Help me pick the right electronics hardware for a filament printer

Printable View

12-15-2017, 12:50 AM
Don McLean

Help me pick the right electronics hardware for a filament printer

Hi everyone. Everyone's favorite person is back (that's me) with some good news. I've decided to join the world of FDM printing.

Yes, I'm still planning on a DLP printer build, but I've realized that if set up correctly, the quality of a filament printer is good enough, and in particular, some of the filament types can do things that you can't do on a DLP resin printer.

The ones that are commercially available just can't do what I want to do. The build areas are too small. Some have very tall Z axis, but this isn't a priority for me. It's not the build volume that is important. I simply want more movement in either X or Y so I can orient a larger part so there is less risk of delaminating than you would get by printing a fairly narrow tall part with a tall Z axis. I've realized that for the cost of a new printer, I can build one with much better linear components.

I have some left over linear components from a CNC build. I have been working on a design for my own 3d printer. The only question is whether I will buy a small printer to 3D print some of the parts for it, or make it all out of metal. Belt drives for the X and Y and two screws driving a fairly large Z plate is the current plan.

I will need help on the electronics side. For example, I understand that the arduino mega and ramps 1.4 can only drive up to 6 motors, is that correct? Also people have been saying that this combination limits the speed of their printers because it can't keep up? Correct me if I'm wrong, I have heard that.

Here's what I want to do........

Dual motors driving the Y axis, I also want to have two limit switches on one side for a homing sequence on startup to make sure the gantry is always square. I want two motors because I prefer to keep the design simple, I'd like to use Nema 17's all around, and because my gantry will weigh a bit more than the average 3D printer, so the extra torque from two motors is a good thing. I realize there are other design options out there, but right now I'd like to pursue this one. This may change if I can't find the electronics solution I'm looking for, but let's focus on the electronics I seek for now please.

One motor driving the X axis.

Two motors driving the Z axis, I also want to have two limit switches on one side for a homing sequence on startup so I keep my build plate level.

Two extruders (actually have been thinking of perhaps even 3). I've been looking at E3D hotends and bowden extruders. I'd like to have the capability to go up to 300 deg C as a maximum.

So that's 5 motors driving the motion, and 2 to 3 for the extruders, with the ability for two axis to have homing to two independent switches each.

I'm also planning on a Z axis proximity sensor / switch and a heated bed,

Also a couple other prox sensors / switches in case of crashes (not related to homing, on the other side of movement to homing), but those can be wired together in series so they only really count as one switch as far as the electronics is concerned.

What kind of electronics hardware do I need to do this?

Is there a different shield available for the arduino mega that can accomodate more motor drivers?

I'm not really sure what options are good ones. Can anyone offer some advice, or a review on something they have tried? I'm good with the mechanical side, but the electronics, not so much.

Thank You in advance for any info!
12-15-2017, 06:21 AM
Trakyan

BEWARE: long post, the first half answers questions from your post, the second is my electronics recommendations.

I'd like to start by saying that a short part with lots of XY surface area is more likely to de-laminate than a tall part with a small XY cross section. Larger XY cross sections mean that each layer contracts more as it cools, this means larger forces between the layers and leads to de-lamination.

As for the arduino/RAMPS questions. I believe it has 5 stepper sockets on the board itself, one for each axis (but you can piggy back two motors off one driver, or assign two stepper drivers to one axis, or three if you really want to...?) and then up to two extruders. There are expansion boards you can plug into it for more stepper drivers. For your dual y setup with two limit switches, you'd need a dedicated driver for each Y axis motor unless you're ok with ramming the gantry against the y axis limit until both switches trigger. Nothing really wrong with this, the motors will just skip steps and as long as your printer is reasonably solid it wont hurt anything. It just makes a nasty noise that doesn't inspire confidence.
Side note, the two Y limit switches will do nothing to keep your build plate level, it could help keep the gantry square if you set them up so they trigger when it's square.

8 bit electronics (ramps/mega) will have no problems driving a cartesian machine at speeds well beyond what would shake it to pieces. It can have some trouble with nonlinear kinematics but even then it can manage unless you want relatively high print speeds.

