I have about a dozen plus robot cells and about a dozen and half robot arms.  I'm in a high volume/high mix shop.  On the outside it may seem that it's a high volume/low mix but each part is different by a few inches.  Our robot cells have are designed to grab the part at the same place on each different part number or close to it.  They also have anywhere from a dozen to 30 or more different robot programs for each robot cell.  We try to minimize all of the variables but in the “welding world”, it can be next to impossible and keep your overhead costs low or manageable.

So, I have a couple of questions:

1)  For welding robots, GMAW, hardwire (0.035”, 0.045”, 0.052”), mild steel, what would be the expect defect rate when the robot cycle is complete?  I understand this is a very vague and loaded question but really, on an average, medium sized part (10-20 piece weldment), about 30-50 welds, 85% fillet welds, 0.3125” and 0.25” fillets, single pass, seam tracking, laser search, touch sense, robot less than 3 years old…what would be expected as far as a defect rate.  Defect rate is defined as no obvious defects such as porosity, overlap, undercut, underfilled craters, welds not centered in joints, anything from D1.1 quality requirements.  My company is just trying to get a feel for where we are at and I have asked a few robot mfgs and a couple of similar companies but I am getting the feeling that I am not getting very honest answers.
2)  At what point do you stop trying to make minor adjustments to achieve zero defects from a robot, how much do you spend, labor, material, overhead?  One of my robots, the weldment is only about 4 to 5 pieces being welded together.  One of these joints is very critical and my defect rate is 0.1% or about 35 for every 35,000 year.  When do I chalk it up to the process and admitted everything that I can do has been tried and at this point I am spending more money than it’s worth?