Tuesday, July 24, 2007
Second pass at the Mk 2 AEM polymer pump
There are some tricks to getting Art of Illusion to do what you want with Boolean operations to make an object like this. I'll write them up sometime when I have a moment.
I've included a GM9 at the right of the pic. It is very like the one Darwin uses on the Mk II extruder though with a different gear ratio. Above and to the left of it is the GM-17 which I plan to use on the Mk 2 AEM extruder that I am presently designing.
I took a few moments and ran the GM-17 at 5 and 9 volts. That thing has HUGE torque. I couldn't stall the drive shaft with my fingers running it at 1.5 volts. I think that it is going to work just fine for pumping HPP and ABS. Dan was right. The mounting holes for the GM-17 are both further apart and at the other end of the gearmotor from the GM-3.
The tiny little pair of motors with brass gearboxes are GM-11a. Those are expensive little mites at $16.80 each. The trick is that they deliver 7.5 oz-in of torque and, I think, will drive the Tommelise xy positioning stage on a diagonal at about 6 mm/sec. As well, these tiny little motors, unlike the ones I am using now, are almost impossible to hear from a few feet away. This is a great idea if a kid decides to print something overnight in his bedroom.
Friday, July 20, 2007
Designing the next generation Mk 2 AEM
That said, I've been thinking about what the next generation extruder ought to look like. Adrian's Mk II was designed to extrude polycaprolactone (CAPA). While CAPA is a tough plastic you can cut into it with the metal thread of 1/4 inch studding relatively easily. That means that the springs that clamp the polymer pump to the filament guide can be lightweight and the torque requirement on the studding pump aren't huge.
HDPE is a tougher plastic than CAPA. It didn't take me long to realise that the #4 springs that worked well with CAPA were not quite adequate for HDPE. I sized those up to the toughest #6 springs that I could find and have since been able to do production extrusion in HDPE. Mounting #6 springs on #4 studding isn't a happy sort of thing to do, but it can be made to work.
The Solarbotics GM3 gearmotor handles both CAPA and HDPE fairly well.
It's safety clutch is set to 60 oz-in of torque. This is way more than you need for CAPA and maybe about 10-15% more than you need for pumping HDPE at the rates that I currently use. I've been able to barely pump HPP (homopolypropylene) at production rates with the GM3. ABS, however, is out of the question.
(Read the full story)
Sunday, July 15, 2007
Bushing wear on the Mk I AEM
You can see the wear circled in red in this pic. The steel bushings seemed to have suffered little more than the odd scratch on the contact surface, a result that leads me to believe that the bushing steel is considerably harder than that of the studding.
(Read the full story)
Wednesday, July 11, 2007
Here you can see the two bits that I want to replace with one new printed one circled in red.
(Read the full story)
Tuesday, July 10, 2007
Definitely a McAllan's 12 year old single malt evening...
Little did I know at the time that it was going to take me another three and a half months to print anything useful with it. Neither did I have the slightest idea how much pain and suffering was going to be a part of those months, either.
Well, it's done now.
Being a congenital idiot, I wanted to go Vik one better and do a part of my polymer pump first off. You can see here what I had in mind. I did some modifications to Adrian's excellent Mk II design to let it work with HDPE and my own perversities. I had to use several chunks of wood to make the Mk II into the Mk I AEM extruder.
I decided rather than simply duplicate a Mk II that I'd redesign it a bit and get rid of that chunk of wood that you can see circled in red just above.
I designed the thing in Art of Illusion and made a straightforward rush to just print it way back in March. Biiiiiggg mistake. I guess I made half a dozen rushes to print that darned thing coming at the problem from half a dozen directions.
Finally in June I decided that I had to solve the corner curling problem with HDPE or I was never going to print anything worthwhile. On 24 June on my 60th birthday I cracked that problem. I had to use a loose diagonal infill to get it to work, but work it did.
Since then I've not had a lot of time to work with Tommelise, what with contract deadlines and the like. Last weekend, however, I'd got a little bit ahead and decided to give it a go one more time.
