Welcome! Log In Create A New Profile


LISA Simpson

Posted by nicholas.seward 
Re: LISA Simpson
December 04, 2013 01:45PM
Maybe a dumb question: why did you go with the Roton Torqspline rather than the originally specified ACME rod? I didn't see any discussion around that.
I am willing to contribute to the cost of a tap, if you need one.

Looking very good. This design can be adapted easily to other methods of support structures, but will be interesting to see how much support is needed.

Cheers to the team.
Re: LISA Simpson
December 04, 2013 03:35PM
I have had some positive experiences with ROTON. I figure if anyone is going to have a straight screw it will be them. Also the Torqspline is the same cost but is more efficient. The coolest thing is that it is non-locking and you can back drive with the same efficiency. (For an ME this is icing on the cake.) Homing for now will consist of me pushing the carriages to the top. (I can have a screw come down for microadjustment.)

We made good progress today. We get to build again on Friday. I think it will be almost all done by the weekend. First moves and possible first prints are next week. (or my students don't get a good grade :-) )

Edited 1 time(s). Last edit at 12/05/2013 04:02PM by nicholas.seward.
Re: LISA Simpson
December 05, 2013 01:47PM
I know two things..

It is very easy to make a tap for a specific thread if you have a piece of threaded rod with that thread. Can be done with a grinder or even a dremel tool. The info is on the interweb. Have done if many times as acme and simular thread taps are pricey. If you don't want to harden them make two and you can even tap a set of brass nuts.

Threadless lead screws work very well and maintain incredable accuracy but the 'screw' needs to be hard and very round. The 'screw' gets marked pretty quickly by the angular contact of the bearings (which may be the reason for its accuracy. Funny though, because it is only a mark and I have never been able to measure any wear). Out of round, like drill rod or cold rolled rod does not work well as the drive is lumpy if the 'nut' is ridgid. I have only buit them with metal bearing holders, maybe a printed plastic 'nut' would have enough give that you could ignore out of round.

@ nicholas.seward: Have been watching reprap for a long time with mild interest until you started in on truely alternate forms (wally, simpson and Quenton's morgan). It is getting exciting now. I love that you went from "Lisa is just a concept, I'm not going to build it" on the 16th to ordering parts on the, what, 21st. I feel honored to get to follow along. Thankyou for sharing all of this.

@ Dannydefe: I love that you jumped right in there and started to build one..

I just inherated a mendalmax from my tool and die maker brother (He's been trying to build it for a year and keeps breaking plastic). It wll be a few months before I can get to it. Can't wait to see what has happend by the time I can print. Really like the parrallel scaras but Lisa Simpson even appeals to my heavy metal cnc brain. Chears and thank you all. Jan
Re: LISA Simpson
December 05, 2013 03:30PM

News flash: one plastic nut won't keep you straight on the shaft. There is some play. Backlash is one thing but being able to rotate no matter how little will mess with the Simpson geometry. The image above shows my intended solution. I have already printed the shoulder and I want the threads close to the pivot to reduce friction. This dictated that I go with a sleeve above or below the shoulder. Above decreases the build volume and I already had nut traps so I had to put it below. I think it is actually the best solution even though I just tacked it on to keep from printing as much.

With only one screw 3/8" that only has a .25" solid core I should have realized that we are going to have a lot of flex. (For reference a rail delta will use at least 2 8mm rods so the rigidity is about 5 times that of this one screw.) I think the forces generated in a normal print will keep us on the good side of the line but I want to go up to the 7/16" that has a a .327" solid core. That doesn't sound like much but the rigidity of a cylinder is proportional the r^4. So The 7/16" screw is .327^4/.25^4 or 2.9 times more rigid. That should do it for sure. I am going to carry through with the screws I have. It may turn out to not be a problem even at high speeds. The next step up of the Torqspline is 3/4". This would be 20 times stiffer. I think that would only be needed if someone wanted to use a dremel.

The PVC structure seems to be rigid enough. As expected, all the flex is from the 1/2" boards. I think it is more than sufficient.
Re: LISA Simpson
December 05, 2013 07:08PM
What's the cost impact of making the rods larger in diameter? If it's modest, bumping up the diameter is a pretty easy decision.
Re: LISA Simpson
December 05, 2013 07:18PM
Re: LISA Simpson
December 05, 2013 10:30PM
I know you are trying to keep this with ONLY leadscrews
would one guide rod on each corner help
A change to your trolly to capture a linear bearing riding the rod?
It would prevent the trolly from twisting and enhance rigidity

? Is this just for extruder -- or are you designing for a mill tool?

confused smiley
Re: LISA Simpson
December 05, 2013 10:44PM
@cozmicray: A linear rod would completely change the geometry. It would cause a cascade of changes that would essentially turn this back into a regular Rostock with lead screws instead of a belt. BTW, the twisting won't be hard to account for.

