Are there guidelines or specifications for using silicone to secure large parts against vibration? 2

[sp8472]

On a lot of electronic assemblies, I see silicone, or even hot glue sometimes, used to secure large components against motion or vibration. I know that solder is weak structurally.

Some assemblies have silicone on darn near everything that has more than several millimeters of height. I've seen computer power supply teardown photos that look like that: Whole sections are buried in silicone. Others only put it on very large components.


-- Is there an IPC or EIA spec, or some kind of standard that says when it is necessary or advisable to use something to secure a component in addition to its solder joints?
Right now, the silicone processing step is seen as nothing more than a waste of labor. Unless I can provide some good evidence and proof of it from some external source, that step will likely be eliminated. I'm all in favor of reducing labor and making an efficient process, but not if it means creating failed parts.

Most of my experience on the mechanical side of things with regards to fatigue loading was with steel (good old endurance limit) and aluminum (tends not to show a flattening of that particular curve). I know that tin/lead solder is fairly weak in general, and I wouldn't want to entrust it with very much stress.

 

[IRstuff]

Vibration will induce flexing of the leads and solder joint. So, what you need to consider is some sort of structural analysis given the vibration environment to determine which components, if any, are prone to flop around in a vibration environment, and restrain them.

TTFN
I can do absolutely anything. I'm an expert!
homework forum: //

http://www.engineering.com/AskForum/aff/32.aspx

faq731-376 forum1529

 

[MacGyverS2000]

I would think a few hours (days?) on a vibration table would provide a wealth of feedback useful for/against the argument.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[MikeHalloran]

A random electronic assembly, mounted on a shaker table, and randomly excited, without silicone or other auxiliary attachment of bulky components, will probably fail within a few minutes, sometimes less than one minute. Copper doesn't have real good fatigue properties.

I'm guessing that the silicone was added to your subject assembly at some time in the distant past, in response to vibration testing or analysis of failed assemblies, and whoever did it may not have recorded the complete reasons for the change, or you did not find the historical record, or did not look.

Anyone who suggests removing silicone adhesive from an existing assembly is an idiot, or worse, an MBA.


Mike Halloran
Pembroke Pines, FL, USA

 

[sp8472]

I do have some kind of vibratory feeder hardware here that could be modified easily to hold a circuitboard.
Even so, some manner of IPC/ASME/EIA spec would go a long way. Even if I show a failure on a shaker table, because it's a small in-house test and not an exact replica of the vibration seen in the field (which will vary greatly depending on installation), it may not be seen as valid evidence.


The silicone is a recent addition to our processes. That means that it's new, and new things can meet resistance. Old designs used heavy transformers that were bolted to the PCB. The new design is lighter, but has some parts like bulky capacitors. I think that using silicone is quicker than attaching nut+bolt+insulating washer, and is simple and cheap way of preventing failures due to a known mechanism.


- I got some free CRT TVs over the years that had been trashed because they didn't work. Reason: Cracked solder joints on anything from a 1x0.5x0.25" capacitor, up to the bulky flyback transformer. Reheat joint = working TV.

- Solder is not strong to begin with, and I can only imagine what tin/lead would do under fatigue loading, or copper foil, especially if it's got a capacitor flopping around on the other end.

- Vibration results in behaviors that you wouldn't normally expect an object to do under static loading conditions. ("What do you mean, I need to 'properly' torque these screws?")

 

[MacGyverS2000]

Old designs used heavy transformers that were bolted to the PCB. The new design is lighter, but has some parts like bulky capacitors. I think that using silicone is quicker than attaching nut+bolt+insulating washer, and is simple and cheap way of preventing failures due to a known mechanism.

The best proof would be to place several copies of the old and new design through the same torture tests... first units to fail win the argument for one side or the other.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[IRstuff]

"Even if I show a failure on a shaker table, because it's a small in-house test and not an exact replica of the vibration seen in the field (which will vary greatly depending on installation), it may not be seen as valid evidence."

I think you're a bit confused about this. In any halfway plausible design specification, the vibration environment is specified, and is already assumed to be representative of the envelope of actual environment. Therefore, when you shake the unit under test to the specified environment and it fails, it is, by definition, noncompliant to its design specification. If there is argument about whether the test is "real," it's irrelevant, because that's what it's specified to. If the specification is too stringent, then you should lobby your customer to change the requirement.

TTFN
I can do absolutely anything. I'm an expert!
homework forum: //

http://www.engineering.com/AskForum/aff/32.aspx

faq731-376 forum1529

 

[VEBill]

The military spec is often MIL-STD-810.