(Read the full story)
Monday, July 09, 2007
Lamarckian evolution at work... :)
I created a 2cm cube in AoI to use as a test object for printing - it's big enough that you can see what's going on (and what is going wrong), yet small enough that it doesn't waste too much material. Below is a picture of several stages of troubleshooting the printing parameters and software while printing the 2cm cube, as well as two rather laughable attempts at a minimug:
The square-and-blob at the far left is an example of what happens when you try to print at an angle with the current software if your XY speed is set in the high range (as it is when repstrapping using threaded-rod drive for the X and Y instead of belts - my XY speed is set to 244). Since the speed setting is not very linear at the high end, when the software adjusts the speed to account for driving the head at an angle, it tends to slow the motors down far more than intended, and you end up with puddles instead of nice pretty lines. I have created a quick fix for this in the software, but since it is only necessary for repstrap machines that use threaded rod for the X and Y axes (Darwin uses belts) and Adrian mentioned fixing the non-linearity of the speed settings at some point, I won't be providing the fix as a patch into the main code unless asked. If any of you decide to repstrap using threaded rod for the X and Y, please feel free to contact me and I will help you add the fix into the software for your setup.
The second square in the series is a good example of what happens when the extruder feed is too fast and the head is too close to the table - it's kinda hard to see in the photo, but instead of nice distinct lines, everything is a bit smeared.
The next squares in the series show the progression as I tweak the various settings - getting the initial Z position of the extruder to the right spot, adjusting the extruder speed, and getting the layer height right.
The minimug attempts at the top of the picture were done partially to see what would happen, and partially because I had not done more than a few layers at a time yet, and wanted to know what changed or went wrong ten layers up or so. The rather wobbly extrusion you can see at the upper left of the short mug is a result of the layer height being incorrect - as the layers progress, the extrusion head is placed farther and farther away from the object being printed, and the plastic just goes everywhere. The 'lacy' look of both mugs is partially due to the extruder head height, and partially due to an apparent bug in the software that causes it to pause between sections of a circle as it is being drawn - the extruder is extruding too fast for the XY movement, and since the head is a ways away the thread tends to wobble back and forth. The threads that are crisscrossing the hole in the middle of the mugs are due to the head moving from one side to another when starting on the next layer - the CAPA tends to leave a long string behind, but this is quite easily cut away after the extrusion is done.
Below is a closeup of the rightmost test cube:
As you can see, it is looking a lot more like an organized extrusion. It is about 10 or 11 layers, and I am quite happy with the infill, and the outline extrusion is much closer to what I want to see. Unfortunately, I have a bit of drift in the Y axis (you can see the effects in the leaning of the minimugs), but hopefully I can clear that up by adding some oil to the Y slide - it's a bit stiff in a few spots, and I think it is making the Y motor skip a step or two here and there. Hopefully I can get the Y axis drift sorted out in the next few days, and I'll be drinking a toast with a new minimug!
Printing a polymer pump
If this works it is going to be very crude. That's all right, though. This is all about developing knowhow with HDPE.
So far I've printed the raft and am just about through printing the base level of the pump.(Read the full story)
Saturday, July 07, 2007
I'm gettin' places!
Okay, today has been an excellent day, and not just because it was so fine to go down to the beach riding my motorcycle.
Today I opened the boxes that arrived this week, with new products for me to go *Frankenstein* with my resins (meet my resin bottles to the right).
1st box: N-methyldiethanolamine, the stuff that stops oxygen scavenging in it's tracks.
Remember how that effect was slowing down the setting and hardening of my resins, specially for polyesters? If not, you can read about it on the post below. Well adding a little N-methyldiethanolamine has reduced the curing and hardening time from something around 40 minutes to.... UNDER 5 minutes!
Here is my big rig with the stronger fluorescent tubes, the one that's on here is the germicidal UV one.
With this setup I will nos start fine tuning the mixes so as to get something that will work perfectly well in a syringe and will be fast curing!
It's really close now! Really, this thing hardened so fast that I literally didn't see it happen. In fact it was so fast that it wrinkled the surface of the resin because it hardened before the inside got time to do so.
Just a minor problem that will be tackled in my next experiments.
2nd box: Titanium Oxide powder, a common white pigment that should act as both a filler and a co-initiator, improving depth cure. Well that hasn't worked so well. It has such a diffusion power that the UV radiation gets swamped before it can make its effect. This pigment has some serious covering properties. I will need to use a lot less to see the desired effects. To be continued...
3rd box: Copper powder. This one is straight forward: I wanted to fill my air-drying acrylic i bought the other day with a fat load of copper powder to see if i could get a conducting thread.