This version is just going to be for 3D printing. You can scale up the steppers and the screw and be rigid enough for a mill.
Re: LISA Simpson
December 06, 2013 07:21PM
Could you print a supporting threaded piece that applies slight pressure against the top/bottom plates using feet and thread tension? Maybe you could gain some stability here by having two of the legs wrap around the PVC, or drill into the PVC somehow? The arms shouldn't hit bottom, so there is some room to play with?


    |~|     |       |
    |~|     |       |
    |~|     |       |
    |~|     |       |
    |~|     |       |
    |~|     |       |
    |~|     |       |
    |~|     | _____ |
   //  \\__/|       |\
 //||  ||\            \

  __       /  __----__ \ \ 
  \   \__/  /          \  \__ 
     | 0   |                  |
     -----  \          /  /----
  /__/    \  \_      _/  / 
            \   -__-    /

Edited 1 time(s). Last edit at 12/06/2013 07:24PM by jason.fisher.
Re: LISA Simpson
December 06, 2013 08:34PM
We had a snow day today so I came in and my students and I got LISA done!

LISA Simpson is together and printing!!!! Above you can see as far as I got on two squirrels. The purple one was using unknown settings. The orange one was Slic3r defaults. (The couplers slipped on both and the filament drive melted during the orange one.) They obviously failed.

Video of first moves!

*The inverse kinematics is all worked out. I will attach my ugly preprocessor if anyone wants to dig in. I was coding to get it done not to make it look pretty or be maintainable.
*I forgot to trim the potentiometers so I was over currenting the steppers. They got hot enough to make the couplers start slipping. I will print new couplers that are a little better and fix the current.
*The 3/8" screw is good enough. I did have one that wobbled due to a a coupler misalignment. That can be fixed with a bigger screw or two bearings at the bottom. I think I will add the extra bearing. It will just be a printed standoff, a bearing, and another thread clamp. I could then technically remove the top thread clamps but it will be better with them.
*The PVC with all thread holding it together is phenomenal. The rigidity is perfect. You can pick the bot up from anything but the screws and move it during printing. Printing upside down or sideways won't be an issue.
*Bed leveling is perfect before calibration. Once I fix all the problems listed above, I want to print something huge.
*Dimensionality it perfect.
*The sleeves I put below the nuts work brilliantly.
*Putting about 5lbf on the extruder only moves it about 1mm. (I didn't measure that is just my guess.)
*Even with wobble in one screw I don't think you will see the artifacts it creates for most prints. It would take a print that was designed to show it.

I am now going to leave this at school and try to get home in the ice so you will have to wait until next week to see a video of it printing.

I will post files soon once the last few bugs are worked out. My question for you guys is... should I convert this to the 7/16" screws that are 2.5 times stiffer before I publish? It will cost $60+ more.

Edited 1 time(s). Last edit at 12/06/2013 08:35PM by nicholas.seward.
open | download - simpson segmentize.py (4 KB)
Re: LISA Simpson
December 06, 2013 10:56PM
Many moons ago the starting point was:

Heavy smooth rods + support rods / what ever cost something
Drive screws cost << the normal rod set for a printer this size

I think I can still convince my self that even with the bigger drive screws, it's still a bit cheaper.

The rotational mass of the screws will be higher. I can't see that slowing the printer down.


Speaking of motor current - how are the NEMA-17's doing when you try to speed this thing up? Put another way - could it use more torque? It will never win the Indy 500, but it might go pretty fast....

Edited 1 time(s). Last edit at 12/06/2013 10:56PM by uncle_bob.
Re: LISA Simpson
December 06, 2013 10:56PM
I want one, it looks so smooth! thumbs up

Lead Screw Wobble:
A bigger lead screw is the way to go, or you'll hear complaints.
Straightness of the lead screw varies, poor shipping and handling practices will more readily bend the smaller lead screw.
I vote for the larger lead screw.

