LISA Simpson

Good suggestion to screw the platens together, and too identify the holes (a,b,c) to reassemble in the same orientation.

Quote
nicholas.seward
I actually take it a step farther.

draw a quarter circle.
Pick a point low on the quarter circle and swing an arc to intersect the arc you are on.

To verify you did everything right go to that intersection and make sure you can swing an arc directly through the other two points of the triangle.

Can you make a sketch?

Basically, put your compass on the top hole position (A), draw an arc (a) below it that covers >60 degrees. Pick a point ( B ) on that arc so that it draws an intersecting arc through itself (a), and this is the third point (C) of the equilateral triangle. To double check, put your compass point of C and it should draw an arc through A.

Edited 3 time(s). Last edit at 12/10/2013 10:27AM by dpharris.

Tks, good description

Quote
nicholas.seward
....... If you bump up to a 1" screw you can make it 8.5' tall. (For fun I priced this last one. It comes to less than $2500 for the whole printer. For 5 times the cost you can make a printer that has 75 times the volume.)

So to get to 75X the volume you are also scaling the X and Y in addition to the Z. Unless I'm very confused (again) about the build height of the standard design.

Looks like the piggy bank maxes out somewhere around 4' tall. That would be roughly $500 for the 3/4" Torqspline screws. At that diameter you have the option of a 1.5" thread in addition to the 1". More decisions...

@uncle_bob: The 75X was scaling in all dimensions. You may just want the height or the width. What is your target envelope. I would love to crunch out the recommended geometry. I would really like to see a megaLISA.

Edited 2 time(s). Last edit at 12/10/2013 12:45PM by nicholas.seward.

Poking around on the Roton site, the Hi-Lead screws have a few interesting things in their pricing:

3/4 x any (including 0.5) at $13-14 per foot
1x 0.25 at the same price.

That would get the price of a true mega back down to piggybank range. I'd probably need NEMA 23 steppers, but three of them should not be to crazy. I'd also (maybe) cut my speed in half. The nuts are similar in price. The pitch may not allow you to push the head to the top by hand. I would get a bit more pecision at the lower speed.

Edited 1 time(s). Last edit at 12/10/2013 06:37PM by uncle_bob.

@uncle_bob: You can get even cheaper.

Nuts
Flanges
Rods

So for a 6' version with 1" (1" lead) screws I come up with $771 for the screws, nuts, and flanges. Not so bad.

If you really want to go cheap you can try the threadless ball screws mentioned earlier in this post. A 3/4" rods will only cost you $175. 608 bearings are practically free.

NEMA 23s are recommended.

uncle_bob, Tks for pointing out the Hi-Lead screws, yea more to consider, or not!
That's a really good observation!

Hi-Lead
The Dia. 1.000, Lead: 1.000 (in./rev.), Price: $27.94 \ft., is a big cost savings compared to the
smaller diameter 3/4” inch, 1.000 lead Torqspline ($43.08\ft), and you get a much larger lead screw compared to the
7/16” inch at $20.08/ft.

I think this is a no brainier if you desire the larger diameter print envelope, require a more rigid setup,
and you're on a tight budget.

I think that the Hi-Lead Dia. 1.000, Lead: 1.000 (in./rev.) will allow backdriving by hand.

I too am looking at this for the precision but, I think if you can only get .004” inch in the X-Y then the
Z should be about the same, or do you see an advantage with a higher Z axis resolution?

The Z axis resolution with a 1.000 inch lead, with 200 steps: 1/200 = .005” inch.
This seems acceptable given that the X-Y will be .004” inch. I much prefer +-0.002” inch from the X, Y axis (one of my targets).

Or has Nicholas calculated a +-0.002” inch in the X, Y axis for LISA?

Screw Weight of a Hi-Lead Dia. 1.000, Lead: 1.000 (in./rev.): 2.089 (lbs./ft.) x 48.00” inch = 8.36 pounds.

Could a NEMA 23 spin 8.36 pounds without issues, and what mods will the DIY microcontroller require?

