I haven't had a problem with high nickel steels with this because I tend to weld it more slowly and deliberatley, however I do run into this with regular carbon and low alloy steel...The problem could actually be that the welder isn't putting filler metal in as he's pulling out.... as you make your tie-in run over the top of the original starting point by about 1/2 inch...then as you're pulling out dab a little filler metal in and make sure that he pulls out straight toward the filler metal so that even if he makes an arc trail out, chances are he won't suck any of the weld back and create the fish-eye..just a suggestion...also if the welder holds his tig torch close to the workpiece for a couple seconds after breaking his arc, this supposedly helps cool the metal gradually as the torch has the argon hot(unless you have water cooled torches)

Ah, nothing like a little homogenous, single pass welding to bring out the little nuances of welder technique.

You’d be surprised at the number of young welders I’ve had ask me “what’s a fish-eye and how does it happen?” It appears many welding instructors never explain “why” to do something, just do it “this way”. So I’ll pass on what I’ve explained to them over the years.

During the normal root application, the phenomenon referred to as “fish-eye” is a natural condition caused by the effects of the arc current on the molten weld pool. The physical fish-eye indication is what’s observed after the weld as solidified. Possibly a better term could be used to help explain what it really is. Right or wrong I’ve always used the term “voltage swirl” in helping explain what you’ve experienced. When using proper arc current and travel speed to achieve complete penetration, directly under the arc on the molten weld pool is where you’ll observe this phenomenon. It’s not a fish-eye yet; it’s more of a vortex occurring in the molten weld pool. It extends vertically through the puddle in the shape of a miniature tornado (my personal description) or tube and if improper arc termination technique is used (as you’ve experienced), the miniature tornado (tube) is left intact through the solidified puddle with a shrinkage cavity on the inside surface as an unwanted bonus!

In order to prevent fish-eyes from occurring, the arc condition has to change from a penetrating arc to a non-penetrating one. It sounds pretty simple and it is. Instead of striving for penetration we need to gradually reverse the penetration process. If you just cut the current off or pop off the arc, you’ll have achieved the perfect fish-eye, so the opposite has to happen. Gradual, controlled chilling (solidification) of the puddle at termination is required to prevent fish-eyes (or leaving the tube). One way of accomplishing this, especially if your welder doesn’t have an amp foot control, is to overlap the tie-in point (after fusion has taken place) and rapidly increase his travel speed while gradually directing the shrinking puddle off to the side of the joint. A ½” of overlap is all that should be necessary to accomplish this. An old adage for a proper overlap was to overlap 3x the width of the puddle. In addition, an increase in torch angle immediately upon or just preceding tie-in will help facilitate this technique as well (care should be taken to not cause long-arcing on the wall of the pipe). By using this technique, the weld pool is gradually chilled, closing the fish-eye from the underside of the root on up to the surface ensuring a complete overlap and a smoothly blended inside surface. If the welder has a foot amp control he can do the same thing by gradually decreasing his current, which will accomplish the same end result.

Hopefully I’ve offer some usable advise for you and your welders and not created too much heartburn for respected others here with my somewhat, un-technical description.