Very interesting question! Why are the solder balls there in the first place and where did they come from? Are they from the wave soldering operation where the solder splashed and the solder stuck to an uncured solder mask? Are they there from a solder paste that did not reflow and was no longer solderable? Are they there because the pads were incorrectly designed and the solder balls are beneath the chip components?

These must be defined first so the proper method of prevention can be implemented into the process.

Now if you cannot define why they are there, then according to IPC-A-610, page 5-14, Acceptable Class 1, 2, 3 - solder balls are entrapped/encapsulated and do not violate minimum electrical clearance, with an added Note: Entrapped/encapsulated/attached is intended to mean that normal service environment of the product will not cause a solder ball to become dislodged.

Therefore if they violate any of these requirements they are a cause for rejection and are identified as a defective condition.

Now how can they be removed?

One must determine what is keeping them in place, if it is flux, then the flux must be removed and the solder balls will also be removed with the flushing action of the flux removal operation.

If no clean flux is used and there is no cleaning process, then we have an issue. If they are stuck to soldermask then we also have an issue.

The best way I found was to use a bristle brush and rub them off them try to flush them off with an air knife or air nozzle. Trying to use the air nozzle by itself may not be powerful enough to dislodge the entrapped solder balls.

Hence establish your process to minimize the formation of solder balls by proper process controls in both the solder paste and the reflow profile.

We have found that a low pressure (35 -40 psi) saponified cleaning with a 50 psi rinse removes more solderballs and flux than any water cleaning only.

Remember that solderballs attached to edges of pads or partially trapped below a component will not typically be removable by standard cleaning. These have a small mechanical connection to hold them in place and are difficult to remove.

It would depend on what the solder paste chemistry is. If it requires the use of a chemicals to dissolve it the cost might be prohibitive for an in-line type cleaning system.

Typically the amount water flow and pressure required to dissolve the paste is found in the higher flow higher pressure systems (40-60 GPM) (60-80 psi). The dishwasher type systems maybe worth a look as they have come a long way in the last few years.

Also should be concerned with close loop and chemicals. Impact on resins beds and discharge effluent

Spray-in-air is the most effective method in both the removal of post reflow flux residues and post reflow solder balls. Vapor degreasers generally lack positive pressure and only rely on solvent contact for defluxing. Ultrasonics may be effective but both real and perceived issues with ultrasonic energy and wire bonds will most likely eliminate this method.

For solder ball removal, consider machines will higher impact pressures. Impact pressures vary widely from brand to brand. Send samples to equipment manufacturers for evaluation.

Solder ball contamination can manifest itself during several steps of the PCB assembly process. If you are referring to post reflow, then only "fugitive" solder balls are of concern and a good spray-in-air system with good filtration may be adequate. However misprinted PCBs and pre reflow solder paste is a different story and a spray-in-air cleaner may exacerbate the problem.

An article was published in the March 2007 issue of U.S. Tech magazine that explains the pros and cons of varoius cleaning technologies titled Reducing the Cost of Misprinted PCBs. A PDF copy of the article is available by clicking on the link above.

Ultrasonic or Spray in Air technologies could both be used for solder ball removal from misprinted boards. The preference could depend on the following:

1) Equipment Design:
a. Spray in Air System: Wash pump, spray nozzle, spray nozzle population per bar, heated or non-heated wash tank.
b. Ultrasonic System: Transducer, power density, locations of transducers, heated or non-heated tank.

2) Test Substrates:
a. Geometry of the substrates
b. Double sided or singled sided PCBs
c. Fragility of the components (i.e glass diodes)

3) Cleaning chemistry:
a. Surfactant and non surfactant based cleaning technologies

4) Process:
a. Waiting time between printing and cleaning
b. Solder ball diameter
c. Viscosity of flux