Tell us a bit more about your GMAWP power supply....
Make, Model, and year of manufacture.
Miller Axcess and Lincoln Powerwave has been notorious for burnback issues... Both Red and Blue have taken measures to address this but if your machine does not have the updates than you may be suffering from that.
Inco 625 has very low comparative thermal conductivity so I don't see a swelling issue taking place in the contact tube as the first place to look.
But first....Look at simple things like both ends of your work lead connection... This is by far the biggest cause of burnback in high production GMAW. Any powersupply with adaptive feedback controls (like synergic GMAWP) grows in sensitivity to poor work lead connections.
Also take a couple of hours and visit www.weldreality.com This has some great data on just what your doing... you will have to pick the gems from between rants... Eat the meat and spit out the bones.
In addition to what Lowrence has said, 625 work hardens very quickly, so residual cast in the spool can cause feeding issues. You may need a wire straightener very close to the feed rolls to help things out. A gent I work with has had instances where he required two straighteners, 90 degrees to each other, to eliminate the feeding issue.
Also, where is your voltage sensing lead clamped? You won't get anything good without that being clamped to the work piece, but separated from the ground clamp.
I have been involved with high nickel alloy pulsed GMAW and wire feeding was an issue. This was semi-Auto and the issues were often related to the welder using the wrong size tip (too big). The electrical characteristics would vary as the tip got bigger. The curve in the wire could elongate the tip hole rapidly. If the wire is stright coming out of the tip then thats no help.
Its my opinion that changing to a larger tip to correct a problem is not the answer. From what I understand pulsed GMAW needs pretty consistent electrical contact at the tip.
We tried to assure the drive roll pressure was enough to drive the wire (without overtightening) and that the correct drive rolls were installed on all machines (You have probably done that if you have a robot). Good grounding was also a problem and could cause inconsistent welding charateristics. The machines we had were ESAB Synergenic power sources and Wire Feeders and seemed very picky about things like that.
Hope it turns out for you .
Given the fact that it consistently happens at about the same time and not at once leads me to believe that pipewelder is on the right track in that it is somehow thermally related.
Try vee knurled drive rollers with just enough pressure to make very light impression on the wire.
Are you using 30# spools of wire or feeding drum a bulk drum which has "cast less" wire?
As far as liners go, I tried about every different type of nylon ect. and found that the best liners were heavy duty steel liners.
Get the correct ID but wound from the heaviest gauge wire that will fit in your gun. (The OD will be bigger than a standard liner)
This will help the gun lead make a bigger diameter loop and stop kinks.
If you do play with nylon or such, skip the serated drive rolls, your liner will be cut in no time. (been there) The serated ones work very well with a steel liner.
Get higher quality contact tips, not bigger ones.
What brand of welding gun?
Is this a water cooled mig gun or air cooled ?