Your thoughts on Burn-in and Vibration Testing

[NW27]

Hi All,
I have some products that I'm trying to improve the reliability. The products range from assemblies with 1 PCB 40mm*40mm to assemblies with three 100mm*250mm boards stacked together.

I'm thinking of doing some burn-in and vibration of these PCA's and assemblies to remove the infant-mortality components and interconnect assemblies.

So my thoughts were:
Step 1. Burn-in stressing of the PCA's after receipting into the store.
Step 2. Vibration stressing of the PCA's.
Step 3. Makeup the PCA assemblies ie the three stack of boards. (Significant labor assembly content at this stage)
Step 4. Burn-in stressing of the Assemblies.
Step 5. Vibration stressing of the Assemblies.

All of the PCB have some kind of integrated circuits. Ranging from a single sensor interface to a 676 pin FPGA (BGA)'s.
Some PCA's can be powered up as individual PCA's, others required the assembly of the three stack PCA before power can be applied.

So what are your thoughts on:
1. Temperature to use?
2. Constant elevated temperature for 1 week (duration as an example only)?
3. Cycle temperature between ambient and Max Temp, say every 4hrs for a 24hr period ie 8 cycles?
4. What about vibration stressing.1 Axis or 2 or 3 axis? G Force level.

At the moment, I only have a large industrial oven. No environmental chamber (Big $$$$). So I can only heat.
I could get a freezer to put the PCA's into?

Recommendations on a vibrating table? for PCA's etc.

Thanks,
Neil.

[T3sl4co1l]

Interested to hear others' insights on this; we currently have a parts-per-thousand level failure rate on one product, possibly due to cracked capacitors (the products pass production testing, but somewhere between that, and packaging and shipment, failures occur.)

Tim

[ChristopherN]

This topic is very interesting, we have a similar thing with one of our more complex measurement tools (contains 5 PCBs, display, battery and so on).

They do test fine after production but some fail in shipping or the first weeks after delivery. I suspect stress on cables or one of the PCBs due to the tight fit in the housing. This does not affect the measurement results so it's just annoying and a bit expensive. We did improve things by using silicone on some of the cables but we still have a few failures.

I think that I'll go with vibration testing as a first step.

[schmitt trigger]

Everyone is suffering from reliability problems caused by the ever-denser boards, fine pitch & leadless components, and RoHS solder.

Vibration is great at discovering intermittent solder joints that very likely could fail in the future. 
Having said that..........The key to effective vibration stressing is that your circuit must be electrically monitored during the test. It could be as simple as a power-up and the monitoring of its current consumption, to elaborate software that  feeds all sorts of stimulus and reads back the responses.

[Twoflower]

Unfortunately I don't have details. But maybe some points you might consider.

I've worked close to a shaker for vibration tests with automotive parts. If I remember correct they stressed samples for lifetime within hours/days timespan. There I learned a bit: Not only Gs are important but also the frequencies. If they ran some frequency sweeps you won't be close to that because of the noise. Putting that one in a chamber was problematic as it's not easy to reduce the low frequencies. But that shaker was quiet powerful. It ripped the threads out to the mounting parts on a regular basis. So they were experienced to fix this.

You should also think about the masses you want to shake. That is an important point you need for the size shaker. The shaker I mentioned above was kind of drum-shaped, about 1m diameter by 1m height and was heavily bolted to the floor. It was used to stress devices below 1kg more like in the 100g area. But probably it was powerful enough to do more.

Also they tested the devices in several orientations.

[NW27]

Hi All,
Some good comments so far.

Virtually all (9 out of 10) of the products are fully encapsulated withing stainless steel enclosures.
The PCA's have a layer >5mm of encapsulate that has the consistency of jelly. This removes some of the physical stress's from the components but not all. It does however allow the components to expand and contract as necessary with the temperature variations ie some equipment is rated for an ambient of -20 DegC to +125 DegC. In general though, it is -20 DegC to +40 DegC.
These assemblies are then covered in an encapsulate similar to araldite.
So basically, there is no chance of repairing these boards, hence the push for greater reliability of the produce product.

So still looking for how other manufacturers are performing their Heat and Vibration testing.

Thanks,
Neil.