Hi-Lead
Dia. 1.000, Lead: 1.000 (in./rev.), Price: $27.94 \ft.,
Screw Weight 2.089 (lbs./ft.)
[www.roton.com]

Some HI - LEAD screws will backdrive under certain conditions and all TORSPLINEs will backdrive.
[www.roton.com]

Edited 1 time(s). Last edit at 12/10/2013 02:28PM by A2.

Nick
Got any experience with threadless leadscrews?
Ever experimented with them?

Would like to hear your take.

Sure would make it much cheaper, if it has enough precision.
Perhaps using them long enough they would become threaded leadscrews?

Quote
cozmicray
Nick
Got any experience with threadless leadscrews?
Ever experimented with them?

Would like to hear your take.

Sure would make it much cheaper, if it has enough precision.
Perhaps using them long enough they would become threaded leadscrews?

No experience. That is why I went with what I knew. I will eventually try them out.

My theory on the electronics is that they are the cheapest part of this whole thing on a large printer. I can go with high current drivers (3-4A) for < $20 each. Interfacing them to a Ramps board is either a bunch of plug in jumpers (4 per stepper) or a evening layout of a pc board to plug into the stepper module socket. Given the confusion factor, I doubt the bunch of jumpers solution would work long term.

Once you have gone to bigger electronics, NEMA 34's are also something to look at. For about $200 ($70 ea delivered) or so you can get some massive motors.... Adding torque to get speed is (relatively) cheap.

My main reason for avoiding the Acme screws is that I really like the "push it up to zero it" idea. I also like the drive efficiency of the "better" screws. The torque is cheap thing only goes just so far.

Edited 1 time(s). Last edit at 12/10/2013 07:23PM by uncle_bob.

I just had my hands on a 1.000" inch 8 pitch, 72" inch long lead screw, and it's heavy.
Suspending the lead screw with one hand, I attempted to rotate it quickly, the torque required was unimpressive, (i.e. easy to rotate).
What I think could be a potential failure mode is the weight of the lead screw resting on the bearings that reside within the stepper motor.

Motor bearings should never be axially loaded, as they don't have thrust bearings.
Deep V-grove bearings is an attempt to address axial loads, but I don't know what they put in a NEMA motor.
If the single shaft NEMA motors has a bronze bushing on the end of the shaft it wouldn't last long.
If there are 2 deep V bearings, and the bearings are up against a stepped edge, then it might work for the DIY'er/hacker/maker.

Has anyone taken a look at the bearings in a NEMA 17 or NEMA 23 stepper motor?, it would be good to know what's in there.
Not knowing what bearings are being used, I would test/use a NEMA 23.
The best solution is to take the weight off of the stepper motor bearings.

uncle_bob, maybe you can specify an electronics package, or a guideline to follow?

@A2: Even with small LISAs it is important to register you screw between the two plates with the coupler on one side and the thread clamp on the other. The stepper should be fastened to the coupler after this. There should be no axial load on the coupler.

I suggest that you loosen and re ighten the coupler every 3 months or so just to be sure that there is no axial load.

Quote
cozmicray
Nick
Got any experience with threadless leadscrews?
Ever experimented with them?

Would like to hear your take.

Sure would make it much cheaper, if it has enough precision.
Perhaps using them long enough they would become threaded leadscrews?

A threadless ball screw would save a ton of money.

I think what you are suggesting is that the bearing edge would roll form a grove into a soft metal shaft, and that would act as a guide.
I've had my hands on a threadless ball screw that used an aluminum block to house the bearings, lets assume that there was no give/creep in the metal bearing housing.
It was unforgiving, and would jitter on a non harden shaft. It was difficult to adjust the tension to find the sweet spot.
In some areas it worked awesome, and in other it would bind, and in some sections it would slip.
If the grove was not worn in evenly you will have regions on the shaft that would still have these issues, (i.e. a variable and meandering pitch, and varying degrees of fit).

Now what I have not considered was a bearing housing made of a flexible material.
It might act as a dampener, and compensate for imperfections of a soft metal shaft?

I like the idea, I love threadless ball screws, the cost savings would be huge, but the idea needs some R&D investment.

