Taz_Mega - Unofficial large-scale Taz-derived machine.

Design work towards substantially upgrading my Taz3 has begun. This will eventually support at least one direct drive extruder, four bowden extruders, a CNC mill spindle, and a <1µm resolution laser autoleveling/lithography tool.

Some gantry brackets have been designed thus far. IMHO, AlephObjects should really consider reinforcing the frame with something similar.

Next step is to implement a better Z-axis. Currently planning to port my FlexReplicator Z-axis system, since it provides both anti-backlash and inelastic coupling, which even the Mini/Kauri does not seem to offer.

Interesting design concept. Are you noticing alot of flex with your existing taz frame? I’ve not noticed any on mine in particular. Which part of the frame seems to be flexing the most?

All Taz machines are adequate for 3D printing in their stock condition. However, particularly for the older Taz3, that seems to imply some routine maintenance during normal use.

Although not a flexing problem, Z-axis backlash is just enough to compromise first-layer alignment at the start of about one-in-five print jobs. Added to this, the Y-axis can deflect several millimeters under light presssure. Overtime, it seems to change position significantly, requiring steel shims or washers to maintain orthogonality. Gantry frame flex (side-to-side) is not particularly severe, but it could be contributing to the Y-axis slipping out of orthoganality.

Taz5 rigidity is comparatively outstanding, better than any of many 3D printers I have access to. However, the frame, Y-axis, and especially linear rails still deflect too much for my use case, deforming slightly under light pressure. CNC milling would probably not be good for the health of the machine. Fine nozzles, precision laser cutting/lithography, an similar uses would likely require regular manual recalibration.

Notably though, the LulzBot Mini is much stiffer than any of these machines, probably adequate for light CNC milling. Honestly, I am a bit disappointed the sheet metal frames have been dropped as Kauri was simplified to Lancewood.

The biggest culprit of Taz 3 layer slipping I found before I upgraded to the 4/5 style ends was the Z endstop target screw. it didn’t have enough friction to hold it in place, and it would push up if heated too much. the larger M5 screw on the 5 helped with that aspect quite a bit. The 3 style X to Y frame mounts also can move significantly as I later found out. The 5 style ones fix that particular issue. The rod deflection is definitly not a great thing though. I like the openrails I have bolted to mine now, they work great!

My Z openrail mod is going to incorporate several back stiffener ribs I think, mainly to keep the rail itself in allignment, but potentially to stiffen the main frame as well. I might end up adding some extra corner bracing as well when i have things taken apart to install that.

Interesting. However, attempting to go through the semi-automatic leveling wizard, I found the Z-axis drive mechanism did have significant backlash as well.

Looking forward to seeing your Z-axis openrail mod.

EDIT: By the way, what about C-Beam for the Z-axis?

releasing designs under a free software licence, designed using software that is not free software

I will never understand this.

You seem awfully critical of Aleph Objects. Keep in mind that every design has compromises, even yours. Manufacturers compromise between price, source-ability, performance, manufacturability, aesthetic, and more. Making one machine that can hit [insert overkill specification here] is totally different than making thousands upon thousands of them, in a way that is self-sustaining as a business, not to mention selling, distributing, marketing, supporting, and developing on them before someone copies you and undercuts you on price.

If you are interested in input from or collaboration with Aleph, it’s not a great idea to start out by saying in essence “look at all of these problems with your merely adequate product in which I am disappointed, you should really have done [insert your relatively inexperienced opinion here] instead, which is [insert way in which the machine is improved, supposedly]. Oh, by the way here is my half-baked design which currently does nothing, but is way better.”

What makes a machine good is not rigidity, speed, price, features, reliability, efficiency, safety or anything else that can be quantified. Those things are very important, sure, but not as important how you feel when you use the machine, which can come from the performance that the specs enable, but more importantly comes from things like whether the machine can do the things that the manufacturer claims it can do, whether the machine enables you to do the things you acquired the machine to do, how easy or difficult the machine is to maintain, whether the machine was developed in a way that resonates with your philosophy (as is the case for many Lulzbot customers, I’m sure) and the practical benefits that that philosophy entails, like in the case of free hardware companies like Lulzbot, freedom to use it however you want, access to the source code to study it if you want, freedom to modify if you want, and distribute copies with or without your modifications if you want.

Good luck, I think you should re-evaluate your approach.

Pardon me? My wiritng has not been the best lately. I have very limited time to pursue these upgrades and similar projects, so attention to politeness may have slipped.

First, no criticism of AlephObjects or their products was ever intended. The Taz3 is a fairly early model, my machine is in refurbished condition, and some difficulties are entirely reasonable. Especially considering I have been using a Budaschnozzle with a 0.25mm nozzle.

