Cannibalize one hobby printer to build a "set and forget" machine.

I've built a number of 3d printers over the last 5 years and currently have a P3Steel and generic corexy in service. Both of these printers are "hobby-grade" meaning one is usually "in the shop" for fixes/upgrades. I now want to build a high-quality workhorse that is much more set-and-forget then my previous builds. The best 3d printer I've seen to date is the Taz5... but given I have enough quality parts laying around I'm thinking I should build my own machine.

Key Upgrades:
- 300x heated bed with dedicated PS, SSD, pei or builtak?
- Duet Wifi.. I run a duet.6 on my P3 and its been excellent
- I'm game for pretty much any extruder setup but currently have a Titan and Bulldog in serve to steal.
- Direct drive extrustion
- Clean wiring.... somehow despite how many enclosures I print.. I never manage to fully nail this down.

I know many of you have followed similar upgrade paths, I'm curious what you've built? Voron, D-Cube seem to be solid choices... The Taz uses a super-high-quality i3 design and I know there are people building these. There are a couple Medal Max XL machines that look enticing but I'm curious what the least maintenance options are in 2018? It looks like a box of ALU extrusion is in the future any gotchas with that?

Best,

Andy

When looking at low maintenance, you'll probably be looking at individual part choices like bushings vs linear bearings and so on rather than motion systems. What I mean is which components, rather than motion systems, are low maintenance. The motion systems themselves aren't usually inherently high maintenance. Deltas need a bit more calibration and care (at least to begin with) I guess and H bot will likely rack and skew your frame, but that's about it.

The only other piece of advice I can give is to avoid tape on build surfaces like buildtak if you want low maintenance. They may be good but from what I hear they need to be replaced reasonably often because they rip and tear when taking parts off. Haven't heard too many complaints about PEI other than it can be a pain to get off the bed and replace, but it seems durable.

For wiring, my personal favorite is drag chains, even printed ones. Those plastic spiral wraps aren't as good in my opinion and tend to get stuck places and create excessively large loops in some parts of the printer's motion. I have a cube style printer with a spiral wrap leading to the extruder through a hole in the rear cover of the printer. When the extruder moves away from that hole, it pulls the wire bundle through the hole just fine. The issue comes when it moves back towards it the wires don't get pushed neatly back through the hole, they get stuck trying to back out then bunch up and kink. Drag chains are cheap enough, and even cheaper to print. While printed ones aren't as smooth as injection molded ones, they'll break in and work fine after a few prints.

As far as motion systems, I'd avoid mendel/i3 style designs, especially for larger print volumes. They aren't very space efficient and throwing the print around is just a bad idea that gets worse as your prints get taller or faster.

Sorry, triple post.

Edited 1 time(s). Last edit at 02/20/2018 12:51AM by Trakyan.

Sorry, accidental triple post.

Edited 1 time(s). Last edit at 02/20/2018 12:50AM by Trakyan.

I have Cartesian, Delta and SCARA printers. My go-to machine is the delta, because it is mechanically simple, virtually maintenance-free, and prints quickly. Its possible disadvantages are that it has a long Bowden tube, so print quality might be better with a direct-drive extruder; and it is single extrusion, so I can't do multi-material prints on it. But I get good prints from it, and I haven't wanted to do a multi-colour print for over a year.

However, deltas are less forgiving of poor mechanics than Cartesian printers, so I caution against building one cheaply. Mine cost well over £1000 in parts.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

One important part in set_and_forget printers is the Z-probe. Make sure you get a good one.
I went from Microswitch/Allen key probes to 12mm proximity sensors and ended with 18mm proximity sensors ( for their bigger probe distance ) or a Smart Effector on the XL Delta.
Then I'd always prefer linear rails over end supported rods or V-rollers, where possible.

Set and forget means different things to different people. To me it means using construction that includes a flat bed plate, a mechanism that allows it to be leveled, and is stable enough that it doesn't have be releveled (or autoleveled) after initial setup. My latest printer uses PEI on cast aluminum tooling plate, with a 750W line powered heater, supported on a kinematic mount, lifted by belts. The frame is made using 40 mm square t-slot and it uses linear guides in all axes. It's fully enclosed and heated to allow reliable ABS printing. I print from SD cards to eliminate USB connections to the printer, though you can do the same with networking. Details in the UMMD link in my sig, below.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

I havent built one, but a CoreXY machine like Tech2C on youtube made.

He also did several comparisons of print quality of steel, aluminum rod and carbon fiber rod on the X carriage.(reducing weight).

I also like that it is quite scalable, and can be enclosed pretty easily.

I made a SmartRap mini, but thats definitely NOT set and forget....

Thank you for the comments... good to hear the feedback on the linear guide system.