The RAMPS/mega combo can use bed sensors, and it can run a heated bed though I suspect not one as big as you're looking to use. Could you please let us know how big of a print bed you plan to use?

By other sensors do you mean X/Y/Z max homing switches? RAMPS supports both a min and max endstop for each axis I believe. Even if it doesn't, it's not hard to add.

With all that out of the way, on to electronics suggestions.

Starting with the stepper drivers. You need 4 drivers for your movement axis, and are wanting up to three extruders. This means between 5 and 7 stepper drivers required. RAMPS can handle the single extruder case, but would need an expansion board for the extra extruders. I recommend against the ramps, spend a few extra bucks and get an all in one board with integrated drivers, it costs no more than 5 dollars extra but saves you a lot of wiring hassle and helps with stepper driver performance (important on a big machine where you'll be pushing the motors). Another reason I'd recommend against RAMPS for you specifically is that it's limited to 12 volts, this hinders the performance of the stepper motors which again is important for bigger machines. Your best 8 bit option (read: cheap) would be something like the RUMBA (no integrated drivers, so slightly poorer performance, but has 6 drivers which means you can use two extruders and double y axis) or a mks base or similar (integrated drivers will perform better, but this only has 5 drivers so it can only run one extruder with your setup). You can use stepper expanders with all of these boards to use an extra stepper or two.

As for the rest of it, min/max endstops are supported on just about every board, so are bed sensors. Most boards have pretty low powered (or even underpowered) mosfets for the heated bed, and they're usually gears for the standard 200x200 heated beds. To use a larger heat bed you'd probably want to use an external mosfet module. Again, I'd try and use 24 volts if I were you (i.e. not ramps), it boosts the performance of the stepper motors and heated bed/hotend. You could get your machine running with two extruders (or even three with an expander) using a RAMPS but you'd be making some pretty big compromises, so I highly don't recommend it.

Sorry for the long post, wanted to be thorough, let me know if you want a TL;DR
12-15-2017, 06:23 AM
curious aardvark

how about having a fixed bed and moving the z axis.
That way the size of the bed is not a factor, you can have a larger bed with thicker aluminium, and not worry about moving the weight around. The z axis setup will be much lighter. Plus your bed should not go out of calibration anywhere near as quickly.

So while you're not buildinga delta - with a fixed bed you get many of the benefits and none of the drawbacks.
12-19-2017, 12:38 AM
Don McLean

Thank You everyone for your replies so far! I really appreciate the feedback. Sorry for the delay in writing back, I've been doing alot of thinking about this, and I still haven't made my decisions.

Quote:

Originally Posted by Trakyan

I'd like to start by saying that a short part with lots of XY surface area is more likely to de-laminate than a tall part with a small XY cross section. Larger XY cross sections mean that each layer contracts more as it cools, this means larger forces between the layers and leads to de-lamination.

There has to be a way to mitigate this with the use of rafting and support material?

If you look at this video, in particular 2:51, it explains the "wood grain" strength of 3d printing which is why I can't simply adapt an existing printer to give it a really long Z travel, which would be very simple to do from a mechanical standpoint, although if I did this I would replace the Z axis entirely with one that has square rail supported linear bearings and a proper ballscrew to accommodate the extended length.

https://www.youtube.com/watch?v=JGhgaypou6E

Quote:

Originally Posted by Trakyan

By other sensors do you mean X/Y/Z max homing switches? RAMPS supports both a min and max endstop for each axis I believe. Even if it doesn't, it's not hard to add.

No, I mean the crash stop sensors. For example, with two motors on the X axis, you'd have a total of 3 switches, two for the homing to ensure that the gantry is square each time you start up (they also act as crash sensors on that end) and one at the other end of travel that acts only as a crash sensor. So that's three for one axis, however the one that is by it's self can be wired together with other "crash only" sensors so it only takes up one spot on the board for all of them. The two homing sensors would each need their own board slot. Do you typically home to both ends of an axis? I've never seen that before.

