Back in October of 2017 (it seems so long ago, and yet so recent), the temperature started dropping, and I kept on building ribs at my slightly glacial pace. Soon, it was 40° F in the garage. This poses a bit of a problem. BTW, Deep Nerd alert -- if you think the idea of a technical discussion of epoxy mixing sounds boring, you can probably skip this one.
System Three's T-88 Structural Epoxy lists the minimum application temperature of their glue at 35° F. I can't find the reference right now, but I have read from System Three that the epoxy will cure at any temperature, but every 18° F rise will halve the curing time. So it'll cure at 40°, but it might take a while.
Test joints; the one on the right has failed badly
Along with some of the ribs, I would make up these little test joints, to be destroyed later to see if the glue holds up. I've made perhaps a dozen of these as I've made the ribs, and never had a problem with any of them. Until December.
I tested the joint I'd made along with ribs #22 and 23, the little rib nose pieces that go in front of the gas tanks, in the upper wings. The test joint separated neatly at the glue line with a mushy-jam kind of feeling as they pulled apart. What it should do is tear out the wood somewhere. This was bad news.
My first suspicion was that I was mixing incorrectly. A variety of people on the Biplane Forum had told me that T-88 is tolerant of sloppy mixing -- that is, mixing ratios that aren't exactly 1:1 by volume (or 1:0.83 by weight). I was mixing up these little tiny batches, though, and I had quickly decided to use a scale to measure my mixture when I discovered that the bottles were uneven after a few batches of measuring by eye.
So, I got more or less the cheapest little scale I could find. It measured down to 0.1g, which seemed like plenty of accuracy to me: a tenth of a gram is pretty small. I happily mixed away with my little drug-dealer's scale, until that failed joint happened.
Then, I stopped what I was doing, and started testing my technique. First, I deliberately mis-mixed a batch at a ratio of 1:0.9 by weight, which I was sure I'd done multiple times on previous ribs ("It's very tolerant of ratio errors," they said). I tested these joints after varying degrees of being left out in the cold, and brought into the relatively warm house, and they failed in the glue line regardless of temperature. So I definitely have some suspect ribs in the stack, though I hadn't been taking notes on my glue weights until after I discovered this problem, so knowing which ribs are affected is impossible.
However, compounding my confusion was the fact that I'd tested a handful of joints previously, and they'd been good, and probably (or maybe not?) at least one of those joints was made with a poor mixture. So now I was unsure. The house isn't kept ridiculously warm, perhaps 66-68° F, and that only for the parts of the day we're actually home and awake. Perhaps the temperature just wasn't getting high enough.
One thing I was reasonably sure of was that the low-resolution scale was causing problems. On my test joints where I mixed at 1:0.9 instead of the recommended 1:0.83, part of the reason my mixture was so off was the scale. I had poured out 1.0g of resin, which meant I should have been aiming for about 0.8g of hardener. Ideally, I'd pour out exactly 0.83g of hardener, but the best I could hope for was 0.8, with only decigram resolution. As I was carefully squeezing hardener out of the bottle, the scale jumped from 0.7g to 0.9g. How much over 0.9g? Who knows! I shrugged, and made my test pieces, figuring this was a valid test.
So, the next order of business was a more precise scale. I located another inexpensive scale, but this one read down to milligrams (0.001g). I didn't need that last digit, but better too much precision than too little. This scale, at least, wouldn't jump from 0.7g to 0.9g without ever showing 0.8g.
Of course, the larger the batch of glue you mix, the less important it is to have a super-precise scale. Unfortunately, for now, I need to mix a roughly 3.0g and 2.52g batch, and it doesn't take much variance on the hardener to take the ratio far from its ideal place. The new scale gives me much greater confidence that I've got the mix right. My tests with the 1:0.9 batch are too worrying to allow that kind of mixing error to continue.
Discussion on the forum started to convince me that what I really had was a temperature problem. That my summer ribs were all good because they'd cured in a 60-70° F average temperature, and the cold Northwest winter was causing my problems.
The Easy Bake™ Rib Oven
I pondered my possible options, and decided I would make the grown-up version of that venerable child's favorite: the Easy Bake™ Oven. I cut up a sheet of plywood I had sitting around from some previous project, and scrounged around in my project supplies to come up with a pair of light bulb bases, an electrical box, a short electrical cord, a discarded computer fan, and some Romex to connect it all together.
In two days of casual bodging, I put together my Easy Bake™, and after scientifically determining that it wouldn't catch itself on fire by leaving it on overnight with nothing inside, I stuck a temperature probe inside the box, and measured 82° F. Perfect.
My next test was to make up some test joints and put them in the Easy Bake™ to see if that made a difference. I mixed up a very close-to-perfect batch of T-88, and assembled three test joints: one with a thick layer of glue and no pressure before stapling, one with heavy glue and normal pressure, and one with a normal thickness of glue and normal pressure. I had read over the T-88 FAQ again, and developed a vague fear that I might be suffering from glue starvation, where there's not enough epoxy in the joint to form a good bond. A starved joint sounded like it might fail in a way similar to the failures I was seeing. All three test pieces went into the oven for a days' cure.
The heavy-glue, no-pressure joint failed on one side but not the other, which suggests that the glue at least isn't the issue (and one day may not have been enough to fully cure it, as System Three says it takes 72 hours at 77° F for a full cure). The heavy-glue, normal-pressure joint broke exactly where it should, in the wood. Likewise, the normal-glue, normal-pressure joint failed in the wood, confirming that at least my normal technique wasn't causing problems.
So, with any luck, I now have a system which will result in full-strength ribs going forward. Doesn't help me with past ribs that may or may not be strong enough, but I can deal with those later. At least now I can bake my ribs to be sure they're getting the cure temperature they need, with a mixture of glue that's as close as humanly possible to perfect.