- I agree with the leveling sensor: dc42's IR sensor is a no brainer (again using it on my p3steel for years now).

The bed plate is one source of annoyance still. At this point I feel the bed should be non-adjustable and be fixed. Obviously don't see many machines with that design but the newer mesh leveling capabilities of Duet and other 32bit boards are making this more reliable over having a 100% perfect bed. Eliminating bed adjustment screws that can loosen over time or get bumped seems like a good idea. Has anyone seen a good design for a fixed mount bed? I assume you would build, level it once and locktite it with fixed bolts... At this point I think the traditional springs could be skipped.

I used steel GT2 belts for my z-axis and while it performed fine, It's just a bit messy for my taste. It seems to me eliminating "belts" wherever possible will also improve reliability and durability. Another issue I've experienced is quality issues with the bed/bed mount. Using aluminum for your bed plate sounds like a good option, how thick is it? Did you have it custom fabricated?

Also, I've always used a Aluminum mk* heat bed but I think for the larger bed I will want the silicon heating mat with the separate heater etc... I have seen the PEI surface work really well on the TAZ printers... BuildTak on my aluminum has been awesome too... 100% reliable for years now... Curious to hear any thoughts on that as well. Eliminating bed problems once and for all is high on my priority list.


So....

- Linear Rail motion system
- Bed with minimal to no adjustment once installed
- As few belts as possible

What's wrong with belts? The other option I know is lead screws and they come with their own issues too. ( eg. not maintenance free and risk of wobble; usually lower max. speed )

The bed of my XL Delta is fixed, but I'm to ashamed to show a pic . I placed some 5x10mm wood into the t-slot and then clamped the bed down. ( glue didn't work at 105°C ). I agree, its not necessary to have manual bed leveling screws.

Will be interesting to see how many cannibalized parts you'll actually use on the new machine.

You want the bed to be as flat as possible and levelable, even if you have autoleveling. And even if you have autoleveling, you should level the bed as well as possible before turning it on. If you do otherwise it will be impossible to print if the autoleveling fails for some reason, and it has been known to fail now and then.

If you're concerned about the leveling screws loosening, the answer is to not screw them into a thin piece of metal in which they will wobble, but drive them into heat resistant plastic blocks that will grip them tightly. I used teflon /torlon blocks in my last two printers and the screws never loosen or move from where they are set. In SoM, I had springs pushing the bed plate up against the leveling screw heads, but when I realized how stable it was, I replaced the springs with nuts. It hasn't needed any leveling adjustments in about a year. My most recent printer, UMMD, uses a kinematic mount in which the bed plate sits on the leveling screws heads and is held down by springs. That mount allows the bed plate to expand when heated without putting any lateral forces on the leveling screws or causing anything to flex. I'll be converting SoM to that type of mount in the near future. Details of that design here. It has been working without any adjustment since the last time I had it apart, maybe 6 months ago, in spite of being transported in my car laying on its back to the makerspace, two maker faires, and home, at least 6 trips. The kinematic mount is the definition of "just works".

I bought a piece of 13.5 x 15" 8mm MIC6 from a local supplier for $15 and cut and milled it at the makerspace. I cut it to match the size of the bed heater, so the heater goes right to the edges, and the machine can print right to the edges of the bed.

SoM has a precision ground ball screw in the Y axis and it is noisy. All the metal to metal contact is very good at transferring vibrations from the motor to the bed plate. SoM lives at the Milwaukee Makerspace and UMMD was built so I'd have a printer to use at home that was quiet enough that my wife wouldn't complain about it. Belts are much quieter than screws.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Quote
atheimer
- I agree with the leveling sensor: dc42's IR sensor is a no brainer (again using it on my p3steel for years now).

I'm glad it's working well for you! However, its consistency is somewhat dependent on the build surface. See the fitting instructions at [miscsolutions.wordpress.com] for the limitations, in particular when using PEI.

Quote
atheimer
The bed plate is one source of annoyance still. At this point I feel the bed should be non-adjustable and be fixed. Obviously don't see many machines with that design but the newer mesh leveling capabilities of Duet and other 32bit boards are making this more reliable over having a 100% perfect bed. Eliminating bed adjustment screws that can loosen over time or get bumped seems like a good idea. Has anyone seen a good design for a fixed mount bed? I assume you would build, level it once and locktite it with fixed bolts... At this point I think the traditional springs could be skipped.