Quote
nicholas.seward
@A2: Even with small LISAs it is important to register you screw between the two plates with the coupler on one side and the thread clamp on the other. The stepper should be fastened to the coupler after this. There should be no axial load on the coupler.

I suggest that you loosen and re ighten the coupler every 3 months or so just to be sure that there is no axial load.

To reiterate, the thread clamp located on top of the upper platen is supporting the weight of the lead screw, that will work fine then.

In view of that, it's possible then that the little NEMA 17 could spin a 1.000" by 48" inch long lead screw.
But I'm kind of on the fence to suggest it's up to the task to spin the heavy weight, it would have to be tested.

To the (very limited) extent I have any plan for the electronics here's what it is:

1) If possible go with a ARM based system. That *assumes* that the firmware on the ARM is up to date enough to handle any of this. That adds $10 to the processor board and massively increases your compute horsepower.
2) Stick with a Ramps (if not ARM) or Ramps like control board if it is ARM. They are cheap and easy to get. You need something to run the extruder(s) anyway.
3) Find a motor that makes sense. They will cost more than the controllers, so optimize on the expensive part first.
4) Get a couple of Chinese opto isolated stepper controllers to match up with the motors. Start with the $18 ones and work up to the $30 something ones if needed.
5) Fab a little board to plug the opto controllers in place of the normal stepper modules. The cost is trivial and the error potential goes way down.

All of that is more in the line of "making a list" rather than making decisions. It raises as many questions as it answers. I believe that the 3.5A or so that the $18 controllers should drive most of the motors I found to a reasonable torque level. These motors will get just as hot as a normal motor at full current. I'd expect to run them below max. If I go with a fine pitch cheap lead screw my RPM's will be 4X higher than with a low pitch. That likely pushes me up to a 24V supply The cheap controllers can go up to 48V. Is it needed? Power supplies are equally cheap at 12, 24, or 48 volts.

1) Screw at 0.25" pitch (worst one I've found so far)
2) 10" /sec ( 250 mm/sec) feed rate
3) Screw is running 40 revs / second (2400 rpm (yikes))
4)1.8 degree / 200 step stepper is running at 40x50 = 2,000 Hz drive
5) if it's a 3mHy winding, that's 37 ohms

One amp of drive at 3mHy at that speed etc would indeed like 48V a bit more than 24V. It's also a bit questionable whether the controllers run PWM fast enough to do a 2KHz with 16 micro steps properly. 32KHz is a bit close to (or above0 where they are running.

Do you need 250mm/sec on the rods? I don't know. Never done one of these here Delta things before. Seems pretty fast ....
Is 3mHy a reasonable winding inductance? At least I've got a number to compare to as I shop motors.
Do I *really* want to go with 0.25" pitch? .... not if I can help it.

@uncle_bob
Tks for sharing your notes, I'm looking forward to seeing what other ideas you may have, and what your final configuration will be.
I don't have an electronics background, so I appreciate all the help that you are able to share.

I'm assuming your analysis is for the NEMA 23?

From what I have read, and the information that I was provided on the SmoothieBoard IRC, the
SmoothieBoard allows for quick, and easy machine reconfiguration.
Meaning you can switch between a 3d filament printing, laser cutter, router, EDM, pick and place, etc.
One microcontroller with a heavy duty printer will be very utilitarian.
I don't know if the Marlin firmware, and associated software will work.
I was told that Marlin would have to be recoded to work with a SmoothieBoard, but I'm unsure.

Quote
Hack A Day
the Smoothie uses a 32-bit ARM chip in the form of an NXP LPC Cortex-M3 chip.
Not only does this allow the Smoothie to do some very cool things with your machine – native arcs and circles,
for example, but this better hardware also allows for Ethernet, drag-and-drop firmware,
and exposing the USB port as both a serial port or mass storage device.

The Smoothie comes in three flavors, with either 3, 4, or 5 stepper motor drivers.
These Allegro A4982 drivers are good enough for any 3D printer, laser cutter, or small mill,
but the broken out pins allow for stepper drivers supplying more than 2A of current.

