What they said...but a tale of caution
Mustang said:
The way I did it, was to dimple the spar first, then I countersunk the doubler, and laid the doubler onto the spar. At first, it would not lay flush on the spar, so I countersunk again incrementally, until it laid flush to the spar and then I clecoed and riveted. It came out perfectly.
Cheers, Pete
That sounds right...but in some places be careful you don't overdo it with the countersink. Sorry for the long tale that follows, but stick with it...it might not apply directly to you right now, but there are important points at the end that don't make sense unless you read the whole thing.
I've been helping to build an RV-7A for a couple of months, stepping into the project right at the start of the fuselage construction. On the firewall (which is steel), you need to dimple the firewall itself and countersink the stiffeners that attach to it.
When the countersink was done so that it looked right (rivet just a hair below the surface, with a barely detectible shiny line around the edge), the stiffener would not sit quite flush with the firewall surface (it would have if the firewall were aluminum, we believe). When we looked at the clecoed assembly edge-on with a light behind it, we could still see the dimples in the firewall, like little posts, with light between them. We asked another builder who knows more than we do about this and he said, "light between the stiffener and the firewall is bad...the strength of the assembly comes from the fact that the two are flush against one another." (It should be noted that he did not have an opportunity to actually see the parts...I described the situation to him in great detail, though.)
So, off we went, aiming to eliminate all gaps between the firewall and the stiffener. We countersunk a bit deeper, and still saw daylight between the stiffener and firewall, but a lot less...deeper, and still a bit of light...deeper, and common sense finally took over. The countersinks were now right at the ragged edge of going all the way through the stiffener and enlarging the hole. Worse, they were way too deep...they weren't engaging the dimple, and the stiffener was no longer positively located (i.e., you could slide the stiffener around to an alarming degree with it laying on the dimpled firewall, but not clecoed). One trashed part, the replacement for which still has not arrived. We concluded that the steel firewall was not completely flat (duh) and that the material was stiff enough that it wasn't going to get completely flat. So the correct course of action was to maximize the contact area while keeping the dimples in contact with the countersinks.
What we wound up doing in this particular instance (meaning the firewall) was to split the difference. The "standard" countersink was clearly not deep enough, because the stiffener was sitting too high off the firewall...there was little physical contact between the stiffener and the firewall itself. Making it a bit deeper broadened the contact area considerably, while maintaining contact between the dimple and the countersunk hole. We arrived at the appropriate depth the way we should have done it in the first place, by taking a piece of scrap and doing some trial-and-error tests on it. We went until the countersunk piece just barely lost contact with the dimple and then stopped. We figure that the very slight loss of contact between the dimple and countersunk hole is okay because the act of setting the rivet will push the dimple down into the countersunk hole and re-establish full contact.
SO...the moral of the story is twofold:
1) The point of all of this dimpling/countersinking business is to get the parts to mate with one another completely flush, because that's where the strength of the assembly comes from. Countersink deep enough so that this is accomplished.
HOWEVER...
2) There are two things going on with a countersunk/dimpled assembly. One is that the parts are supposed to mate to one another in a flush manner. But they're also supposed to fit together precisely, peg (dimple) in hole (countersink). If you go too far with the countersink trying to accomplish a perfectly flush assembly, you lose that custom fit.
When in doubt, try a new idea with a piece of scrap before proceeding to mangle a real part.