Quote:

Originally Posted by Trakyan

To use a larger heat bed you'd probably want to use an external mosfet module. Again, I'd try and use 24 volts if I were you (i.e. not ramps), it boosts the performance of the stepper motors and heated bed/hotend. You could get your machine running with two extruders (or even three with an expander) using a RAMPS but you'd be making some pretty big compromises, so I highly don't recommend it.

That's something to consider for sure. Yes, I'd want 24V minimum. Higher voltage means higher torque at speed for stepper motors. Higher current means higher torque at low speed for stepper motors, but doesn't necessarily have any effect on higher speed torque. Depending on how the system is geared, more voltage is the way to get better performance.

I'm thinking of a 12" x 36" bed with about 10" x 30" as the printable area. Alternatively I could still decide on a 12" x 24" bed with a 10" x 22" printable area. What would you recommend as a heater for those sizes?

Quote:

Originally Posted by curious aardvark

how about having a fixed bed and moving the z axis.
That way the size of the bed is not a factor, you can have a larger bed with thicker aluminium, and not worry about moving the weight around. The z axis setup will be much lighter. Plus your bed should not go out of calibration anywhere near as quickly.

So while you're not buildinga delta - with a fixed bed you get many of the benefits and none of the drawbacks.

I have thought about this. There are actually lots of drawbacks from having the Z axis attached to the gantry. For a CNC router, the more Z axis travel you have the stiffer and heavier you have to make everything because when it accelerates back and forth, the gantry will flex due to the moment from the weight hanging below it. So for a 4" Z height, yes it could make alot of sense, but for an 11" Z height it doesn't, unless you want to add more rails / bearing blocks and make the gantry assembly 10x heavier than it would otherwise be. Have you seen any 3D printer designs that use this methodology and work really well? If you have, I assume they look more like CNC routers with big gantries and linear components? The Delta design is it's own kind of beast, however if I could get over the wood grain strength issue previously mentioned, a tall delta printer might make alot of sense.

I am currently building a steel CNC machine, haven't finished yet, but have made others in the past, also very large, weighing in the 1000's of lbs range. I was thinking about adding a 3d print head to my CNC that I'm building now but in reality I don't want to run the bohemeth for 48 hours straight doing a print job, it's somthing that shouldn't be left unattended. Also I'd have to redo all the electronics to accomodate the extra motors for 3D printing as my current board couldn't handle that, and forget about trying to enclose it to keep the heat in, plus the debris etc from the CNC part would just make it messy.

And after your suggestions about the type of head on the bnc3d sigma, I am seriously considering this. The only problem is that it takes up way more space. This would require two moving Z axis heads (if I had a fixed table), and I think that would be a nightmare.

I've also been having a much more serious look at the H-Bot and Core XY designs. Have you ever seen a printer with independantly moving heads that uses H-Bot or Core XY? It is totally possible from a mechanical and design standpoint, but I don't know if the software would get confused with that in addition to the H-Bot or Core XY architecture?

Thanks again for your replies guys! I still don't know what I will do here, so continued advice is appreciated.
12-19-2017, 12:47 AM
Don McLean

One more question, I've been looking at gear reduction for the stepper motors / belt system.

Have you ever come across an 80T + 20T combined idler pulley? Or a 60T + 20T combined idler pulley?

Running a 20T GT2 pulley from the stepper to a 60T idler that is connected to a 20T gear turning the drive belt would give me a gear reduction of 3:1, which if I run the steppers at a good voltage could give me better native resolution and better / smoother performance.

Alternatively I've been looking at 16T gears, but the idlers only have a shaft of 3mm, and that seems too small for my liking.

It looks like most people are simply running 20T gears without any reduction for their pulley systems?
12-19-2017, 05:41 AM
curious aardvark

Quote:

I assume they look more like CNC routers with big gantries and linear components?

Thinking about it - yes they do :-)
So what's wrong with building one of those ?
You can simply use 20x20 or 30x30 aluminium extrusion for the gantry. It won't bend :-)
Has the advantage of being cheap, very easy to attach things to and slide things along and super rigid. Much better than mucking about with rods and screws and the like.
You could build a horizontal carriage with the hotend below the extrusion and the motor above.