The test procedures are precisely specified, but there are plenty of options representing different environments.

Consumer gear would have other, perhaps less strenuous, requirements.

It's all about "Requirements".

 

[sp8472]

I think you're a bit confused about this. In any halfway plausible design specification, the vibration environment is specified, and is already assumed to be representative of the envelope of actual environment. Therefore, when you shake the unit under test to the specified environment and it fails, it is, by definition, noncompliant to its design specification. If there is argument about whether the test is "real," it's irrelevant, because that's what it's specified to. If the specification is too stringent, then you should lobby your customer to change the requirement.

The vibration environment hasn't been specified, and we are essentially the customer for our own manufacturing line. If there is to be an environmental specification, we'll be the ones writing it for our own products. The market I'm in isn't heavily regulated. I don't want to take that as a sign to start pushing the limits of functionality. I prefer to make robust things that work well, and continue to do so for a long time. If labor or production steps can be removed without compromising the product, yes, gladly. I have no interest in sacrificing quality for the company's benefit and at the customer's expense.




VE1BLL: Thank you for that direction. I'll have a look at that standard. We've referenced a military spec before for our solder joints, and used it as a baseline for testing resilience against thermal cycling, which is of course one more way of potentially causing fatigue failure.

 

[IRstuff]

OK, so you should have a pretty solid idea of where and how your board is being used. Which means you can determine an enveloping vibration spec. If there's no spec, how did your engineers design the board?

TTFN
I can do absolutely anything. I'm an expert!
homework forum: //

http://www.engineering.com/AskForum/aff/32.aspx

faq731-376 forum1529

 

[Comcokid]

The amount and location of RTV needed for vibration and what productions ends up actually applying can sometimes be two different things.

Excessive amounts of silicon can be added for non-technical reasons also. I recall once taking a tour of an electronics assembly facility of one of the big US auto manufacturers. Knowing how cost sensitive they are I was rather surprised at one high volume but electronically simple module that was potted with an excessive amount of a pricey silicon compound, and pointed it out. The two engineers giving the tour suddenly had a dear-in-the-headlights-stare, and one quickly took me aside to explain. A corporate executive had a close friend who sold the compound to the corporation, and there were 'orders' in place to use as much of it as possible. One past engineer tasked with cost reduction pointed out the cost savings that could be had using simple acrylic coating of the circuit board and was immediately fired from on high. The engineer asked me not to mention the module and to quietly continue my tour.

 

[VEBill]

Apologies in advance for drifting off topic, but...

"If there's no spec, how did [the] engineers design the board?"

I believe that the formalities of Requirements Traceability, even just writing down a spec, is a process that arose to common use far more recently than, for example, the field of circuit design. It's intended to be a value-added risk reduction process, but it indisputably adds time and labour to the schedule. It's certainly possible to completely ignore the entire process (go straight to design with only a mental understanding of requirements), provided that everyone understands and accepts the increased risk.

It's even a reasonable approach to skip it for small simple rushed projects. It may be obvious that the process would not be worth the price.

The process may be mandated of course. Even then, it can be 'tailored out' by explicit agreement to meet schedule targets.

Even invoking the entire formal Requirements Traceability process (e.g. DOORS end-to-end) certainly doesn't eliminate risks. It just allows the design engineers to point back at the requirements authors as being at fault. I've seen product specs where the revision letter had gone past 'Z' and was into the double-letters (e.g. Rev AH); each revision representing a 'new' previously-unforeseen ('Oops') requirement. Spec authors are human too, so an efficient change process is valuable.

The answer to the question is obvious...

Nike's trademark: 'Just do it.'

 

[IRstuff]

The question was somewhat rhetorical. Nevertheless, even if it's by word of mouth or by tribal knowledge, there needs to be something. Not having even a semblance of a requirements process mean that questions like this wind up AFTER the fact. Usually, the rationalization is "we have no time to waste" on no stinkin' requirements process. If you don't have the time to do it right the first time, how do you have time to work your way through issues like this after the product is design?


TTFN
I can do absolutely anything. I'm an expert!
homework forum: //

http://www.engineering.com/AskForum/aff/32.aspx

faq731-376 forum1529

 

[VEBill]

My paragraph 6 ("Even...") was responsive to that question.

 

[zappedagain]

"revision letter had gone past 'Z' and was into the double-letters" - Ah, continuous improvement!

Z