Add the extra bearings, it will help average out the slip fit, even for the larger lead screw.
A hack to get the lead screw to press fit into the bearing (which it should be) is to straight knurl the OD of the lead screw.
A DIY knurl could be done with a chisel and hammer.

The bearings over time, and especially if there is a wobble will work loose, and migrate out of their pocket.
Use a button head cap screw, or use over sized screw heads, which will allow you to move the screw hole away from the bearing pocket.
Or use a large washer to cover the bearing, drill some holes in the washer so you can retain it.
Or use a piece of sheet metal as a strap to bride across the bearing, and screw it down.

Super structure (base):
Best to have some separation of the bearings.
Consider gluing together two 3/4" inch pieces of MDF/Plywood/etc.

Axial end play of the lead screw, To aid in retaining the lead screw against it's reference shoulder:
Add a Teflon or "soft" Nylon washer between your clamps and the bearing to help prevent translation.
Wave Disc Spring.
Belleville Disc Springs for Ball Bearings.
Corrosion-Resistant Acetal Spring Washers.

Wave Disc Spring

Belleville Disc Springs for Ball Bearings

Corrosion-Resistant Acetal Spring Washers

Lead screw clamp placement:
Both lead screw clamps should be located on the bottom of the platens.
This will allow the wave disk springs to apply a force in the same direction (i.e. push the lead screw down).
Adding a wave disk spring to opposing sides of the platen cancels out the spring force.
Locating the top lead screw clamp above the platen forces it to work against gravity.

I can see that the top lead screw clamp aids in retaining the bearing, but as I mentioned above there are simple alternatives.

Edited 6 time(s). Last edit at 12/07/2013 01:55AM by A2.
Re: LISA Simpson
December 06, 2013 10:59PM
Thanks, that was nice present for Friday evening. :-)
While it would be nice to see the large screws, I am not convinced it is worth the extra bucks. I am willing to buy the screws for you, if you want to try, though.
Re: LISA Simpson
December 06, 2013 11:21PM
Cross posted on this thread:

DIY $1500 Metal Printer
Re: LISA Simpson
December 06, 2013 11:38PM
Excellent Work! You and your students are fantastic.

I would not publish the 7/16 screw version without test building it first. You never know about the gotchas until you try. Unless there is an obvious benefit for it, that is a significant difference in cost.

I was thinking about the 480mm Tr8*8 or 8*12 integrated NEMA steppers available today. It seems that something like that would make an even simpler setup, but more limited in size.
Re: LISA Simpson
December 07, 2013 12:04AM
If you want to get exotic with the wood, you can get 25mm birch plywood. It's not as easy to find as the 18 mm stuff. There are a few places that used to go even thicker than 25mm. I have not tried to get anything really thick in a while...
Re: LISA Simpson
December 07, 2013 01:28AM
Arm jitter:
At 35 seconds into the video I believe I saw the arm furthest from the viewer jitter.
The other two arms nearest the viewer made the transition smoothly.
Any idea what happened?
Re: LISA Simpson
December 07, 2013 06:38AM

I would like to see Hall Effect Sensors employed, as they are very accurate, and would complement the lead screw drive nicely.
I think this style is cheaper and easier to install than the Hall-0 type.

They just have three pins, GND, Vcc, and signal. The signal is normally open collector, get a magnet close and it drains to ground.
That’s right. The three pins are _exactly_ the same pins, in the same order, that RAMPS expects its endstops to have.
The only setup you need is a 3 pin jumper cable you can stick the sensor in. Simple servo cables from pololu or any RC store work perfectly
and there is no need to solder anything. When looking at the larger flat side of the sensor,
the left leg is the signal so should be oriented towards the outside of the RAMPS board when plugged in.

Video: Hall effect sensor on fridge magnet

eBay Hall Effect Sensors for end stops, cheap ~$3/qty10, including shipping + the cost of magnets.
Accuracy about 0.01mm.


Cheap Hall effect sensor endstop:
These things are very exact, completely without wear and really easy to exchange with the mechanical switches.
Different to the Hall-O (See 2.), they trigger on a fixed strength of a magnetic field and can't be calibrated. Even more different they only cost 20c per piece

John used A3144 sensors (which are temperature resistent, which is good) but I used OHN3019 for no other reason then availability and price.
Both have an integrated voltage regulator and can be driven with 4.5 - 24V. This allows the part to be directly attached to the endstop connector of the sanguinololu without any extra logic.

