What Ed Craig had to say on the subject:
copied from his site below
E-mail. Oct 2008: I am emailing you because I have come to a questionable snag with my pulsed MIG equipment. I have the equipment set in the spray mode. I am welding on 5/16” carbon steel material, my settings are set to spray transfer (29 volts 500 wire speed in/min).
When making a 3/16” fillet weld with the 0.035 (1mm) wire I have noticed that at the end of the weld, the weld flattens out and has what I have been taught to refer to as a “fish eye” ( I am not sure if this is the right term for this problem ).
The attached photo will show you what I am referring to. When coming to the end of my weld I back over the weld about ¼” instead of just stopping. I don’t pull my nozzle away before I let the trigger go, so I don’t think this issue is caused due to the length of the stick out. My gas is set to 35cfh argon/CO2 mix.
Could you please advise what may be causing this poor finish is this just cosmetic or an issue that needs to be addressed? If this is an issue that needs to be addressed could you please explain the proper procedure for fixing. These parts are under extreme vibrations and some stresses Vertical / Horizontal and Lateral. Thank you.T Eason.
Ed's Reply. Two things going on here.
[1] First the weld picture indicates poor side wall fusion. As you are using good spray parameters the lack of fusion is likely a result that the weld surface was wire brushed and the mill scale has been left. If you are concerned about fatique properties you don't MIG weld over mill scale. Grind the weld area before welding, I am sure you will see a difference in the weld appearence. As for the crater and crater hole.
[2] A fish eye is typically a pore evident in a failed weld and the bright shiny appearence in the pore indicates the presence of hydrogen, so you dont have a fish eye. You do have a pulsed power source that has a built in defect. This is a a commom classic issue with pulsed equipment in which the machine controlled end parameters or burn back parameters are set too high, (more evidence that pulsed equipment manufactures don't correctly test the equipment they build.) I see this defect all the time in pulsed equipment in robot cells. At the end of the weld, the high voltage spike applied for the burn back causes a suck back effect in the arc leaving that classic hole in the crater. In many instances if you examine with magnification you will find shrinkage cracks around that hole and with your fatigue concerns, this defect has to be ground out and the crater filled in. My MIG process control training resources deal with this issue and provide process solutions, however you would be well served to send the power source back to the company who manufactured it. It's ironic that this defect would not occur on a lower cost traditional CV power source.-end quote from Ed Craig