The Skeinforge Dimension plugin subsumes the obsolete Reversal plugin’s features. At the end of each thread, if the nozzle will move more than the Minimum Travel distance (1 mm by default, which is what I’m using) to the start of the next thread, the extruder yanks Retraction Distance of filament out of the hot end at the Retraction Speed.
Some experimentation at 30 mm/s showed that 2 mm of filament would eliminate all drooling, 1.5 mm left thin threads, and 1.0 mm wasn’t nearly enough.
Similar experimentation suggested 60 mm/s as the upper limit for Retraction Speed, with the SJFW acceleration limiting parameters set for 250 mm/s2. The usual extrusion speed isn’t much faster than a crawl, so the distance required to reach a backwards 60 mm/s is:
dist = (60 mm/s)2 / (2 * 250 mm/s2) = 7.2 mm
What that means, of course, is that the extruder doesn’t have enough torque to reach the programmed speed in the required distance. Assuming SJFW uses trapezoidal limiting, it will accelerate to some maximum speed at the halfway point and decelerate to a stop at the same rate. Pegging the midpoint at 1 mm, the extruder will reach a peak speed of:
v = √(2 * 250 mm/s2 * 1 mm) = 22 mm/s
In order to hit 60 mm/s in the middle of the retraction, the extruder must accelerate at:
a = (60 mm/s)2 / (2 * 1 mm) = 1800 mm/s2
Which requires way more torque than the piddly little motor I’m using can provide.
While I could swap in that larger motor, crank the current up a bit, and goose the extruder acceleration, the current Reversal Zittage is small enough for my purposes. I’d rather expend that effort on doodling up a direct-drive extruder, but that’s on the back burner until something horrible happens to the current extruder.
One easy alternative: lower the perimeter speed sufficiently far as to reduce the pressure in the hot end enough that the current speeds can suppress the zits. Notice the difference in the pix below; what you can’t see is that the first layer has no zittage whatsoever. Of course, that means the perimeter must trundle along at maybe 10 mm/s…
Herewith, a Reversal Zittage bestiary at various perimeter speeds, with Dimension set as described above and these extrusion settings:
- 0.25 mm layer height
- 0.50 mm thread width
- 60 mm/s infill
- 250 mm/s travel
A Dishwasher Rack Protector vertical tube at 30 mm/s:
The tube’s interior had equivalent zits that cleaned out easily with a twist drill.
Some of the half-tube ends came out slightly angled with zits here & there, but remember that they’re 4.5 mm tall:
The Zire 71 Protector had a lot more infill with very few perimeter joints. This corner shows a few zits at 30 mm/s:
One of the Dr. Who Cookie Cutters showed much more conspicuous zittage on the inside of a corner at 20 mm/s:
Than on the outside of the same corner:
The zits on the other cutter fell along one edge. The inside:
And the outside:
The Dr. Who set included flat cookie presses with patterns. Although these islands show some zittage, they’re about 1 mm tall and perhaps 5 mm long:
The rest of the perimeter extrusions look essentially perfect, so these really are very minor imperfections.
The Smell of Molten Projects in the Morning 
When are you baking these cookies.
You obviously need a tester. I’ll bring the chips and beer!!
Gave ’em away, on the promise of getting a handful of QC samples. I’ll perform a careful analysis on them, fer shure!
Chips & beer & cookies? You’re a manly kind of guy…
Is an acceleration curve really even necessary on the extruder motor? The main intent is to eliminate jitter, increase extrusion speeds and reduce wear and tear on the hardware when throwing around mass in the form of a build platform. There can’t be that much mass in the stepper motor, filament, and metal extrusion gear…even in your geared stepper setup…perhaps a direct drive extruder is a more pressing matter than originally thought? :)
Should have said the rotational mass of the stepper…the motors themselves can be quite massive…at least by ToM standards!
Now that any serious machine must have at least two extruders, there’s a premium on light weight.
Eventually, somebody will crack the code on having interchangeable extruder heads in an automatic tool changer, using what’s common practice on milling machines, and then we’ll have something!
The problem isn’t the rotating mass in itself, but the requirement that mass and the required acceleration put on the motor. I think nobody (except, most likely, major companies with actual engineers on their staff building big-money machines) has a good model of extruder operation, which means all the DIY extruders get built from whatever seems to not fail immediately. That development model seems to be running out of steam.
So I’d like to figure this out from scratch. Not that I know a lot about fluid mechanics, but perhaps some doodling can eliminate obvious blunders and get me directly into subtle screwups. They make for better stories, too, which is not to be sniffed at! [grin]
There are very classic motor control equations for servo systems. As you know
I bet there are also the same for stepper motor systems.
It might make some sense to start by assuming servo motors and do some system analysis on the motion.
See what that uncovers.
Part of the problem is that we (well, I) don’t have any useful numbers to feed into the usual control-systems equations. Apart from nophead’s early tests and a few quick-and-dirty measurements (mine among them), there’s nothing out there. Unless, of course, I’ve been looking in all the wrong places, which is entirely possible.
Like, for example, what’s the relation between pressure and speed? How much compression is there from solid filament to extrusion nozzle? What’s the viscosity of that goop vs. temperature? And on and on and on…
Science, that’s what we need, science… [grin]