Only downside so far: depending on the used parts you get a notable hysteresis.
An example: You drive your axis to the endstop, the sensor triggers at a specific distance. Then you drive your axis 0.2mm back.
If you now try to drive these 0.2mm back, it's not possible because the endstop is still triggered. It needs some distance that the sensor pulls the signal down again.
This distance is 0.5mm in my current setup, but I'm using a small neodym magnet which is properly a bit to strong for the application.

I measured the accuracy after installation with an cheap micrometer. I made 10 tries, always driving a few steps down and full up again.
It seems to differ by about 0.01mm (I got between 835,2 - 836,1).

-- Installation --
These steps need to be made to exchange an endstop with a magnetic one:

1.) Print the part (I forgot to do that first. ;-) )
2.) On the back of the sanguinololu, by the endstop connectors are solderpads which need to be closed to choose the supply voltage of the endstops. This depends on your sensor. I chose 5V.
3.) You need to change a firmware parameter (See 3). In the file "Configuration.h" set '_ENDSTOPS_INVERTING' to true on the axis you are modifying.
4.) Take three 35cm long wires and solder them to the sensor pins.
5.) You can reuse the connector of the old mechanical endstop, just pull the old wires out.
6.) insert the new wires in the back and fasten with a small screwdriver. The pinout again depends on the sensor. Look at my pictures, I used red for voltage, black for ground and yellow for signal.
7.) Use Pronterface to see if the endstop reacts to a magnet. Mine only reacts on the side with the imprinted dot, and only in one direction of the magnet field.
Make lots of dry tests, if the endstop does not pull you will damage your printer! You have been warned!
8.) Glue the sensor into the part and install it and a matching magnet.
9.) Carefully test again


Edited 3 time(s). Last edit at 12/07/2013 06:53AM by A2.
open | download - ScreenHunter_278 Dec. 07 06.29.jpg (19.9 KB)
Re: LISA Simpson
December 07, 2013 06:51AM
This is a great design! Would there be any reason NOT to enclose this on all three sides (using a door, obs)? My workshop has constant traffic in and out and drafts are in abundance. 2 full plywood sides and 1 clear acrylic door would add to rigidity and I could add some fans for cooling and ventilation.
Re: LISA Simpson
December 07, 2013 07:43AM
Homemade precision nuts

You can make your own precision (zero backlash) nuts with a low melt point bismuth allow, Cerrosafe is one of the more popular trade name.

Print a nut shell with maybe 1 mm clearance over the threads.
Add an undercut to key the cast nut permenatly to the shell.
To prevent the liquid metal from running out the bottom use a putty to seal it.

Oogoo might work to seal it:

If you make a mistake casting it, don't worry just remelt it, same goes if you wear it out, it's reusable!

The casting might be too perfect, so you may need to lap the threads by running it back and forth over the nut with a lubricant, and a fine abrasive.
But wait for 8 days as it will expand by about .0025" inch (0.0635 mm), which might make it perfect for this application.
I use a double boiler of water to melt it.
1/2 lbs cost about $14.00 usd.

Cerrosafe shrinks during the first 30 minutes of cooling and then at the end of an hour, is EXACTLY chamber size.
At the end of 200 hours it will have expanded approximately .0025". This factor is well known by all toolmakers
Cerrosafe melts between 158° - 190° F.

For some reason Cerrosafe does not show up on ebay, use this term: low melt bismuth.
McMaster Carr calls it: Low Melting Temperature Casting Alloys, and is quite expensive.
There are sellers on Amazon as well.

Edited 4 time(s). Last edit at 12/07/2013 09:52AM by A2.
open | download - ScreenHunter_280 Dec. 07 07.38.jpg (4.3 KB)
Re: LISA Simpson
December 07, 2013 09:26AM
You can get the casting alloy from:


It has lead in it, so it's going to be a bit of a chore to dispose of in this day and age.

They also have most of the traditional Babbitt bearing alloys at the same sort of price. They are higher temperature materials, so not quite so easy to play with.
Re: LISA Simpson
December 07, 2013 10:08AM
Withouth springs there is no endplay I could detect se we are good there.

This design is end stop free and will probably remain that way. It is very easy to manually home the design and I have it return there after every print. Quickier than electronic. Just as accurate. Less points of failure. Quicker to assemble.

One screw does jitter. Doesn,t affect prints but I will track down the cause.

The speed in the video is way under the top speed. You can probably do 75% of rostocks speed. Faster than we need.

