How much torque?

I'm poking around asking my search for good motors, but it's got me wondering, how much torque do we need? How much do we want?

1kg-cm = 98 mNm = 13. oz-in = 0.87 in-lb

There's two different drive methods in common use. One is the Darwin belt drive, the other is the threaded rod drive.

Solarbotics motors are doing something like 40rpm and 100oz-in before clutch slip. Nanotec steppers are doing 80 oz-in holding torque, and do 2000 step/s, so 300rpm. I've found a gearmotor at Lynxmotion that can just about match that and has position feedback.

[www.lynxmotion.com]

That's 225 oz-in stall and 200rpm with a 12 ppr encoder (x 30 gear ratio = 480 ppr). $40US.

So, low RPM, high torque benefits the Darwin design, and high speed but low torque benefit the leadscrew design. I know most hobby CNC setups require a 100 oz-in torque holding/stall rating. Anyone have any other thoughts on this?

DC motors such as the last one in this PDf would definitely pull its weight (cost is $7.10 singly too) would need an encoder still but can hit very high speeds, letting leadscrews move fast if you can get past the friction.

[www.anaheimautomation.com]

I'm also researching brushless motors and low cost linear bearings and drive shafts, but that's probably beyond the scope of this. Once I have a motor and encoder on hand I'm planning on testing out a few different designs. I'm trying to determine what would be a good starting point and what you can get away with.

I have presently ordered motors from this company [www.kelinginc.net] they are the kl23h256-21-8b I will let you know if and how I got them to work. This way if you get different motors then we can get different options. I do not know how well the encoder would work or how it would be set up for a motor using one?

Bruce Wattendorf

Hi,

I may be off--I'm not that good with all this "stuff" yet so bear with me--but can't we measure the torque required on one of the few repraps already built? Or, barring that, can't we calculate it from the given mechanics of the idealized raprap? I'm a student (mechanical engineering) so let me know why a thing isn't possible. I like to know.

Demented

Calculating can be tricky. Setting outside boundaries, otoh, is fairly easy to do.

I'm using Solarbotics GM8 and GM9 motors for the x, y and z axes on Tommelise. They stall at 43 oz-in. In ordinary operation, where I haven't let one of them get stuck at the end of an axis, they never stall.

Similarly, I use a GM3 for the extruder polymer pump. The clutch on that one engages at 60 oz-in. When my extruder is operating properly I never hear that clutch engaging.

My x and y axes run at about 1.8 mm/sec. Tommelise uses threaded drive rods.

You can take from all of that that Tommelise is running using well under 40 oz-in per axis.

If you're building a machine from the start, with known machining tolerances and mechanics, you can determine your torque. Here I'm looking at what people have found useful in their Repstrap experiments. We also have to look at torque for the extruder, something that's being experimentally gathered.

I look at various drive mechanisms. Now, the losses in Acme and related screws are well documented as shown here [www.roton.com] . You can also get that of linear bearings, such as here: [www.igus.com] . Ideals can be calculated given surface area, normal forces, and coefficients of friction, but experience can show the pitfalls you might find. Especially since I suspect that allthread is no better than 20% efficient in power transfer, and THAT needs experimental testing. And lubrication.

Now, your standard CNC package for a small steel/cast steel bench mill uses 200-300oz-in at stall motors. Darwin currently uses 100oz-in holding torque belt drive steppers. Tommelise and other 'Straps use 50 to 100oz-in gearmotors that drive the allthread rods, but it's much slower in non-dispensing rapid mode. This is probably sufficient, but as the design gets more complex and the head gets heavier, both the inertia and the load friction is going to possibly overpower these issues. Now, I'd like to speed up the screw designs in my own designs.

Now that I've got a good idea about how to build a n x 12V PSU, I can get the brushless controller working. Might pursue those KL23BLS-95 on the keling web site. It's aboult 2/3 the price I was looking at and double shafted.

Let me know how those stepper motors work!

SOI,

Can't we just move to a stationary head design where the xyz stages move and not the head. Someone, somewhere in the forums, has mentioned this before--I remember reading something about it. That would take care of increasing head weight and friction. The xyz stages would only be expected to take a minimum of weight in total.

Demented

Zaphod works that way.

Demented Chihuahua

You still have changing mass due to the item being built and it will be different every time. You also need a lot more desktop space as the table needs 4 times its area as an overall envelope. The current design i feel is an extremely good first start and incidentally is how most of the commercial printers work

The motors being used have loads of torque for this application and most stuff to do with friction etc is more about bearing design not motor torque, the torque only really affects acceleration after the friction is taken care of, that's why strong men can pull airplanes albeit very slowly.

Well I got the motors in from www.kelinginc.net and they worked after I reloaded the newest firmware in my chip's they seem to be a small bit of vibration but if I adjust the torque they are not bad at around a setting of 210. They drive everything with no problem and are double shafted and the same size as the nanotech motors. and in seeing the first Darwin last week seem to be running very close to it's steppers. And after seeing Zach's work with the l297 will only run even better.

