New Member Seam finding question

LochinvarDan

I am a new memeber to your forum.  We have had a welding robot in our operation for the last 6 months and have been constantly battling problems.  Our parts are cylindrical tanks with multiple fittings that need to be welded.  We are using a three jaw head and tail stock to grip the tank.  Our problem is that we can not hold tight tolerances(+-.075") on our tanks and they end up being positioned in the chuck diffrently for each part.  It is enough that we have to go back and hand repair roughly 30% of the parts.  We have been looking for an add on seam finding system and think we have found one but would still like other options.

 

Has anyone else out there used a seam tracking system like the one I am looking for?  If so what was it and how well did it work for you.  We are trying to weigh the options of adding more tooling to try and locate the tanks better or would it be in our best intrest to use this seam finding system.

 

Thanks.

graemepaulin

Hi,

 

Have you looked at using SmartTac (an option from ABB) for searching for the location of the fittings and tank surface?

 

This will slow down the production rate as performing the search is not a fast process.

 

Regards

Graeme

LochinvarDan

We started out using the SmartTac but did not have any success with it.  Since these fitting are round and sit on a round tank it has a funky looking joint. With them being round if it doesnt sense directly on the tangent then that will throw the calculation off.  The other obstacle with that is they are roughly 3" tall and if I touch off the top of the fitting that is no garuntee that the bottom of the fitting is directly below that.  LochinvarDan2010-08-12 18:00:10

graemepaulin

A possible solution would be to use the search to give you three points round the fitting then calculate the center of the circle (there is a utility routine available from ABB which has this function which will save some programing time). Use the difference betwen the centers (as programed and actual) to give you the offset required.

 

If you can not search close enough to the weld to negate the fitting not being perpendicular you would need calculate the center at 2  different hieghts on the fitting and do some maths to get the correct offset.

 

You would need to do the same sort of calculations for the tank if its postion also varies.

LochinvarDan

LochinvarDan

Here is an example of what the tank looks like and what joints look like.  We are capturing the coupling closest to us with an arm so that all of our programming is based off its position.  The three fittings off to the left only have one weld pass on them and the ones in a line have two passes, one for the backing ring and a second for the nipple.  You can see if we use the Smart Tac then the only place to really touch is threaded section and that is a taper. We dont have a brake on the wire drive or a wire cutting station so the wire length varies after each weld.   Also we aren't able to get it to weld good using only arc moves we also have a couple very short straits built in to accomodate for following the contour of the fittings and the tank.  I am going to an ABB training course at the end of September but I am trying to figure out what are the best questions to ask when I am there so I can get the most of the training. 

LochinvarDan2010-08-13 16:45:55

mattdavis

It is possible without too much modification (depeding on your type of wirefeeder) to touch sense with the gas cup rather than the wire which looks like it would work reasonably well in your situation, I would still recommend a torch cleaning station to keep the gas cup free of spatter. 

I think you could search each side of the tube in 2 places to determine the location and angle of it and then move down just outside those locations to determine height at both sides.

The backing ring looks like it will be very hard to locate, but as long as it is always the same distance from the tube it should be workable.

As a method of doing the touch sensing that avoids having to do alot of trigonometry you can just program one part and then touch sense in the desired locations afterwards and just take the results it gives you as correct. Then when you next touch sense you just need to correct for the differences in results. It might work out simpler in your case.