On this picture yous see a number of threads i spread over a Polystyrene board with my copper filled acrylic resin, after 1 hour drying. The shiny part that looks like copper I achieved by polishing the hardened thread a little, with a rounded metal object. Isn't that just the most amazingly unexpected thing? I was not hoping this good a result! And I didn't even load the resin to the maximum ability! It's loaded with about 50 volume % copper, i could probably go up to 70v%! And you know what? you could probably use any regular glue to get this stuff, it's so damn easy I tell you!
Now, not everything is perfect. The threads still have a quite high resistance.
Here I made a measure of the polished track's resistance. those are 10 kOhms by the way... And the non polished, quite porous thread has a resistance through the roof! :$ Anyway, the thread varies in resistance and sometimes it's a lot lower. This thing needs to be copper plated still, and that should get us where we want. By the way, the thread sticks pretty nicely on the board, but maybe using strong glues (epoxy?) would give us a better result, specially at very high copper loads.
Anyway, I'm very happy with my results today, I feel like I made a big leap! Cheers to you all RepRappers, and thanx for all your ideas and inspiration! I'll have some drinks on us all to celebrate this :)
Tuesday, July 03, 2007
Round two: UVC versus photoinitiated resins
The Philips TUV-36W wich delivers a good 12 Watts of UVC power at around 253nm, which is Benzophenone's excitation peak and also Mechelers Ketone's secondary excitation peak.
This tube costs a hefty 50 euros, because it's glass walls are made of pure SiO2, to let through UVC radiation, when normal glass would block it.
I have also purchased a Black light tube of 40Wats (PHLF40BL) delivering 13Watts at the UVB frequency of Mechlers Ketone's main peak, at around 360nm.
Results are: UVC is a LOT more effective than UVB!
I divided the hardening time by 3 using Benzophenone and the UVC lamp. This configuration has proven to have a much higher efficiency.
Benzophenone is also cheaper than Mechelers Ketone and safer to handle, which compensates for the higher price of the lamp.
Another advantage of Benzophenone is that it dissolves a lot better in the resins I have tried. Actually, I have not reached saturation levels yet with 6 weight % BP, where as in polyester at 3w% for Mechelers I'm already close to the limit. Equally, Mechelers seems to have very little solubility in acrylic resins, which is a very negative point.
Things to be careful about when using UVC: Don't be around when that thing is on. UVC is a really energetic radiation and it will eventually burn your skin and retina. It's one of the radiations coming from arc-welding (although less intense). It also leaves a distinctive smell of Ozone in the room, like in a printer or photocopy room.
Ventilate the room after or during the operation as Ozone is not to healthy. To filter UVC, you can use regular glass panes, but not plexyglass.
To attack the problem of oxygen scavenging I was seeing with previous UV curing (in which air-oxygen solved in the resin impedes the catalyzer to do its work) I decided to try an acrylic varnish that you get from art-shops. This varnish dries on contact with air, so that I expected the combination of this varnish with my photoinitiators to improve setting times.
I was correct! Another 4 fold improvement in setting times. I goth this one to set and harden in a little over 10 minutes, with half the photoinitiator content I'm using for my regular resin.
The resulting hardened film (around 0,5mm thick) was not very tough and had yellowed a bit, as this was a surface varnish, not really thought for thick sections, but i guess that a very attractive research path would be to find out what commercially existing air-drying acrylic or alkydic varnishes could be used for our Reprap needs...
Anyway, this week I'm receiving the Diethanolamine I ordered 10 days ago(prevents oxygen scavenging) as well as Titanium oxide which will be a very useful additive to my resin mix as it should accelerate the setting as well as increase the depth of curing in the resin films.
I'm also getting copper powder which I will use with the acrylic varnish to create conducting polymer films...
It will be an interesting weekend, a lot to do!!
Monday, July 02, 2007
It's Alive II - sort off...
"Richard" is assembled and has been tested. Things are moving, but there are still a few major problems. Not enough capa is being extruded, squares look like parallelograms, belts skip - sometimes even off the (Z) gears - and there is some odd behavior when starting on a part.
I guess some major tweaking needs to be done, or this is where the inaccuracies of my tools and skills really make a difference.
Anyway, it is still a pretty sight, specially when it moves around on my command :-)