Issues with metal contacts onto PCBs

[mikeselectricstuff]

A customer of mine has proposed a mechanical arrangement whereby 15-odd PCBs are stacked on brass rods. Picture a set of window blinds.
 The rods will carry power & data.
The proposal is for brass collars to friction lock onto surface pads on the PCBs, the PCBs will be mounted on aluminium strips with a slightly larger outline to the PCB. there are also some ntlon bushes involved but not sure of they are under any pressure, 
The proposal is to tighten up the whole stack with nuts a each end.

I am worried about the long-term contact integrity (brass to gold plated PCB), as I'm aware of issues like PCB cold-flowing over time, as well as possible thermal cycling causing changes in contact pressure. I've suggested they put a spring in somewhere to maintin pressure

I also have concerns that unplated brass against PCB pads may have long-term relibility issues.
Power will be low voltage, about 8V, feeding a max of 1A to each PCB. I'm not sure 8V would be enough to get the oxide punch-though that high current contatcts rely on.
Oxide on the data line is potentially more of an issue.

I don't know enough about the metallurgy to say how likely it is that it will have problems over time, but just have a nagging feeling that it may do.

Any experience/suggestions? Would nickel-plating the brass help?
 
 

[Precipice]

Is this the one time when star washers are actually the answer?
Turn your contact areas into points, and add a bit of spring to the stack?

[T3sl4co1l]

Hmm, you're probably right to worry about brass on gold.  Nickel plate will solve that (diffusion barrier), but nickel is probably worse than brass for contacts.

Brass is partly passivated by its zinc content (as stainless is passivated by its chrome content), and so is nickel.  But nickel's oxide layer is somewhat tough and forms rectifying contacts, so it can be undesirable.  I guess the zinc oxide layer is soft enough not to matter under most other applications.

I suppose sending the brass through ENIG is inconvenient or expensive...

I don't like star or split washers on PCB (they chew up the plating), but I'd be happy with cone/wave washers, one per PCB hole I suppose.  Tighten the whole thing down and it should sit okay.

I suppose there might be some advantage to using thinner rod, so it's stretchier.  The difference between clamping a whole stack versus pinching each board between a pair of nuts is, the sprung length of thread is only a board thickness in the latter case, but includes the board-to-board distance in the former case.  So you have some mechanical advantage, having more length of rod versus PCB stack, even though the PCB is that much more flexible.

Tim

[mikeselectricstuff]

I don't like star or split washers on PCB (they chew up the plating), but I'd be happy with cone/wave washers, one per PCB hole I suppose.  Tighten the whole thing down and it should sit okay.

Beryllium copper wave washers perhaps?

[Richard Crowley]

What is the motivation to do this?
This scheme has serious metalurgy and mechanical issues.
Either of those by themselves would be a show-stopper IMHO.
And TOGETHER, they seem like a really terrible idea.
There are at least half a dozen ways of doing this properly.
The proposal seems like a horrible kludge.

[David Hess]

I agree with T3sl4co1l about the star washers; they tend to chew up the printed circuit board although I have seen them used successfully this way.

I have seen brass standoffs used this way in old designs and occasionally new ones for ground and power.  The thing I would worry about is the temperature coefficient of expansions resulting in compression of the printed circuit board fiberglass.  The old designs had brass rivets through the printed circuit board which would protect against crushing but compression washers should work also.

[mikeselectricstuff]

What is the motivation to do this?

Art

This scheme has serious metallurgy and mechanical issues.
Either of those by themselves would be a show-stopper IMHO.
And TOGETHER, they seem like a really terrible idea.
There are at least half a dozen ways of doing this properly.
The proposal seems like a horrible kludge.

Mechanically it's generally fine for the application, problem is it's being designed by a mechanical guy who assumes that as long as two bits of metal are sort-of touching then it will be OK
I have serious reservations about the electrical implications over time, I just can't tell them for certain that it won't work reliably, or offer detailed reasons.

[Richard Crowley]

I would at least use bell washers between the standoffs and the boards to concentrate the contact pressure. I would think that would increase the long-term reliability significantly. A coil spring wouldn't provide nearly as much pressure.  Or, at very least, precision grind the ends of the spacers to make them flat and perpendicular.



Can you solder the spacers to the boards after mechanical assembly?

[AndyC_772]

How long does this product need to last? Is "long term" reliability a problem, or does it need to survive an exhibition and that's it?

[Richard Crowley]

Indeed, as AndyC_772 observes, what is the expected life-cycle of this item?

I would also consider using some kind of spring-contact soldered to the board adjacent to the spacer to exert continuous sideways spring-pressure.

[mikeselectricstuff]

How long does this product need to last? Is "long term" reliability a problem, or does it need to survive an exhibition and that's it?

Permanent install, hung from a ceiling so access will be a problem.

[mikeselectricstuff]

Indeed, as AndyC_772 observes, what is the expected life-cycle of this item?

I would also consider using some kind of spring-contact soldered to the board adjacent to the spacer to exert continuous sideways spring-pressure.

That's exactly the sort of thing I suggested quite early on, but mechanical people have gone off on their own & come back for the PCB design, thinking it's just a case of providing flat pads on the PCB for their fixings to land on.

[Precipice]

Can you decouple the mechanicals from the electricals? Pick power off the rods by some method other than the PCB landing pads? Maybe clamps or tapped holes in the rods?