Everything on the Smoothieboard is modular, meaning this board is equally capable of powering a RepRap, mill,
laser cutter, or plotter. There’s even a planned control panel called the Smoothiepanel, making this a great choice for your next CNC build.
[hackaday.com]

[www.kickstarter.com]
[smoothieware.org]
[smoothieware.org]

Quote
uncle_bob
1) Screw at 0.25" pitch (worst one I've found so far)
Do I *really* want to go with 0.25" pitch? .... not if I can help it.

It appears that you are price driven, it is a huge savings.
I'll wait on my decision, as I want to hear more about your thoughts of the electronics/hardware that is required to drive the finer pitch.
I'm assuming that you are not motivated by the Z axis resolution, but by the total project cost?

There will be an increase in cost due to the larger ID dia 1.000" inch bearings required.
An ebay search indicated about $12.34 each x 6 = $74.00 usd.

I'm aiming to keep my expenditures total to $1000.00 usd, but if I get a hydra machine out of it,
well it's like getting a bunch of machines for one price (my justification for going over budget).

1 X .250 Right Hand Hi-Lead®
Screw Weight: 2.280 (lbs./ft.)
Price: $13.99 \ft
[www.roton.com]

1 X .500 Right Hand Hi-Lead®
Screw Weight: 2.000 (lbs./ft.)
Price: $24.34 \ft.
[www.roton.com]

1 X 1.000 Right Hand Hi-Lead®
Screw Weight 2.089 (lbs./ft.)
Price: $27.94 \ft.
[www.roton.com]

The Z axis resolution with a 1.000 inch lead, with 200 steps: 1/200 = .005” inch.
The Z axis resolution with a 0.500 inch lead, with 200 steps: .5/200 = .0025” inch.
The Z axis resolution with a 0.250 inch lead, with 200 steps: .25/200 = .00125” inch.

This is a really neat design
but
these massive leadscrews are getting out of hand?
I can see heavy leadscrews on a CNC that has to push-pull a big spindle / tool around
but
for squirtin plastic?
The elegance of the ONLY leadscrew device is one thing.
but Hundreds of dollars for mechanicals is not good.

Especially for a printer that maybe has a life span of maybe a year,
the way 3D printers are moving so fast?

Keep up neat development --- keep in mind what it will take for others to do a build?

Axial load on common ball bearings is a problem!!

So shouldn't you guys be looking at thrust bearings
or thrust bearing -- axial bearing combination?
and
Is plywood and PVC strong enough to handle all of this




"A2 --- Motor bearings should never be axially loaded, as they don't have thrust bearings.
Deep V-grove bearings is an attempt to address axial loads, but I don't know what they put in a NEMA motor.
If the single shaft NEMA motors has a bronze bushing on the end of the shaft it wouldn't last long.
If there are 2 deep V bearings, and the bearings are up against a stepped edge, then it might work for the DIY'er/hacker/maker."

@cozmicray: Regular bearings are rated for an order of magnitude more than what we are asking of them. 6' of 1" threaded rod is only 16lbs. A 1" ID bearing can probably handle 250+lbs.

I only use thrust bearings when absolutely necessary. My first hobby mill that I designed and built used a thrust/radial/radial/thrust stack on one side of a floating lead screw. Maybe I did something wrong or got bad thrust bearings but I threw them away and my mill is better for it.

Cost is the main reason to use cheap radial bearings.

@cozmicray

Quote
cozmicray
these massive leadscrews are getting out of hand? ... ... for squirtin plastic?

For my self, I want the rigidity so I have the option of utilizing different manufacturing technologies.
The current design is not rigid enough for my needs, it will need to be redesigned.
I'll probably scale back on the height, as the majority of the parts that I have designed are low and wide.

I'll start with a normal size platform, due to the arm length limitation, and budget constraints.
As my skills increase, and time/money allow, I'll make a set of longer chunkier arms, wider platens, replace the PVC with steel tubes, ect,
I will continually reevaluate the design as I learn, my goal is to evolve it into a ridged multi use platform.

I like the idea of printing a plastic part, but hole diameters and tolerances are lacking.
Switch the end effector tool to a Dremel, and the software in the microcontroller,
and you have a mini router to clean up the holes, and perimeters.
.
So to do some things the platform needs to be rigid, hence the large lead screws.
I hope that this is possible, I don't know, this is my first attempt at building a 3d printer.

