Hadron Collider repairs cost £14m

They should have X-rayed all their weldings. I know this because I have an aluminium welder diploma besides a laurea in theoretical physics.
Tullio

You would think that would be a given... considering the cryrogenic temps the component would be operating. I remember all the failures in the early space program that were blamed on bad solder joints.. you usually learn that lesson when you have to build more than one..:-)

TIG and MIG but I am told today there is a better method. My diploma was obtained in 1980 at Istituto Sperimentale Metalli Leggeri in Novara. I agree with Dagorath, most inspections are performed with portable gamma ray sources.
I remember using a Co-60 phial to get neutrons out of a beryllium target. The neutrons were to calibrate a bubble chamber neutron detector we were building at the Istituto di Fisica Sperimentale in Trieste under contract with the International Atomic Energy Agency in Vienna. Being a theoretical physics student, I was given a few germanium transistors and the task to modify its electronics from vacuum tubes. I remember soldering and testing a resonant circuit. Happy days!

Common lead/tin solder can be used to join dissimilar metals/alloys, for example steel and copper, when dissimilar metals do not weld easily. Usually dissimilar metals do not weld at all or else the weld is so weak/brittle you're better off using epoxy or even duct tape. Even if the joint was copper-copper, as many electronics joints are, the solder itself would likely be a lead/tin alloy, lead/tin/nickel alloy, silver/tin alloy or some other alloy with coefficient of volumetric expansion different from copper. However, soldering and low temp work has been around for years and engineers know how it works by now and know what to do and what not to do. So, in my opinion, it boils down to poor adherence to procedure (solder jockey error or laziness) followed by quality control mising a bad joint. Add some internal stress due to unequal expansion and pop goes the joint.

Soldering is, in many ways, more difficult than welding. For many metals and their alloys, typical welding temperatures melt or burn off surface contamination (metal oxides and foreign material such as oil and bird poop) long before the metal itself melts. Therefore the flux (if you're using one) removes the contaminants from the weld zone long before the base metals flow and mix to form the joint. That's why top quality welded steel joints are so easy. Aluminum is different because aluminum oxide melts at lower temp than aluminum or any of it alloys. So when welding aluminum you always have to deal with those nasty non-liquid contaminants that don't want to flow away from the molten base metal. The new solid-state power electronics in new inverter based (transformerless) welding power supplies are able to produce waveforms that blast away contaminants better than ever but not everybody is savvy about that so they weld aluminum the old crappy way. This is perhaps what Tullio was referring to when he said there are new and better ways to weld aluminum.

With soldering you have the same problem as you do with aluminum... the surface oxides that do not even come close to melting because soldering temperatures are nowhere near the melting point of metals or their oxides. So you use a corrosive flux to remove oxides but that doesn't always work like it should. Or you use an abrasive to remove oxides/contaminants but then you run the risk of embedding abrasive in the base metals.

Getting back to the difficulty of using radiography (X-ray or gamma ray) to look into the joint to see if there are faults... radiography works when the joint is of fairly uniform cross section and of materials that are equally transparent/opaque to the radiation. If the cross section is non-uniform then the picture will have areas that are over-exposed and/or areas that are under-exposed. If you shoot obliquely through several sections then you get shadows as well as over/under exposure. The result is a pic that tells you nothing. Now imagine the difficulty in getting a decent pic of stranded copper wire impregnated with solder. The variation in density would, in my opinion, make it impossible to get a decent pic of the interior of the joint. You can use ultrasound instead but if the operator doesn't have the skill then on non-uniform joint shapes and densities he'll miss lots of faults and I know that because I've put many of them to the test by purposely leaving a fault in a joint then having it examined with ultra-sound. It's depressing to see how often they fail to detect the fault. BTW, I would either throw the test joint in the trash bin or repair it.

So that's why wings fall off of brand new airplanes and why chemical plants explode :) Be nice to your radiographer and test your ulta-sound tech every day.