Detecting small cracks in stamped metal parts

[PLCDontUQuitOnMe]

I am looking at an application where I will need to detect small hairline cracks in stamped metal parts. The sensing will need to be done in the stamping die in the press. The cracks will be of a hairline nature and could be anywhere from 5mm to 30mm in length.

Originally I was looking at a camera system but I have been told that eddy current sensors can be used for this. I have hesitations about putting a camera into a stamping die as well.

Anyone have any experience with this type of application using an eddy current sensor and can provide any kind of insight?

 

[I_Automation]

I have used eddy current sensors for checking tapped holes for good threads.


I found them to be very reliable and repeatable in detection's.



I wouldn't hesitate to use them for an application like this.

 

[joseph_e2]

I've seen eddy current sensors used successfully to detect delaminations between layers of a bimetal strip as it went by. I would hesitate, however, to try to put them inside a stamping die. I'd reach out to some folks that make them, though, to get their insights.

 

[parky]

There are two main methods of crack detection Induction (eddy current) & thermographic.
Induction has been used for many years, I remember in the 90's I was involved in a machine that stamped a boss into a gear wheel, although I had very little to do with the crack detection it was just a matter of getting a signal from the detector i.e. good or bad. I seem to remember seeing the gear placed on some sort of platter & being spun but I'm not sure if this had to do with the crack detection, it was spun I believe when inserting the boss (have no idea if the boss was stamped i.e. push fit & then spun to fuse it in or if it was to do with the crack detection). Induction only works well if the part is reasonably smooth & consistent, Never done thermographic but apparently this is an ever emerging process & may be better.

 

[PLCDontUQuitOnMe]

Thanks for the input. I will investigate the eddy current sensors and thermography a bit more.

 

[LadderLogic]

I have seen a machine using an inspection like this: https://www.modalshop.com/ndt/products/ram-drop

The idea was to drop a small metal part against some surface and to detect the sound pattern with microphones. The sound patterns of a part with cracks differ from those without.

The plant people called it "ding inspection"

 

[PLCDontUQuitOnMe]

I have seen a machine using an inspection like this: https://www.modalshop.com/ndt/products/ram-drop

The idea was to drop a small metal part against some surface and to detect the sound pattern with microphones. The sound patterns of a part with cracks differ from those without.

The plant people called it "ding inspection"

Interesting concept, but for my particular application the part must be inspected in process. Thanks for the idea though.

 

[joseph_e2]

I have seen a machine using an inspection like this: https://www.modalshop.com/ndt/products/ram-drop

The idea was to drop a small metal part against some surface and to detect the sound pattern with microphones. The sound patterns of a part with cracks differ from those without.

The plant people called it "ding inspection"

Did it work well in that setting? I've seen something similar done with brake pads to detect cracks in the friction material. There was a pick & place that loaded the finished brake pad onto a foam bed. Then a really small microphone extended out to touch the friction before a plastic plunger was launched into it from underneath. It was called a Grindosonic machine. Honestly, it was a pain and a little finicky. It had to be as isolated as possible with vibration damping and an enclosure and placement of the brake pad had to be just right.

 

[dwoodlock]

Have used Eddy current, but not for an application like this, was exclusively part of a secondary process related to fuel injector tubes. EC probes are pretty pricey for the stuff we were doing as they were hand made. The cracks we were detecting were in the microns range too.

The cracks you're talking about are pretty huge by EC standards. Perhaps something simpler would work like pressuring the cavity with air, and a precision flow control? We opted to go this route in future tooling so that we were identifying a crack at the source instead of in the next process hours/days later.

I also did an Investigaton/study using capacitive inspection, and it really showed a lot of promise, although I can't remember who the hardware provider was as it's been 5-10 years. They were using their capacitive probes primarily for inspecting threads in engine blocks and had a lot of success. The material and its thickness were an issue for EC detection, which is how we got to capacitive. Customer ultimately declined to move forward with the system because the investigation was meant to address a natural material defect, which had occured only 8 or so times in >20million.

 

[PLCDontUQuitOnMe]

Have used Eddy current, but not for an application like this, was exclusively part of a secondary process related to fuel injector tubes. EC probes are pretty pricey for the stuff we were doing as they were hand made. The cracks we were detecting were in the microns range too.

The cracks you're talking about are pretty huge by EC standards. Perhaps something simpler would work like pressuring the cavity with air, and a precision flow control? We opted to go this route in future tooling so that we were identifying a crack at the source instead of in the next process hours/days later.

I also did an Investigaton/study using capacitive inspection, and it really showed a lot of promise, although I can't remember who the hardware provider was as it's been 5-10 years. They were using their capacitive probes primarily for inspecting threads in engine blocks and had a lot of success. The material and its thickness were an issue for EC detection, which is how we got to capacitive. Customer ultimately declined to move forward with the system because the investigation was meant to address a natural material defect, which had occured only 8 or so times in >20million.

Pressurization would not work for us since the part is not able to be pressurized on the die.


We did used to have a helium leak tester machine for another obsolete part we used to make. It was quite the machine and was a nightmare for maintenance but was very accurate when working properly. For my current application though it sounds like eddy current sensing is probably my best option so far.

 

[LadderLogic]

Did it work well in that setting?

Cannot tell, sorry, this was not my project, just saw the machine in the plant. BTW, it also used the eddy current inspection as well.