The 'wood grain' analogy is a good one and angus is probably the only 3d printing youtuber I have any time for.

But you have to bear in mind that it's also material specific. Flexibles - for example - have almost perfect layer bonding no matter what orientation you print them at.
A stiff flexible pla is brilliant for parts that need to be virtually indestructible, have flex and resist impact but also need structural stiffness.
I've yet to see any kind of layer bonding issue with any flexible.

Pet-g is an interesting one. sometimes it seems to have great layer bonding and sometimes not - I'm starting to lean towards moisture absorbance being the issue with my cheaper pet-g.
The sun-lu stuff - at the moment- I have is definitely nowhere near as tough as the colorfab xt.

Thinking it needs a session in the dehydrator.

But anyway, the basic gist is that layer bonding is as much material driven as print orientation driven. So don't necessarily let that be a defining factor in your choice.

And i suspect the reason people don't muck about with gear ratios that often, is simply that most of us have no bloody clue how to rewrite firmware to get custom gearing actually working :-)
12-19-2017, 07:45 AM
Trakyan

Quote:

Originally Posted by Don McLean

Thank You everyone for your replies so far! I really appreciate the feedback. Sorry for the delay in writing back, I've been doing alot of thinking about this, and I still haven't made my decisions.

There has to be a way to mitigate this with the use of rafting and support material?

Rafting is to help with bed adhesion and support material is for overhangs. Delamination is basically the same as prints coming unstuck from the bed, except instead of coming unstuck from the bed, the layers unstick from each other*.

The way to fix this is to make sure the prints cool slowly and uniformly, one way to do this is with a temperature controlled print chamber. Not always necessary and this is related to a print's XY area, not Z area. Prints with larger XY areas have larger layers, larger layers will contract more as they cool.

Quote:

Originally Posted by Don McLean

If you look at this video, in particular 2:51, it explains the "wood grain" strength of 3d printing which is why I can't simply adapt an existing printer to give it a really long Z travel, which would be very simple to do from a mechanical standpoint, although if I did this I would replace the Z axis entirely with one that has square rail supported linear bearings and a proper ballscrew to accommodate the extended length.

The 'wood grain' analogy is about the strength of printed parts, and how they tend to shear along the layer lines, nothing much to do with warping or delamination during printing. You can absolutely give a printer more z height by doing that if you wish, just look at what ultimaker does with their 'extended' edition printers.

Quote:

Originally Posted by Don McLean

No, I mean the crash stop sensors. For example, with two motors on the X axis, you'd have a total of 3 switches, two for the homing to ensure that the gantry is square each time you start up (they also act as crash sensors on that end) and one at the other end of travel that acts only as a crash sensor. So that's three for one axis, however the one that is by it's self can be wired together with other "crash only" sensors so it only takes up one spot on the board for all of them. The two homing sensors would each need their own board slot. Do you typically home to both ends of an axis? I've never seen that before.

No, you don't typically home both ends of an axis, it's used as what you're calling a crash stop. I was just saying you could use the second endstop plug as the second homing switch instead of a max endstop switch. You could then wire another endstop in parallel to one of them on the other side as a crash stop, since you shouldn't ever have both the min and max trigger at the same time it shouldn't be an issue.

Quote:

Originally Posted by Don McLean

That's something to consider for sure. Yes, I'd want 24V minimum. Higher voltage means higher torque at speed for stepper motors. Higher current means higher torque at low speed for stepper motors, but doesn't necessarily have any effect on higher speed torque. Depending on how the system is geared, more voltage is the way to get better performance.

I'm thinking of a 12" x 36" bed with about 10" x 30" as the printable area. Alternatively I could still decide on a 12" x 24" bed with a 10" x 22" printable area. What would you recommend as a heater for those sizes?