Or, if you're mostly worried about the data comms, then cheat and use RF, or IR between the boards, or something?
I'd not normally propose either of those as 'easier than copper', or more reliable, but...

And yeah, Belville washers do sound like a better plan than star washers. Will I ever find a use for them? They must exist for some purpose.
 

[mikeselectricstuff]

Can you decouple the mechanicals from the electricals? Pick power off the rods by some method other than the PCB landing pads? Maybe clamps or tapped holes in the rods?

Or, if you're mostly worried about the data comms, then cheat and use RF, or IR between the boards, or something?
I'd not normally propose either of those as 'easier than copper', or more reliable, but...

been through most of that already - need power anyway, so little point not using hardwired data. Th problem has always been the desire to combine mechanical & electrical connection, but they want an "invisible" join. Support rods are 2mm dia brass (though this may change to copper+kevlar cable) so no scope for holes!
 

[Someone]

Are we missing the obvious of actually soldering these rods onto the pads? Or is there too much thermal cycling?

[nctnico]

A customer of mine has proposed a mechanical arrangement whereby 15-odd PCBs are stacked on brass rods. Picture a set of window blinds.

Any experience/suggestions? Would nickel-plating the brass help?

I had a similar problem where I had to bolt a busbar to a backplane. I used nickel plated brass standoffs and I used another nickel- plated washer + star washer to retain pressure. The reasoning to use nickel plated brass standoffs is because the gold plating on a PCB is copper nickel gold. Large capacitors are often bolted to a PCB so cold flowing shouldn't be an issue.
In your situation I'd insist on having the stack being pressed together. Maybe even use threaded pieces (round standoffs) to build up the stack.

[Richard Crowley]

If the support rods are only 2mm OD, then they amount to just a large "component lead".
Can the boards be "threaded" on to a continuous rod and soldered into place?

Or drill and tap these things and solder them onto the boards.
Then use a set-screw to bite into the rods (or flex cables?)

[Dave Turner]

Mike - if I understood correctly you mentioned suspended and inaccessible. At what distance from an observer? Without knowing full details I'd agree with precipice and Richard C have the way to go.

Are you after the electronic version of this?

http://www.amusingplanet.com/2012/08/magical-floating-faucet-fountains.html

[lovro]

Saw this a few days ago in Electronica
http://www.mac8japan.com/

Maybe use some of their solderable pcb threads / spacers?

[David Hess]

Can you solder the spacers to the boards after mechanical assembly?

Are we missing the obvious of actually soldering these rods onto the pads? Or is there too much thermal cycling?

I have seen this done and it did not help at all.  Standard solder does not work well for mechanical joints.  With time either the pads got ripped off of the board if the solder held or the solder joint cracked.  Similar problems can occur with designs which use brass rivets for through holes.

[mikeselectricstuff]

Are we missing the obvious of actually soldering these rods onto the pads? Or is there too much thermal cycling?

been there, done that. Never again ( for a few reasons)

[sparx]

Could you mill a hole into the edge of the pcb, and use something like a brass pin with a hole through it perpendicular to its body, to support the pcb on the rail? Therefore stopping the issue of pcb plating on rails?

My description was pretty crap:

Also, conductive lube?

[kizzap]

Can you use something like a spring nut, where you have a nut threaded onto the brass rod that is threaded through, and said nut has a spring welded/soldered/permanently joined to it, which you can tighten down onto the PCB. If you get really keen, you can solder the spring to the PCB and having a nut behind the PCB will lock the PCB in place, while allowing it to move if it needs to.

[free_electron]

PEM nuts may be the way to go.
these are 'pulled' into a plated hole and 'bite' into the plating.

the nut itself is threaded. you can get these with a counter nut. KF2 or KFH series
http://www.pemnet.com/fastening_products/circuit_board.html


another solution is a soldered nut that is guaranteed to get good contact.

we use PEM nuts ll the time for the gournd contact on burn in machines. these things go through thermal cycleing thousands of times without losing electrical contact.
but ! you need to design the correct hole in the PCB and you need to employ thick plating ( 2 ounce copper so you have 1 ounce in the hole) you also cannot rely on the hole containing the nut to contact the inner layers. you need additional via's around the hole to make that connection ( standard construction of a screw hole with 6 via's in its annular ring )

wurth has those solderable nuts

http://katalog.we-online.de/en/em/search/Stromversorgungselemente

another idea would be to solder on a round collar to the pad ( like the wurth ones ) and use a turret spring inside the bushing. the turret spring goes over the rod . hmm kinda difficult to explain. envision those turned-pin ic sockets. that insert is called a turret spring.

Multicontact makes those in various shapes and sizes.

http://www.multi-contact.com/products/productline/7 look at that sa400. that is what i mean. they exist in different sizes. all you need is to pick one that clamps over the brass rod , and inside the bushing soldered on the board. these things are used inside connectors that are to be semi-permanently mated.

[T3sl4co1l]

PEM and other brands' nuts are available as soldered (shoulder, barrel or both), blind or through thread, etc.

In a recent project, we put some on a board to hold a daughterboard; seems to work, though it hasn't been through thermal yet.

They're not quite as available as general hardware from McMaster is, so as you build a design, be careful what you're working yourself into, if all the parts are quite available or not.

Tim