I think I will get the 7/16" screws. It seems like the cleanest solution.
Re: LISA Simpson
December 07, 2013 12:20PM

Dia 7/16" lead screw:
Cool, bigger = better thumbs up

Tell me more about the design, is it's purpose to act as an anti backlash nut?
Does it prevent the main nut from rotating, why do you think it work so well?

*Putting about 5lbf on the extruder only moves it about 1mm. (I didn't measure that is just my guess.)
What component do you think is flexing the most, and/or what joint is the weak link?

Delta router:
Do you think it could handle a flexible shaft attached to a remotely mounted Dremel tool as it is now?
If not what would you change?

It would be very useful to be able to true up holes in the Z axis after/during printing.
A quick change tool holder would be a nice feature to add.

Edited 1 time(s). Last edit at 12/07/2013 12:30PM by A2.
Re: LISA Simpson
December 07, 2013 06:59PM
I have a 4' tall by 2' long/wide MDF cabinet on casters with a glass door that I have been looking for a design to build into.

Would the 3/8" screw be enough with a rigid cabinet? Or at that length, is the larger screw even more important? ~$210 vs $370?

This seems very straightforward, but it may make more sense for me to mount the steppers and electronics on the top of the cabinet for accessibility?
Re: LISA Simpson
December 07, 2013 07:08PM
I've got a variety of cabinets running around here. I'd bet the PVC under compression approach is as rigid as any of them, at least in the directions where it counts.
Re: LISA Simpson
December 07, 2013 08:49PM
LISA Simpson great
I see a very un-happy Wally in corner watching LISA work.

Looking at video:

With the stationary trolley (not being moved by leadscrew) being rotated by the
other two trolleys movement --- doesn't this screw up precise positioning
does software account for this ?

confused smiley
Re: LISA Simpson
December 07, 2013 09:20PM
@jason.fisher: I think that 3/8" is just this side of good for my arrangement. The max deflection is proportional to l^3/r^34. 24^3/.125^4=48^3/r^4 so r=.21in You need to find a lead screw with a solid core of .42in or bigger. Depending on how deep the threads are a 5/8" or 16mm or bigger screw will work. Putting the electronics up top will be great if you are going to have a heated cabinet.

@IntelliTom: Doors would be a good addition. I plan on making some removable boards that have plastic clips that snap onto the the uprights. I don't want them permanently on because it will make a much better demo piece if people can look in from all angles.

@A2: The sleeve keeps the nuts straight and probably add in a slight misalignment that removes all backlash. Without having LISA in front of me, I think that the screws are the weakest link. I would use 3/4" screws (37 times stiffer) if I was going to use this with a dremel. At that point you don't need to worry about removing weight from the hub. I would also make the arms 50% bigger or 5 times stiffer.

@cozmicray: I forgot to bring a spool holder to school so Wally got drafted. Wally was glad to help. Also, the software does account for the translation /rotation coupling of the carriages.
Re: LISA Simpson
December 08, 2013 01:16AM

Here is a visualization of the build volume. It is 260mm tall at it highest. It is 235mm tall close to the screws. (It was a pleasant surprise to see that the top of the build volume is almost flat. I thought the difference between the sides and the top would be bigger than 25mm.) It has a volume of 9.34L. To put that in perspective, an 8" Prusa i3 has a 7.6L volume.

Here you can see the biggest circle and square that you can fit in the area.

I usually just think of the volume as a 200mm diameter by 235mm tall cylinder. That simple approximation is 7.38L or 79% of the true volume.

I really need to see if I can mod Slic3r or Cura to show the outline of the build area to help with part layout.

One other note. Using the equation in my post above to figure out what a practical max height would be for the build volume when I transition to the 7/16" screws, I determined I can add 250mm to the z before I need to bump up the size of the screw again. (That is if I keep the screw spacing the same. If I move the screws out, I lose some height. If I move to a 300mm diameter build area it could be 350+mm tall.) To be clear, I don't plan on increasing the size of LISA.
Re: LISA Simpson
December 08, 2013 01:22AM

I really need to see if I can mod Slic3r or Cura to show the outline of the build area to help with part layout.

Now that would be great!
Re: LISA Simpson
December 08, 2013 01:57AM

This is what I have in mind for the sides. The printed plastic clips should just snap over the PVC pipes.
Sorry, only registered users may post in this forum.

Click here to login