For future reference, I've just had confirmation from customer service at Permabond that the correct adhesive for structural bonds between, for example, metal and composites is their ES550 and ES558. They are described as:                

If you are adhereing it to smooth (albeit degreased) aluminium then I would have expected the results you have had. This is why we advise keying BOTH surfaces.

We have had people use it for quite high loadings when joining tubes, certainly in the hundreds of kilos range on short tubes as well as additional shock and vibration.

Both the glue and resin are epoxy.  What i think has happened is the thinner epoxy resin mechanically bonded with the fine scratches on the surface of the aluminium, where as the more viscous epoxy based adhesive won't have flowed into the scratches as they are too fine.

Warren Penalver
Easy Composites / Carbon Mods - Technical Support Assistant

I generally consider an adhesive to be something which adheres with a chemical bond, in this kind of setting. Hence I roughed the carbon with 240 grit and degreased the alloy in acetone. Using an adhesive that bonds, this should be sufficient. It would appear that the 515 adheres in the same manner that a silicone sealant does, and not with the tenacity of a cyanoacrylate or - even - the epoxy resin used with the carbon fibre which seems to actually adhere to aluminium pretty well.

Of all of the miles ridden on the framework built, the only failures I had were on joints where the ET515 was used, which I had to reinforce, en route, with garden wire (fancy).

What I learned is that carbon is a material that can be used for fabricated structures, however that any interfacing between it an another material is past what I can handle with my limited experience. ET515 is not the right adhesive, this is for sure. We traveled over 250 miles with 20+kg of luggage on carbon framed racks www.laureltache.tumblr.com

at over £30 when you include the syringe, nozzles and adhesive, ET515 is a very expensive mirror-tack.

As above, I've used that to bond halves of wings together and not had a failure yet

how big is the gap?? I have used ET515 personally for bonding inserts and its been a good bond.  You need to key the alloy properly. Some people go as far as putting grooves in or even knurling the end to be inserted.  

I guess you are using roll wrapped tubing?? Almost all roll wrapped tubes will come from the factory with a heavy coating of release agent still inside the tube, so you really need to ensure it is gone to get a good bond.

Some photos would be ideal.

If you're still having problems getting a good bond after that, then return us the glue and we will test it. It is theoretically possible to have a bad batch, although this has not happened to us yet.

Warren Penalver
Easy Composites / Carbon Mods - Technical Support Assistant

I will post an image of the part on Friday.

When bonding any metal I still use the wet sanding technique using 240 wet and dry sometimes 120 and sand metal with epoxy prior to bonding it really etches into the metal surface,belt and braces on my part.

Hmm, this was a 20mm carbon tube, and the join was to a strut coming in below, by means of an aluminium sleeve which wrapped around the tube like a clamp - made from 20x2mm stock.

Hello Tobias

It would appear that the ET515 has no structural value at all, as it seems to neither bond to degreased aluminium nor to roughed carbon, as in I can scrape it off either with my fingernail after three days of curing. I only discovered this when I had a catastrophic failure of a rather beefy join.

I kind of feel that the ET515 is no stronger than Araldite, no more nor less rigid, and in a similar manner is no more respective of its apparent 15 minute working time.