calstar 

Given tubes of the same size and mtl type how big does a fillet need to be for optimal strength? There is a point at which the fillet's increase in size becomes aesthetic rather than structural(a point of diminishiing returns), yes?

thanks, Brian

unterhausen

ETA: this statement is based on reading technical literature and not actual testing. I like to make medium size fillets.
The main strength of a properly brazed fillet is in the thin area between the tubes. You can replace some of that strength with big fillets. In conclusion, big fillets are stronger, but you probably don't need them.

I noticed in your pictures that you were having a bit of a problem building a fillet; wouldn't hurt to do the tower building exercise where you try to build a tower just out of bronze. It's worth the time to get that control and then you can decide how big you want to make them. You can also just add flux to your already made practice pieces and make them bigger to see what happens. There is a fine line between a perfect fillet and a collapsed mess so practice is important

Andrew R Stewart 

The classic fillet size description that I've heard many times is the fillet should be 3 or 4 times the wall thickness. So a tube with a 1mm wall would have a fillet of 3 or 4mm in size. This is smaller then what most do, in the area of a TIG weld size. I try for medium to small fillets myself.

Far more important then whether the fillet is 4 or 6 or 8mm is whether while finish filing the tube gets undercut, So besides the practice of laying down the fillet one should try to file a few also. You'd be surprised how easy it is to file down the tube at the edge of the fillet. One way to avoid this is to file only on the filler, easier said then done. Another reason to practice the filing is to gain motivation to make nice looking fillets which require less filing. (I have a friend that was employed in a short run production frame company 25 years ago. He hated to do the filing of others' brazing. There was no feed back loop to train the others to do cleaner brazing).

I also agree with Eric's suggestion to make brass towers or other lumpy shapes off a surface. I've made a heart out of brass for my honey. Andy.

Andrew R Stewart 

Brian- Another practice is to build up a lump of brass (this could be in a joint of two tubes or on a surface) maybe 1/8" high. Then using the flame go around this lump and flow it out bit by bit. I find I turn down the flame after flowing the joint (to get an internal fillet) before I go back and lay the fillet down. I'll add brass along about 1/2" of joint length, trying to keep the brass humped up (not trying to get it to flow out and look like a final fillet) then go back and work this length of lump out onto each tube surface. With a small flame you can get real close to the joint but be willing to back off, flick away, the flame often to control how hot things get and watch out where the tail of the flame is going to not ruin what you've already done. Andy.

calstar 

Is the 3 or 4 times the wall thickness measurement the radius of the fillet taken where the two tubes meet? Lets say the tubes below have a .035 wall thickness(probably pretty close since its a tt and ht) then the radius at the fillet is .105. Is the radius/thickness of this fillet too small to be structurally sound with the penetration shown? I'm trying to get an understanding of what a small, structurally sound fillet looks like.

thanks, Brian

MassiveD

The normal thing is the depth of the fillet on a point that bisects the joint, the sides are different.

Normal welding practice is to test parts to destruction. It isn't just what some book says it is, but whether you can actually pull off that dimension. The normal simple destruction test would be an unsupported t type joint, but that is obviously a very severe test relative to a triangulated frame, still can't hurt. Might also explain why the fillets are the size they are.

There is a lot of talk in certain circles about certification for frame builders, but it would be a useful step before getting to that point to just have some simple answers to questions like this so that people who want to develop their skills have something to shoot for.

unterhausen

it's not actually a simple question. The joints above are not going to fail in response to a single pull. They may fail in fatigue. I have considered testing things like this in my fatigue machine, but getting the pull so it doesn't go off axis is tricky

Andrew R Stewart 

My understanding was that the fillet size was measured along the tube surface. Joint crotch to thin edge of the fillet. So a joint with different tube wall thicknesses would have a fillet with different lengths (or a fillet that was a "French" curve in shape). But I really think that when people start to delve this deep into the structure and design of joints their process gets to the point that they build strong enough joints. Or that the caring to learn of the science make one try to get the hands on righter then those who don't bother to think much about things. Andy.