Two comments on this:

- Delta printers have fixed beds, so no adjustment screws needed, assuming you get the bed perpendicular to the towers (or nearly so) by design;

- Mesh levelling on a Cartesian or CoreXY printer can be used for 3 purposes: (a) to compensate for a bed that isn't flat; (b) to compensate for sag in the gantry (i.e. height of the print head varying with X and/or Y position, due e.g. to support rails flexing under the weight); and (c) to compensate for a bed that is flat but not level with respect to the gantry. Your printer may be solid enough not to need (a) or (b) if you design it carefully. For (c) the traditional approach is bed level adjustment screws; however an alternative is to use 3 independently-driven leadscrews, and have the firmware level the bed using the Z probe. The Railcore printer (see [forum.seemecnc.com]) has this as one of the build options, and if you choose it then the bed levelling screws are omitted. The disadvantage is more expensive electronics, because you need to drive 6 stepper motors independently for a single-extrusion printer.

Quote
atheimer
I used steel GT2 belts for my z-axis and while it performed fine, It's just a bit messy for my taste. It seems to me eliminating "belts" wherever possible will also improve reliability and durability.

I don't have a problem using belts for the XY drive, but I'd hesitate to use them on the Z axis - although I've never tried to do so, unlike digital_dentist.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

The last time I looked, (most) delta printers use belts for the Z axis...

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Quote
the_digital_dentist
The last time I looked, (most) delta printers use belts for the Z axis...

Fair enough! But the weight supported by the belts is much lower - just the effector and hot end on most delta printers - so the issue of the bed falling when the printer is powered off (unless you use a worm gear or similar) doesn't arise; and I don't have to worry about belt stretch that might result from lifting a heavy bed (but maybe that's not a problem anyway).

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Stretch isn't really a problem. With two steel core belts in UMMD, I measured 42 um stretch per kg of print mass. If you completely cover the 300x300 mm bed with PLA (1.25g/cc) the stretch in any 250um layer will amount to about 1.2 um.

I used a $108 Rino 30:1 reducer in UMMD, but I've just designed and printed a 40:1 gear box (using steel gears) that will fit a NEMA-17 motor with 8mm output shaft. That will essentially duplicate the Rino function for about $30, making belt lifted Z axes more economical, and eliminating the need for Rube Goldberg type arrangements of counterweights, bungee cords, etc. The output will be 1600 usteps per mm, or 100 full steps/mm, putting full-step layers at 10 um intervals. I'm waiting for the gears to arrive and hoping they match the dimensions on the manufacturer's web page. Lunar New Year is a b**ch.




Edited 2 time(s). Last edit at 02/20/2018 07:49PM by the_digital_dentist.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Having the firmware level the bed is an interesting concept. I came across this yesterday looking at the Voron 2 vids (https://www.youtube.com/watch?v=wRdF0ZTIliY) which I think uses the DuetWifi.

The Voron 2 looks kinda cool especially with the auto-leveling gantry but 4 motors and 4 belts for the z system scare me a bit...

That said my confidence in using belts has increased thanks to this thread.

Edited 1 time(s). Last edit at 02/22/2018 08:25AM by atheimer.

Edit: the filters kept causing me issues and would let me post, all of a sudden two posts go through.

Edited 1 time(s). Last edit at 02/22/2018 04:33PM by Trakyan.

Personally I think four z axis motors is a bad idea, it's over constrained and will warp the bed, costs more and will be harder to get aligned. This is for all the same reasons 4 leveling screws is generally frowned upon. Three is plenty. One of the cubicon printers does something like this, could be worth looking into. I personally think one z motor to lift the bed, one fixed point on the bed and then two other servo controlled points on the bed to mechanically level. Adding 2 servos is easier on most boards than adding an extra two z motors.

Does a proper bed really cost so much more than a crappy bed plus all the extra electronics and motors that enable it to be used?
Does all that stuff result in a more reliable machine compared to one that has an actually flat and actually level bed?

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

TDD, I was just making a suggestion regarding a possible improvement to the automatic leveling system being mentioned. As far as its value, that's up for debate. For most people, the extra 30 seconds to manually level the bed or a $5 probe for software leveling is good enough. But some people don't want to 'tinker with' or deal with the mechanics of the printer, they just want it to run, like businesses, the education sector etc. They don't want to deal with manually leveling the bed. And software leveling with a probe is non ideal as it's just trying to compensate for a non flat or out of tram bed, so I can see why the mechanical auto level would be used.

It's all about the needs of the person or organization. I know a lot of people who most definitely do not want to manually level or adjust the printer, and people for whom software leveling may not be good enough.

Quote
the_digital_dentist
Does a proper bed really cost so much more than a crappy bed plus all the extra electronics and motors that enable it to be used?
Does all that stuff result in a more reliable machine compared to one that has an actually flat and actually level bed?

Most of the original 3d printer development was done by electrical engineers and software developers. When all you have is a hammer, everything looks like a nail.

My pet peeve is people trying to put encoders on everything.