Welding with a robot – the 5 most common mistakes

Posted by Manfred Seipp on Jan 29, 2021 12:40:23 PM

The welding robot has been programmed for hours. The straight paths work perfectly, when welding a curve everything is wrong back and forth – what happened? The industrial robot is synchronized, the component is fastened and the weld seam is still set incorrectly – where is the mistake? On the first day, tens of components were perfectly welded, the next day the seam was displaced – how could that happen?

Welding with a robot is just as efficient as it is difficult – regardless of the welding process used in the industry. Here, essential things have to be observed from the very beginning so that your company does not lose valuable time, because ultimately this has a painful effect on the quality, the output volume and of course on the costs. Welding with a robot requires a lot of knowledge and experience. However, just as essential, your own interest is also part of it, in order to find solutions to challenges and problems and to fully exploit the advantages of a welding robot in welding technology.


In the following we have set out the 5 most common mistakes when welding with a robot – and of course, tips on how to prevent or correct these.

1. Tool Center Point (TCP) is not set up correctly

TCP? It’s hard to imagine that this is possible but there are actually robot users in the industry who do not deal with the TCP when programming their welding robot. Instead, they teach to the max – and then they are unpleasantly surprised when the straight tracks are welded well, but nothing fits when doing curves or circles. Setting the tool center point is part of the robot synchronization and is one of the basic requirements if all seams are to be welded properly.

The TCP is set on a tip or a point. From there, at least the three axes X, Y and Z are calibrated from a level. When changing the torch position in these axes, the TCP must always remain the same. The TCP becomes even more precise if eight or more points are used for the calibration – different numbers of calibration points are used depending on the robot type. The robot welding torch will always maintain this setting, even if the work surface tilts to the side like a tilting turntable. From this point, the robot's control unit calculates the TCP to the robot, to the table or to the tilting turntable in the background and can then weld circles without any problems. Provided, of course, that the workpiece itself does not have excessive component tolerances. See point 3 for information on how you can counteract constant workpiece tolerances.


2. No collision protection on the robot

It can happen again and again that the welding robot collides somewhere while moving fast and the torch neck is affected, meaning it can be deformed. If the front end of the welding torch is then, for example, 10 mm to the side, all welding seams are also offset by 10 mm if the damage is not noticed immediately. For this reason, every robot should have collision protection, i. e. a shutdown protection in the event of a crash. The robot then comes to a standstill immediately. The welding torch is not damaged due to the deflection of the collision protection, and the torch moves back to the original TCP thanks to its reset accuracy. In this way, the entire welding equipment is protected.

Without such a collision protection, which is located between the robot arm and the welding torch, you will otherwise have to repeatedly straighten a deformed torch and calibrate it to the robot, which means to also reset the TCP. The latter is certainly no fun, because there may be 10, 50 or 100 programs set to this one tool center point. In the worst case, this means using a new torch and also having to set it up again. Repeat accuracy is therefore a very big issue when welding with a robot. The advantages of an automatic shutdown in the event of a collision pay off immediately.

3. Clamping device or workpiece are defective

For each component and each process, the clamping device must be so precise that the components to be welded are always positioned exactly as the robot was programmed. This can lead to faulty welds if the clamping device is incorrect or if a component has tolerances. A synchronized robot can only weld properly when the clamping device works, the components to be welded do not have excessive component tolerances and the entire environment – i. e. the TCP with all its measurements – fits. If there are deviations at this point, welding defects occur.

An optical seam tracking sensor that detects the course of the seam to be welded, corrects these tolerances or component changes directly during the welding process. This is a great advantage because you save costs directly: Fewer defective parts means a faster return on investment than you might originally have thought. Such a purchase is definitely worthwhile if you want to produce as little scrap as possible.

There is also the option of tracking the seam with an arc sensor. The benefit is that the paths in the arc can be corrected by processing voltage (or current) differences. The disadvantage of this is the pendulum motion with this technique.

4. Torch neck is changed without paying attention

At some point even the most robust welding torch has to be replaced. With every new torch neck in the process, there is again the possibility that it does not fit 100 percent. Before a new torch head is used, it needs to be checked.

A torch exchange system with several torch necks in a rotating cassette is very useful for robot welding. The advantage: All torch necks are aligned on a neck alignment before being inserted into the exchange system to ensure that each one fits exactly. If, for example, a contact tip is worn out, the robot automatically takes the next torch neck from the torch neck changing system and can continue welding directly. The used torch necks can be removed, cleaned, equipped with new wear parts and reinserted in the exchange system. Such a system shortens the downtime of a robot system to a minimum and always ensures the appropriate supply of suitable torch necks. You can find examples here.

However, not everyone has such a torch exchange system. If you want to manually replace a torch neck in your welding cell, you need to make absolutely sure that the new torch neck has the same angle and length. Only then the set TCP can be guaranteed.

5. No trained specialist for robot welding

Anyone who is professionally involved in welding with a robot in production must definitely be a trained specialist. That may sound like a simple basic requirement, but unfortunately reality can change the picture far too often. Only those who like to deal with robotics and do their job with full passion will be able to think outside the box. Therefore, it is not enough to send a welder to a robotics course and only assign him to program a robot many months later if necessary. Then there is a great risk that basic things like setting the TCP and synchronizing all the essential components will no longer be remembered.

It is therefore advisable to actively use the advanced training programs and courses offered by local robot manufacturers or the German welding society DVS – German Association for Welding and Allied Processes – or any other welding organisation and to make the time available to the employees concerned. The know-how gained can be actively put into practice right away and it definitely pays off in time savings when dealing with the welding robot.


Sometimes you have to learn painfully through mistakes in production. Anyone who has programmed a welding robot for the first time can certainly tell a thing or two about it. We hope that this blog was able to give you helpful tips so that one thing or the other is easier from the start when making your products. If you have specific questions about welding tasks with a robot, please do not hesitate to ask us.

Happy welding!

Cobot welding - Download free eBook now!

Topics: Robotic Welding