The higher voltage has another perk, it lets you drive the motors faster since the higher voltage can overcome the inductance easier. You tend to lose torque with steppers as you go faster since at higher switching frequencies (speeds for the motor) the inductance of the coils has greater effect, and 24V helps overcome that. As for the heated bed, it would need to be custom, I can't think of any off the shelf that fit the bill. For that size I'd go for a main silicon heater, and chinese manufacturers will probably be willing to make a custom sized one for you at a reasonable (but still expensive) cost. Mains power because for that size, running it off 24 volts would require high currents which can get dangerous real quick.

Quote:

Originally Posted by Don McLean

And after your suggestions about the type of head on the bnc3d sigma, I am seriously considering this. The only problem is that it takes up way more space. This would require two moving Z axis heads (if I had a fixed table), and I think that would be a nightmare.

Agreed. If you have two heads, go for a moving Z bed. CNC routers/mills use that type of gantry because it makes it easier to get the z axis orthogonal to the bed (so your cuts end up with vertical, not slanted, walls). With 3d printers having it orthogonal doesn't matter as much, and a tall Z axis is more desirable so a moving bed is more common.

Quote:

Originally Posted by Don McLean

I've also been having a much more serious look at the H-Bot and Core XY designs. Have you ever seen a printer with independantly moving heads that uses H-Bot or Core XY? It is totally possible from a mechanical and design standpoint, but I don't know if the software would get confused with that in addition to the H-Bot or Core XY architecture?

It's absolutely possible, I think dc42 has implemented it for reprap firmware, and I don't think it should be too hard to implement in other firmwares if it isn't there already. Keep in mind this type of set up would need double the amount of motors for XY, which adds to the already high number of stepper motors you're wanting to control.

Quote:

Originally Posted by Don McLean

One more question, I've been looking at gear reduction for the stepper motors / belt system.

Have you ever come across an 80T + 20T combined idler pulley? Or a 60T + 20T combined idler pulley?

Running a 20T GT2 pulley from the stepper to a 60T idler that is connected to a 20T gear turning the drive belt would give me a gear reduction of 3:1, which if I run the steppers at a good voltage could give me better native resolution and better / smoother performance.

Alternatively I've been looking at 16T gears, but the idlers only have a shaft of 3mm, and that seems too small for my liking.

It looks like most people are simply running 20T gears without any reduction for their pulley systems?

If you want 80T and 20T, just get them separate and mount both on the same shaft.
https://www.aliexpress.com/item/LUPU...StoreLevelAB=5
Just type '16T pulley' into aliexpress or google, one of the first results.
Anyways, performance should be 'smooth' enough with 16/32 microstepping. I doubt you'll see much improvement regarding resolution because of other sources of error like extrusion width, mechanical vibrations and slop, belt elasticity and so on. To use them, all you need to do is multiply your steps/mm by your gear ratio, on some printers this can be done through the LCD.

*The more scientific explanation behind this is when a layer cools, it contracts. If the print does not cool uniformly, there is differential cooling and contraction. When one side of an object contracts more than another, it causes it to bend and warp (think bimetalic strips) and if the top layers cool quicker and try to curl up, while the bottom layers stay stuck on the bed, the top layers will delaminate from the bottom layers and result in a crack in the print.

As for 3d printing youtubers worth following, thomas is a good one to look at when it comes to the technical side of printers. He has lots of guides on firmware, electronics, mechanics and more and his technical knowledge is spot on and has some backing as he's sourced and built a couple of printers himself of both the delta and cartesian variety. The things Angus says aren't always entirely accurate.

EDIT: wow that is long...
12-20-2017, 03:24 PM
curious aardvark

Quote:

The things Angus says aren't always entirely accurate.

Pretty sure that goes for all of us :-)

But he's more 'user friendly' than most of them :-)
And I don't spend much time watching youtube videos. lol
12-20-2017, 10:34 PM
Trakyan

He is more user friendly but that also seems to make him less technically orientated. He often uses jargon without understanding it and I find he goes back on what he says a lot.
12-21-2017, 09:04 PM
Don McLean

Quote:

Originally Posted by curious aardvark

But you have to bear in mind that it's also material specific. Flexibles - for example - have almost perfect layer bonding no matter what orientation you print them at.

Quote:

Originally Posted by curious aardvark

But anyway, the basic gist is that layer bonding is as much material driven as print orientation driven. So don't necessarily let that be a defining factor in your choice.