Bruce W.

There will probably be a small bit of vibration. That's normal for non chopper/non microstepping driver setups. The torque/current is high at the edge of a step and it drops as it centers, and it'll usually oscillate past and back again due to the rotor inertia.

I'm planning on a Tomellise 1.0 style design. I have other uses for the structure beyond RepRap, so the structure has several advantages. Only 2x floor space requirement and it still can "roll the part" out for easy access. Simple to add multiple tables to if you have it doing mutiple fixtured parts. This also means that one "table" can be a head caddy for automated head swapping. Heck, you could either use the backside of the bridge (Y axis carrying head) or add on another bridge to allow for simultaneous work on two parts that share a single head repository.

While I'm thinking of drive systems, I also have to consider drag problems on this. Normal Acme screws are 40% efficient, and lower pitch designs are up to 70%. Of these, I'm working on a rule of thumb of $10/ft and $15 per plastic flange nut. I'm guessing that threaded rod is about 30% and has a lot more slop in it. A spring sleeve slide nut (a box or U channel with one nut epoxied in place, a sleeve between the tension spring and the rod, and the other nut captured in the channel) will probably lower this a bit due to the extra tension.

1.8mm/s appears to be a good top speed for current plastic extrusions. Might not be for larger orifice "bulk" extruders, but it'll do for "standard def" for now. I'm pushing a bit higher This will probably hit Godzilla speeds when transitting. I'm hoping this will leave a "cleaner" break from the extrusion and if any repositioning is needed between layers or around layers, this should cut down on the "nonproductive" time.

SOI Sentinel Wrote:
-------------------------------------------------------

>
> While I'm thinking of drive systems, I also have
> to consider drag problems on this. Normal Acme
> screws are 40% efficient, and lower pitch designs
> are up to 70%. Of these, I'm working on a rule of
> thumb of $10/ft and $15 per plastic flange nut.
> I'm guessing that threaded rod is about 30% and
> has a lot more slop in it.
>

LOL! From my experience, I figure that you could get 30% efficiency out of a threaded rod if you sprayed it every hour with teflon lubricant.

Mind, the really fine pitches gives you a LOT of mechanical advantage, so you don't need a lot of torque.

BTW, you can buy spring loaded antibacklash lead nuts for acme screws for about $30, IIRC.

[www.nollinc.com]

Edited 1 time(s). Last edit at 10/17/2007 02:43PM by Forrest Higgs.

Wait, you haven't FDM'ed yourself an auto-lube system yet?

Yeah, I'm probably high on my nut calculations, and I'm working on an Excel sheet (although I should probably move it to an OS Matlab clone) for testing out different efficiency ratings of various thread types and pitches. I'm actually thinking of working with 3/8-16 rod. Higher inertia will absorb vibration better and it's capable of longer runs at speed. This is also because I'm considering using the same hardware to build my father a CNC router table.

That nut cost is about right for an anti-backlash nut. I'm mostly debating whether I'd want to go that route anyway. I'll probably experiment with threaded rod first. I know where to get low friction plastic bushings ( [www.igus.com] they provide low quantity pricing online) for less than two dollars each (Iglide G300 and J materials), so I might be able to dodge needing multiple roller bearings per parallel shaft and keep my cost outlay reasonable.

Of interest, PIC designs has done a lot of testing and characterization of how fast you can spin a rod.

[www.pic-design.com]

Their catalog is almost a bible on correct small motion part selection.

I wonder what the backlash of a 0-80 or a 2-56 rod and nut combo are...

Edited 1 time(s). Last edit at 10/17/2007 03:05PM by SOI Sentinel.

SOI Sentinel Wrote:
-------------------------------------------------------
>
> I'm actually thinking of working with 3/8-16 rod.
>
I'm currently using 3/8-24 rod, myself. It was as close to the 1 mm/turn metric pitch that Adrian and the rest of the crew were using at that time as I could get.

The only problem with threaded rod is that if you buy it from the hardware store it tends to have been roughly treated during shipping and has serious bends that are very hard to iron out. I had to design a sliding thrust collar to account for them on my x and y axes.
>

>1.8mm/s appears to be a good top speed for current plastic extrusions.
I can easily achieve 4mm/s @ 1mm diameter with HDPE in the RepRap extruder without stressing the motor too much. If your head moves slower than that you either limit your build speed or lay down thicker filament giving a lower resolution.

---

[www.hydraraptor.blogspot.com]

how much torque to apply for new rod bearings, when originals had 30lbs?
I used 25lbs and crankshaft wasnt turning.

we are looking for drive 1kg weight shaft with uniform motion at 1 to 500 rpm. how to select is servo motor.