Parametric pulley - lots of tooth profiles

[www.thingiverse.com]

I posted this on Thingiverse last week, but thought it would be worth mentioning here too. It's an openscad model that can create a range of metric and imperial tooth profile pulleys, with any number of teeth, specified in the Openscad file.
Tooth profiles currently supported are MXL, 40DP, XL, H, T2.5, T5, T10, AT5, HTD (3mm, 5mm, 8mm) and GT2 (2mm, 3mm 5mm). There are a range of editable parameters for editing the fit of the tooth (to account for printer variation), the pulley base, captive nut(s) slot for the set screw, motor shaft diameter, pulley height and belt retainers.

This should allow experiments with belts that are better suited to linear motion, as timing belt like T5 and XL are not ideal; see [reprap.org]

Some of the detail of the smaller pulley sizes will be beyond current FDM printer's ability, but powder bed and laser sintering printers should be able to use the detail to make a better-fitting pulley. In fact, I'd really like to know if it does! I only have T5 and AT5 belt, so I have not tested the other belt profiles for fit. T5 and AT5 do fit very nicely though.

I'd appreciate any corrections, comment or feedback.

I think the reproduction aspect of RepRap is in a difficult place with pulleys. We want to print them, but the tooth profiles that suit the diameter pulleys we want to use are too fine for easy FFF. I've been very happy with my GT2 pulleys from Stock Drive Products, but I don't see ever printing them. I wonder if even powder bed or laser sintering can produce a pulley that fits a GT2 belt really well. Do you know of a larger profile more suited to FFF that also performs well with reversing loads?

Has anyone tried modifying the pulley teeth for T5 to carry the belt by the tooth flanks instead of the root or crest? It seems that should eliminate lash.

I don't see any backlash with the T5 pulleys I have printed, metal T5 or metal T2.5, only with the metal ones from Hong Kong which have teeth that are thinner than the gaps in the belt.

I also don't notice any quality difference between T5 and T2.5 or metal and printed. Until somebody produces some measurements to show a discernible difference I am sticking with what is easiest to obtain.

---

[www.hydraraptor.blogspot.com]

I'm with nophead on the whole, I don't think there is much difference in the belt choice as repraps don't overly load the belts. The important thing seems to be getting the hole nice and central for the motor spindle, so there isn't an eccentric movement. After that, I can see that as the speed increases that people drive their repraps, eventually backlash will be revealed unless fit is very good, and timing belt type teeth profile will move around before HTD or GT2 belt will. Having more teeth in contact will help stop movement, which means using smaller tooth sizes or larger numbers of teeth.

My personal preference is slightly larger pulleys. I use 10-tooth T5 pulleys, simply because I find the fit a little better. I think the ideal is probably GT2 3mm belt: teeth big enough to be printable, designed to be used for linear motion, and plenty of teeth in contact with the pulley, but not so small that a loose belt jumps.

@Dale - it's the outside diameter of the pulley that defines the tooth spacing, and holds the shape of the belt. The belt is designed to be supported on the top and bottom parts of the tooth, while the ramps transmit the force along the length of the belt. By running the belt with it supported by the ramps, you will quickly wear the pulley and the belt tooth.

Ultimately, the openscad script is intended to give people the choice of belt for what they have available, and is not limited to reprap-type applications.

The parametric models you have produced will help many people I am sure.

I have been doing a few experiments with printing timing pulleys and I have found that simply trying to duplicate a standard metal pulley in ABS doesn't seem to work that well. I think it is probably due to the variation you get across the tooth pitch when printing on a reprap and also the slight rounding you get at the corners when changing direction. I basically developed my own tooth form which has slightly narrower teeth, but is suited to my machine settings and the belt I am using (Brecoflex).

Here is an image of my timing pulley

You may find that not all T5 forms are the same as there are subtle differences in the way the belt teeth are profiled between manufacturers, so you may be unlucky that the pulleys do not match the belts and you get some backlash on pulleys with low numbers of teeth.

Your parametric models are good because at least people will be able to fine tune their prints so they can get a good mesh with the belts they have.

I am pretty sure T5 dictates the tooth profile, so it shouldn't vary between manufacturers.

That picture looks like the teeth are a fraction too narrow, bordering on the giving backlash.

How does more teeth help? They all have the same profile, so if the profile gives backlash the only benefit of more teeth is more friction resisting it.

