder still wicks up the wire, though.  I'm guessing, with the right wave cycle, it doesn't wick past the crimp part that hugs the insulation, so the strain relief is good, but beware if you're using them with hand soldering, you may end up wicking too far.

Worth noting that consistent and well optimized processes go hand in hand with high reliability.  It's hard to do in the first place, but even harder to do in low quantity.  If it's really critical and not just a nuisance if there's downtime, you may consider contracting out assembly to a house that has this experience.  (And by "really", I mean you'll be looking at 5 digits cost, maybe 6.)

3. Wire-to-board pins.  An example is linked above.  These, I think, solder in without the wire getting tinned at all.  The pins may be a weak point though.

4. All the various kinds of connectors.  These include IDC headers, crimped headers, flex cables, etc.  These also include wire-to-board headers, like ribbon IDC to pin headers which solder in.

You'll probably avoid IDC for the most part, speaking of; for ribbon cables I guess they're not terrible but I'm not sure how much so on the 10-year scale (and, if we're talking automotive environment, probably not as great?).

5. If you don't have enough budget to do super-hi-rel construction, consider redundant methods.  For connectors, sure, you could double up pins or cables for example.  But you might also just make two or three full units, and have them ready to swap out with minimal downtime.  Possibly with a live changeover if downtime needs to be truly minimal.

Tim

[wangxinghe88]

I've seen connectors like this. It's like a wire holder that you need to insert wires into the holes, and then glued somehow. The wires are arranged and can be easily soldered on PCB. Do you know what it is called and where to buy it?

[openyk]

I had a similar design challenge, motivated by my frustration with unreliable mechatronics everywhere. My answer is PC pins with crimped or solder-cup wire-termination.

For example, see Mill-Max crimp terminal pins, metric catalog (for 12-28 AWG wire): https://www.mill-max.com/catalog/download/2019-07%3A223.6M.pdf

Solder or crimp the wire-to-pin connection, then insert+solder the pin-to-PCB connection.

With the crimp-terminal PC pins, the wire is not soldered hence not as brittle.

With the solder-cup PC pins, orient the high-side of the solder-cup wall (in direction of nominal wire bend) and keep the solder joint low enough to get some strain-relief for the soldered-wire section.

When I get more field testing data with this design I hope to return and share good news.

My related point of uncertainty is what insulation-design I should be using with this wire-to-board method. Heat shrink is the intuitive idea but open to any ideas on "completing" this design space.

EDIT: Mill-Max PC pins are only RoHS-compliant with exemptions. In this case they have up to 4% lead content as a copper alloying agent. As I highly prioritize designing with safe materials, this fact compels me to search for a more non-toxic alternative. Molex Board-In connectors with halogen-free material (2.5mm pitch) only go as far as 3A per circuit, too low for high-power but quite good for reliable control-side electronics.

[Feynman]

I'm pretty sure there is a standard that says something about soldering wires directly onto a PCB. I would have guessed it's in IPC-A-610, but since you found nothing in there, maybe it is in some related standard.

[SerieZ]

I have had fairly good success with FastOn Terminals.

[jfiresto]

If you have the room, have you considered spring-cage terminal strips? I used strips like these inside an enclosure to bring outside cables to a PCB, in case I have to replace the board or a cable.