Allowable permanant PCB deflection.

[jacksonm]

HI

I am designing an enclosure to hold a circuit board supported on pillars from below and pressed onto them from above. My design must always deform the PCB in the same direction to avoid stressing some solder joints. The problem is when I do a tolerance analysis the maximum deflection could be quite high (1mm) over quite a short distance (27.5mm - defined by the support spacing). This gives a radius of curvature of less than 400mm which sounds small.

My question: is there any guidlines on a minimum radius of curvature for a SMT PCB that I can use as a starting point in my design.

When I ask electronics/PCB designers they always say 'as flat as possible' or 'test it'. The trouble is that with my current design 400mm is as flat as possible and my timescales don't allow me to start testing from scratch!

Matt Jackson
Product Design Engineer

 

[IRstuff]

Your tolerances sound huge for a machined part.

> Tighten them?
> Use washers to compensate individual assemblies
> Use elastomeric supports to eliminate hard contact



TTFN

 

[jacksonm]

The application is very cost sensitive and fairly high volume (15-20k PA) so there are no machined parts and I don't want to have to vary the build (to cope with tolerance bands) or band components.

The tolerances are cast base ±0.1 + moulded cover ±0.05 and also a PCB ±0.15 + connector ±0.2 (both of which I have no control over).

I am obviously trying to design out the tolerance problem but I would still need a PCB deflection figure.

Elastomeric supports are an option though - I'm already using some elsewhere. Thanks.

Matt Jackson
Product Design Engineer

 

[bvanhiel]

I can't address the permissable PCB bending tolerance, but what you have sounds high. Consider doing a tolerance analysis that considers the odds of a worst case scenario. If you assume that the given tolerances are +/- 3 sigma, you should be able to calculate what the combined tolerance is at that same liklihood of occurance.

Alternatively, you could add a spring, a piece of poron foam, to the stack to absorb the tolerance yet still deliver a load to hold the PCB in place.

-b

 

[Heckler]

I would consider potting the entire PCB. To much bending of the PCB and you will start seeing components seperating from their pads.

Best Regards,

Heckler
Sr. Mechanical Engineer
SW2005 SP 5.0 & Pro/E 2001
Dell Precision 370
P4 3.6 GHz, 1GB RAM
XP Pro SP2.0
NVIDIA Quadro FX 1400
o
_`\(,_
(_)/ (_)

Never argue with an idiot. They'll bring you down to their level and beat you with experience every time.

 

[IRstuff]

I think you need to separate the problem.

Use a bracket that you can control to set the deflection the way you want and then float the assembly in the enclosure.

TTFN

 

[ajamnia]

Do expect any vibrations? Steinberg has some suggestions on PCB deflections in a vibration environment. I wonder if that data can help you.

I have to admit that I like IRstuff's solution.

Ali

 

[jacksonm]

I have considered vibration as a separate issue. I also like the solution from IRstuff but there is a power device on the PCB which is screwed to the cast base for heatsinking and the cast base also provides PCB support so the PCB cannot float (unless I use foam or elastomeric pads).

I really appreciate all the suggestions on a workaround but I have considered most of these things - it's really the allowable PCB deflection dims I'm after. However this seems to be somewhat elusive.

Matt Jackson
Product Design Engineer

 

[Tunalover]

Steinberg's book probably states a maximum allowable board deflection to avoid popping off of SMT components.


Tunalover

 

[ajamnia]

You are right tunalover. His formula is based on the input G, the magnification factor and the natural frequency of the board.

Ali

 

[craigmj]

also depends on what kind of solder you are using. the new RoHS compliant solder is much more brittle than the old leaded stuff.

 

[jacksonm]

Thanks Tunalover

I don't currently own Steinbergs book (but I'll order it ASAP!) can anyone that does have a flick through to see if this dimension is available?

Matt Jackson
Product Design Engineer

 

[Tunalover]

My copy is at wqork. If I can fit it in today I'll reply with that number.


Tunalover