Nylon 12, and Nylon 12 CF filaments (perhaps also some ABS CF filaments?), I realize these are expensive filaments, but I want to make functional parts for a specific application, and some suppliers of these filaments are less expensive than others. Sure I want to do fun projects also, like a starwars inspired helmet, props for friends and family who would feel at home at comicon, but those are not my main motivation.

So how important is orientation for Nylon12 and Nylon12 Carbon Fiber filaments re: woodgrain strength? And how much of a maximum Z height can I push if orientation isn't important?

Quote:

Originally Posted by curious aardvark

And i suspect the reason people don't muck about with gear ratios that often, is simply that most of us have no bloody clue how to rewrite firmware to get custom gearing actually working :-)

I hope you're joking. Are you talking about open source firmware or machine specific closed source firmware? In mach 3 (for CNC routers, mills) for example, this is super easy to set up.

Typically, IMO, for larger CNC machines using Nema 34 steppers you want a gear ratio of 2 to 4 for a rack and pinion system with a pitch circle diameter of 1" on the pinion (alot more for servos) and this is very similar to a belt drive. Not using any gear reduction doesn't work very well. Also using high inductance motors doesn't work well. Also you typically run at 48V as a minimum, with 60V or higher not being uncommon. This gives better speed performance to the stepper motor. 10mm travel per rev to 38mm travel per rev of the motor on X and Y, depending on a number of factors, this is the comfort area for CNC stepper motion, IMO.

So if you look at a 20T GT2 pinion, that's about what? 1.5" to 1.6" per rev? That's on the edge, but perhaps that's fine. I really need to have a better look at the Nema 17 torque vs speed graphs to give a definitive opinion. Also 3d printers tend to use more microstepping, but microstepping can't always be relied on for accuracy in and of it's self.

Originally I was looking at using a single motor to drive the gantry with a connecting rod to two belts. I was thinking a 12mm rod, so that ruled out 20T and left me with 40T driving the two, and about 3.15" per rev, which is too much IMO, so I was looking at a belt reduction from the connecting rod to the motor, and I still haven't ruled out some belt reduction.

I was watching a vid on Youtube recently of a gentleman refitting a stratasys 3d printer, they definitely had some belt driven gear reduction with Nema 23 motors inside that thing.

Quote:

Originally Posted by Trakyan

If you want 80T and 20T, just get them separate and mount both on the same shaft.

So this method is waaaay more of a PITA than using an idler that already has a built it bearing that you can bolt on somewhere, especially if it is mounted vertically.

I just bought a 22mm carbide tooth hole saw to help with making my own custom mounts for Nema 17's,

Quote:

Originally Posted by Trakyan

*The more scientific explanation behind this is when a layer cools, it contracts. If the print does not cool uniformly, there is differential cooling and contraction. When one side of an object contracts more than another, it causes it to bend and warp (think bimetalic strips) and if the top layers cool quicker and try to curl up, while the bottom layers stay stuck on the bed, the top layers will delaminate from the bottom layers and result in a crack in the print.

Perhaps I should really think about some doors and a lid to make an enclosed space.

Quote:

Originally Posted by Trakyan

As for 3d printing youtubers worth following, thomas is a good one to look at when it comes to the technical side of printers. He has lots of guides on firmware, electronics, mechanics and more and his technical knowledge is spot on and has some backing as he's sourced and built a couple of printers himself of both the delta and cartesian variety. The things Angus says aren't always entirely accurate.

I like watching Angus' videos because he seems like a very honest and kind person. Obviously he is more oriented towards using and reviewing commercially available printers than building them, which is fine.

Are you talking about Thomas Sanladerer? I've watched several of his videos. I don't think he knows what he's talking about when it comes to CNC cutting, but in general I like him also. I haven't seen anything to make me think he is more knowledgeable than Angus, he seems to mostly review printers, talk about their features, etc., the same as everyone else. Where are the links for the printers he has built?