The only ‘advice’ I think I offered here was with regard to reinforcing the frame sides with metal, rather than relying on plastic brackets. Not only because it would resolve one of the most significant sources of flex in the Taz5 design, but also because it is easy to apply to existing machines.

Second, please check the readme file on github. CAD models using non-free software will be replaced with FreeCAD documents. In fact, FreeCAD is proving easier to use for modeling the actuator components, so this transition is likely to take place sooner rather than later.

Further, I think I have been clear that I am trying to refurbish a machine, then push it far beyond any normal use case. For normal 3D printer use, LulzBot designs represent one of, if not the most, effective and affordable platforms available.

mirage: don’t mind the trolls. 800mm frame 4 lyfe.


Thanks, appreciate the encouragement. Your large build volume machine is also very impressive, and I am eager to see first-layer pictures covering the entire surface. :sunglasses:

u like?

Wow, absolutely incredible. Good job!

OpenBuilds V-Slot 500mm linear actuator, modeled as a FreeCAD assembly (requires Assembly2 workbench). FreeCAD is quickly proving an excellent, usable alternative to non-free CAD tools, although some steps can be relatively awkward.

Pardon the double posting, let me know if that causes problems.

Project completely rebuilt in FreeCAD, with one Z-Axis actuator attached. Render attached, source files at github.

This is starting to go together nicely.

Z-Axis and frame final design. All-metal mounting hardware consists of only the metal gantry plates, and a six L-Brackets.

@piercet Please take a good look at the design files.

CNC platform completed. Rather simple, but that is probably best. As has been said of the Hexagon hotend, all metal, all good! :smiley:

Please excuse the multiple posting…

EDIT: Threaded rod actuators have been calculated to provide up to 2000mm/sec speed. Belt drive will also be supported.

Design looks like it should work fine so far. For the Z axis bearings, since there is less force on that axis usually you might consider making the Openrail the structural member as well unless you feel you need the extra support back there.
Fast start ballscrews can be used on 3d printer X and Y axis successfully, You can see an example of that with the ill fated Aluminautis printer (which was a basically good design with bad manufacturing choices and a couple of questionable design choices that ultimately doomed them). One of the issues you may run into with an X and Y leadscrew is X and Y axis wobble. Same cause and potential countermeasures as the Z axis. It will tend to show up as excessive layer ripples. I’d probably stick to belts (potentially larger ones) or even motorcycle chain and sprocket before I would want to go with leadscrews for that rapid of a motion segment. You may want another plate at either end of the Y axis for finer bed alignment as well. The C rails you have there should be plenty stiff to avoid deflection, but that would be the area I would expect to see it too. Would also potentially make rail alignment easier. Looks neat so far!

Thanks for the tips!

Will definitely add a few more plates to the Y-axis, good idea. Also, thanks for pointing out the Aluminatus, just from photos it is clear they had a great concept and made some avoidable mistakes.

Interchangeability is one reason for the current Z-axis mounting scheme. Almost any actuator can be (dis)mounted using the side brackets. Similarly, the X-axis V-slot can be used as a mounting bracket for other hardware.

Since this will be used for CNC milling and photolithography exposure, high-precision under high-pressure at relatively low-speed is a top concern. OpenBuilds ACME leadscrews are quite convenient. With anti-backlash nut blocks, I would expect much finer positioning than belt drives or especially chain drive.

Wobble concerns me though. In the past, I have addressed this by restraining the lead screw at both ends, and minimizing linear bearing degrees of freedom. Further, PFTE greese has done an amazing job minimizing friction/wear. Any reason to expect the V-slot mechanism will not control this?

Finally, an updated render image is attached. Some Y-axis components aren’t moddled, but the basic premise should be obvious. The surface is MDF, and will be milled into alignment.

2000mm / s is very ambitious for threaded rod. If you are going to go that route, you will want to add an anti backlash nut. That’s basically 2 nuts per rod with a spring between them. That keeps the nuts engaged with the top and bottom of the thread. For the rest of the potential wobble, a drive nut with motion disconnect in the directions you don’t want travel is good insurance.

Ooh! No way am I actually going to drive any CNC machine at 2000mm/s, or even close. That was mere calculation to determine maximum step rate wouldn’t cripple the machine. :laughing:

More realistically, 45-180mm/s will be desirable during 3D printing, with other tasks like CNC milling performed at <<10mm/s.

Anti-backlash nuts are definitely in the design. OpenBuilds offers an excellent component for this - http://openbuildspartstore.com/anti-backlash-nut-block-for-8mm-metric-acme-lead-screw/ . Once the machine is operational, perhaps I will 3D print an appropriate spring mount to allow a bit of wobble.

BOM uploaded to Git repository. Total cost for the mechanical platform is ~$1.5k. However, that would drop to $1k for a similar Taz machine in bulk production, and even less if CNC mill capability is dropped.