More experiments -- notes after the pics (and some notes about kit release timing):
A question I've been asked by the few folks who've had access to development-level parts is why I haven't designed the rocket body elements to have their connecting parts printed into the body shell parts. The main reason I've given is that I had designed most of these parts to be as tall as the printer can work, necessitating separate "mating collars."
Since I "locked in" the original beta design, I've been able to push the "printable height" I can get out of the bot another 2-3 cm. While I'm working on finishing a demonstration of the new fins and engine section, I decided to do some experiments to see if I could print the mating collars directly into at least some of the rocket body parts. These pics show the first results that are close to what MIGHT be usable.
The first pic shows all of the tank section parts printed in this way. The second pic shows the long, straight-sided first two parts mated. At the level of simple printing, these experiments are a success. The parts come out with acceptable outer contours, and fit snugly. But there are some downsides.
First, one of the challenges that made me not do this before is that the interior contour of the parts has to take into account the maximum overhang that the printer can render in plastic (c. 65 degrees or so). Given this, the print time for these parts is a little more than the sum of the print times for the shell and internal parts separately, because I have to design in a transition from the shell to the mating collar that doesn't exceed that limit.
Second, there's the print time itself. I had gotten the print time for the largest tank section parts down to just over an hour. Now the time for the largest of these parts is three and half hours. By itself, this isn't a problem: the Makerbot is far more reliable than it was formerly, and now that I've identified the need to do long-print-time parts from a cold start, these prints come out without flaws. But long print times do put an additional constraint on my work flow, since I have to find blocks of time in my actual real life to accommodate them. I'm a very early riser, and often work from home in the mornings, so this is something I should be able to accommodate and still stick to the 1-kit-per-week goal I've set for myself.
Finally, though, is an issue that is a consequence of these factors: part strength. The design for the body shell parts I'd locked in some months ago was as thin as I could possibly make it. At the time, I was doing this for reasons of reliability: I had significantly fewer failed prints for parts that printed in under two hours. With the many upgrades I've done to the printer as Makerbot has refined their design, I can start a 2-3 hour print now with high confidence that I won't get a glitch while the printer is executing the part.
But there are real problems in transitioning from a super-thin-walled element of a part to a thicker-walled element, which is inherent in printing in the mating collars. The thin-walled body shell parts get their cylindrical rigidity from the mating collars. Without them, they have about the same x-y crush strength as a thin cardboard tube. With them, they're as strong as thick-walled PVC pipe.
When the two elements are merged into the same printed part, the adhesion between layers degrades in the zone where the part transitions from thin to thick walls. There is much technical detail here, but suffice it to say that the angle of transition is more important than just addressing maximum overhang. The "fill" between shell elements has a LOT to do with whether the layers and shells will have good adhesion. The bottom line on this is that, while the parts pictured above LOOK good, they're actually not strong enough to make it into a released kit. I've already had to inject some CA glue into delaminations at the transition between thin and thick walls on these parts to work with them.
Right now, I'm experimenting with altering the "angle of departure" to get the best fill to combat delamination. Whether that will be a fruitful path of experiment is yet to be seen. With each experiment taking 2-3 hours to print, this is a painstaking process. There's another line of experiment I MIGHT try, which is to tinker with the bot's fill settings for these parts. I'm loathe to do this, though, because my goal has been to design the parts to all work with the same print-control settings. If I have to use multiple print-control settings for different kinds of parts, it adds to the things I have to keep track of in a real kit production environment.
So, for now, I characterize this as pure experimentation, which might or might not make it into a first release kit. If it does, it will reduce the part count in the tank section from nine to four parts, slightly reduce the effort in kit assembly and will decrease significantly the chance that a builder might make a mistake in assembly requiring replacement parts.
Meanwhile, real-world work has once again thrown a kink into my plans for a release date. As some here know, I pay the bills with work as a trial lawyer. One of the problems with this is that my schedule is at the mercy of the courts I work in. I have a major case that WAS set for trial in the first part of December. This trial would have been finished before Christmas, which would have given me the holidays to lock in a final release design, revise and complete kit documentation, and start taking orders and printing kits for sale in January. That case has now been reset to the second half of January, and the scope of the case has increased beyond what it already was. This will mean that I will probably not be able to get to actually selling kits until the second part of February, at the earliest.
... thus it is with garage builders . . .