Don't worry about the long reply, I appreciate the effort you put into it. :)

On one hand I know a few things from other CNC experience and what I have built / used in the past, on another more relevant hand, I am a total noob in this space. So any help I can get is appreciated.
12-22-2017, 05:02 AM
Trakyan

Quote:

Originally Posted by Don McLean

So how important is orientation for Nylon12 and Nylon12 Carbon Fiber filaments re: woodgrain strength? And how much of a maximum Z height can I push if orientation isn't important?

Your layer bonding does /not/ limit your maximum Z height in any way, shape or form. What it limits is the amount of force the layers can take before they shear or pull apart. This is to do with how the part is loaded when in use, nothing to do with how it prints.

Quote:

Originally Posted by Don McLean

So if you look at a 20T GT2 pinion, that's about what? 1.5" to 1.6" per rev? That's on the edge, but perhaps that's fine. I really need to have a better look at the Nema 17 torque vs speed graphs to give a definitive opinion. Also 3d printers tend to use more microstepping, but microstepping can't always be relied on for accuracy in and of it's self.

Most printers out there, including higher and high end commercial ones use 20 or 16T pulleys with no gear reduction and do fine in terms of torque and accuracy. You don't need anywhere near as much torque for a 3d printer as you do for a mill, the only forces on it are acceleration/deceleration no cutting forces. And you'll never be running your stepper motors on a 3D printer to a speed where you'll lose noticeable amounts of torque, unless you've geared them down significantly. You are entirely correct that anything beyond half stepping isn't really reliable accuracy wise, and printers don't rely on this for accuracy. With a 0.4mm nozzle, the accuracy you get with an ungeared nema 17 running a 20/16T pulley is good enough, and far from your biggest source of error.

Quote:

Originally Posted by Don McLean

Originally I was looking at using a single motor to drive the gantry with a connecting rod to two belts. I was thinking a 12mm rod, so that ruled out 20T and left me with 40T driving the two, and about 3.15" per rev, which is too much IMO, so I was looking at a belt reduction from the connecting rod to the motor, and I still haven't ruled out some belt reduction.

I was watching a vid on Youtube recently of a gentleman refitting a stratasys 3d printer, they definitely had some belt driven gear reduction with Nema 23 motors inside that thing.

Your first idea adds a lot of inertia and mass for the motor to drive (and halves or worse your resolution), which brings me to the part about stratasys systems being geared. If you play spot the difference with a desktop printer and a stratasys (I assume you mean an industrial model?), the stratasys systems are much more heavily built and require the extra torque from some gear reduction to lug around those overgrown axis. Even a 500x500mm print bed can use ungeared nema 17s as long as you build the gantry light enough.

Quote:

Originally Posted by Don McLean

Are you talking about Thomas Sanladerer? I've watched several of his videos. I don't think he knows what he's talking about when it comes to CNC cutting, but in general I like him also. I haven't seen anything to make me think he is more knowledgeable than Angus, he seems to mostly review printers, talk about their features, etc., the same as everyone else. Where are the links for the printers he has built?

I agree that Tom isn't too knowledgeable about CNC, and from what I understand he only has a converted mendelmax to cnc (which was more of an experiment in 'can it be done' not 'this is perfect who needs a proper cnc mill') and a mdf framed cnc he was recently sent to review, which didn't seem to go well. He does however have a manual lathe and mill which he's used in a few videos, and most engineers are at least somewhat trained with those.

He has a bunch of older guides going over every aspect of a printer from extruders to hotends, control boards and motion. Those are very useful but some are a bit outdated at this point.

Some useful guides to look at
https://www.youtube.com/watch?v=pRX_...zECor3EbqF8euw
(mostly the later ones down the playlist)
https://www.youtube.com/watch?v=E5Fq...3G9QSpzsRQCVGG
(series on building a printer, but maybe slightly less relevant for your situation)

Two of his printers he's built
https://www.youtube.com/watch?v=gZU-N3qUbRE
(his own design I think)
https://www.youtube.com/watch?v=M8BlJ_8nD68
(his build of a mendel90)
He's referenced an older mendel he's built in a couple of videos but hasn't made a full video on it.