cool TI paper

Interesting read. Dependability of devices is a big part of my day job of repairing arcade games...

I run into problems with long term reliability of SMD LED driver chips where the solder junctions fail. In a particular application this is likely the result of long term exposure to heat/cool cycles and weather where the logic board is exposed to twisting tension where it is screwed to an aluminum tower that is fifteen meters or so tall. This is a set of

I want to make new panels to replace the old ones as the 8 x 8 LED legs are rotting off on most panels and they are going to finally fail within another five years. So trying to figure out what will be most likely to survive for perhaps another twenty years of outdoor exposure...

John :-#)#

Semi-Permanent - it sits outdoors from April through early November, and then is stored under cover at outdoor temperatures. This is Vancouver, BC, so coolest shade temp is perhaps -10C and hottest is 40C (rare, closer to max of 35C in the few really hot [what we call hot!] days in summer), it does rain though...

That is an excellent point, I can certainly make thicker boards. That will greatly reduce the flex...

No, the tower is secured halfway up and near the top to a steel "H" frame but it can still flex a smidgen in the wind, and when it is taken down for storage... No guy wires securing it.

Photos here:

The LEDs are in 8 columns with each panel having 8 X A6275 Allegro 8-Bit Serial Input Constant-Current Latched LED Drivers. Obsolete but replaceable easily enough. The original design has 8 LEDs per 'pixel' in a 4 x 4 array. My plan is to replace the 8 LEDs with a single high intensity RED modular LED so it is easily seen in daylight.

Potting might not survive rough handling...

Oh yes, the real world is a scary place for electronics and other creatures!

Thanks!

John :-#)#

Depends on potting choice. Potting in a box of some sort is generally the single most robust solution, for both mechanical and weather.

I'd look into re-enterable potting agents, which are typically some kind of gel.

Joe

The parts killer is board vibration. On a serious shake table, a pcb will blur when a resonant peak is crossed and parts fly off.

A thicker board at best will have different resonances.

Some damping would help. And more spacers.

I try not to rely solely upon solder for mechanical connections of large through hole parts because of the greater stress they experience due to the larger mass. In your application, even small parts are under stress.

Should you be gluing these parts down before you solder so the solder joints are not going to be under stress as the only mechanical hold downs?

Failures occur at any time. I think that taking the assembly down may stress the circuit board connections as the tower may twist, leading to twisting the PCBs slightly and contributing to the IC solder failure on one of more pins of the 8 ICs per PCB.

The hammer makes zero connection to the tower display - it is mounted on pavement a couple of metres away.

If the amusement park agrees to my cost estimates to replace the displays I intend to keep the same modular idea for the 8 x 8 matrix displays.

Yes. Updating the display circuit boards.

I'm stuck with SMD devices, but the current generation is about 1/2 the size of the ones used previously and this should help reduce solder stress fractures.

Thanks!

John :-#)#

I don't know if gluing them down will help. I think reducing the size of the SMD devices will reduce the PCB solder stress fracture risk though.

John :-#)#

You could increase the area of the surface mount pads so that more solder paste is available, giving a thicker solder fillet. Is it possible to add stiffener bars to the pcbs to reduce flexing? John

Larger pads is a good point. Thanks! There isn't room to add bars, but I can make the board thicker which may also help. It all depends if the organization wants to spend money to update the machine or just replace it with something else...

John :-#)#

The trick is match the mechanical stiffness of the glue layer to that of the soldered connections. Thermal layers are usually not very stiff.

That also helps.

The proof is shake test where a sinewave drive is slowly swept across the likely resonance range. If the chip flys off, deem it failed.

Joe