Maybe some thing more interesting to me will come along, and I don't build this one.
I don't know where this road will lead, but this looks useful for my needs at this time.

@cozmicray:

Read a few post up Nicholas said that he is using a thread clamp to keep the load off of the stepper motor.
The thread clamp is located on top of the upper platen bearing.
So there is no axial load on the stepper motor bearings.
As Nicholas just said the weight of the lead screw is inconsequential to the ability of a Dia 1.000" inch bearing to support it in a thrust application.
It's a fly on an elephants arse

It's the stepper motor bearings you need to consider if you were to place an axial load on the rotor shaft.

We don't know what the design configuration is inside the stepper motor housing surrounding the bearings.
If there is not a shoulder behind the bearing, and/or the bearing is not press fit properly (loose),
the stepper motor shaft has the potential to shift and the rotor will crash and burn.

Over time vibrations can also cause bearings to shift if there is no support.

I may be a little off topic, but maybe we can take some ideas from this new KS low cost Delta printer [www.kickstarter.com]

On LISA
I thought tension of the leadscrews was important.
Thus tensioning the leadscrew causes Axial load on
top and bottom bearings in the plywood plates.

What is the purpose of tentioning the leadscrew?
Can you really tension a steel leadscrew much with a polymer clamp/nut?

If the leadscrew is tensioned between top and bottom plates
the stepper motor is not seeing any load.

Wouldn't a taper bearing top and bottom do the trick?

It looks like you can get NEMA 24 motors with roughly 4X the torque of a typical printer NEMA 17 for about $100 (3pcs delivered in the US). That should be enough torque to create other issues if you ever used all of it.

Stepping these motors fast enough to get a 10" / sec feed with a 1/4" screw isn't a real easy thing to do. I think that with 1/16 micro stepping, the Z resolution (or x or y since they all couple) will be good enough on a 1" screw. It is indeed resolution and not accuracy in this case. There are a number of things that will contribute to accuracy.

As long as the motors are 24's the 3 to 4A driver cards should handle them just fine. Since they are opto isolated, they will lash into almost any controller board you would want to use. I've already got Mega's and Ramps 1.4's lying around here, so I'd start up with those. Once the Marlin (or what ever) on ARM gets sorted out ... and the Adruino IDE for ARM gets sorted out ... and the control cards for 3.3V i/o to ARM get sorted out ... upgrade to an ARM. Even bought new, the Mega + Ramps is $40 or so. That's not the big item on a $1K parts budget. The ARM world is only slowly coming to printers. One really cool new thing at a time...

-------------------------------------

Looking at growing the LISA in X and Y - the arms grow as roughly 2/3 the screw spacing if I read right in a post above. Taking that as an accurate number, for a 600 mm screw spacing one would need 400 mm arms. Those aren't going to print easily on a common printer. Even more so if you put a skirt / brim around them to keep things flat. I'd rather not go to a pair of printed parts mated in the middle with screws, my guess is that they would not be very accurate. Some sort of aluminum bar might work, printed parts plus thread rod might work, neither gets me real excited. Am I being overly concerned here?

@cozmicray: I used just enough tension to remove play. I wanted more tension but this was sufficient. Angular contact bearing would be pricey.

Using the equation ... to figure out what a practical max height would be for ... 7/16" screws ... a 300mm diameter build area ... could be 350+mm tall. - 12/07/13

I would love to crunch out the recommended geometry - 12/10/13

300 x 350. Sounds like a winner. 28" rods. Waste not want not.

Geez taper bearings used on just about every wheel rolling out there.
Can get bearings and races all over the place.
Take them off your go-kart or your neighbors boat trailer

Cnn you give us some urls to appropriate low-cost taper bearings? Thks.

@nicholas.seward
Listen to how loud the bearings are on this impressive speed test of the Mini Kossel printer.

I'm thinking that LISA should be very quite with the thread drive , but I don't know...
Time permitting, please post a video without the sound track, tks.

Mini Kossel Speed printing
[www.youtube.com]