I have found that to print decent T5 pulleys I need a 0.35 nozzle. I used to use a 0.4 nozzle and it made teeth a bit too rounded. That is with the same g-code and layer height: 0.3mm.

---

[www.hydraraptor.blogspot.com]

nophead said
"How does more teeth help? They all have the same profile, so if the profile gives backlash the only benefit of more teeth is more friction resisting it."

Unfortunately in the real world when producing any part there is a natural variation. These most often form a normal distribution around the mean value. This means that the teeth will vary in thickness around the pulley. Having the belt over more teeth will reduce backlash as over more teeth the errors are averaged out. This is the main reason why there is less backlash in timing belts as opposed to spur gears. You have to allow backlash in spur gears otherwise the gear jams when a wider tooth is encountered.

Friction is independant of area, so you your friction comment is incorrect. Friction is dependant on frictional coeficcient and force (in this case it will be belt tension applying the force). You should also consider the increased force required to bend the belt around a tighter curve.You will get less backlash if you tighten the belt as the friction increases and the belt is less likely to slide across the teeth.

The T5 profile is shown in DIN7721, but I have a few belts at home where the corner radii vary between suppliers, which I don't think is covered in the standard. Some of them also have a different shaped notch between the teeth which makes them sit in the timing gear slightly differently.

Edited 2 time(s). Last edit at 02/05/2012 05:54AM by martinprice2004.

The standard seems to define the crest and the angle of the flanks and the pitch so all belts should have the same size teeth. I think theoretically they shouldn't have corner radii and metal pulleys appear not to, they have sharp corners.

Belts are fundamentally different from spur gears because they are compliant. As long as the pulley teeth are slightly too big, and not too small, you don't get backlash.

In practice the printed ones I use don't give backlash with 8 teeth. I suppose that if they were printed with incorrectly calibrated filament width the teeth could come out undersized. I got the tooth profile from the original Mendel STL. I assume it is correct, but I don't know how it was arrived at.

I haven't tried the ones from the Prusa repository as I stick with what I know works. Having just looked at the original Prusa STL I can see they are not good because the inner circle of the pulley, i.e. the base of the troughs doesn't have many facets, making it lumpy. I assume it should actually be straight lines between the teeth and not an arc anyway. Maybe this is why printed pulleys get dissed so much.

Edited 1 time(s). Last edit at 02/05/2012 07:37AM by nophead.

---

[www.hydraraptor.blogspot.com]

I print using a 0.5mm nozzle at 0.5mm layer height, so my gear form is not the best, but the image I posted seems to work OK and I cannot feel any backlash by hand so far. I haven't run my SCARA robot yet, so I cannot say for sure whether they will be OK when running around an 8 tooth motor drive but I am more confident the 40 tooth pulley should be OK. At least if I can print the large pulley and use a metal motor pulley it will keep the costs down.

No promoting intended in the following link, except it may have something on tooth gap profiles. I think at one extreme, the bigger gap is more important with increased/relative high loads and friction, and on the other extreme, the "zero" gap is more important for positional systems which typically have very light loads.

The standard itself is centered along geometry of design therefore cant cover different applications specifics. So at this point differences like these may come along to make these differences.

[www.brecoflex.com]

That being said, generally in power transmission e.g. in Kw - range, should stay out of reduced play because it potentially wears down the belt extremely fast. Reason for higher power std belts to have round profile not trapeze, better spreading the friction onto the semi-circle. But for us, reprap is not "power transmission" per se, but its rather "positional" transmission, so reduced play or no play fits right up reprap alley.

The above is theory. Generally speaking, the aluminium pulleys most ppls use may be reduced play already, as i think technical documentation doesnt usually specify. I think the reason is for how they are made industrially. Its a very special machine which has a intricate cutter tool with lots of individual teeths. This is probably the most expensive cutting tool in industry, and that costs regularly like from 1 to 5+k eur (and more crazy price if its cobalt), and on the other hand pulleys are cheap at factory. Thus makes sense to let the cutting tool a little "over wear" before replacing, which may rather fit "reduced play" even if its specified or not. Especially these pulleys are aluminium and low teeth number, for these reasons they are not likely to get employed in a system of significant power.

Edited 1 time(s). Last edit at 02/06/2012 09:57PM by NoobMan.

As far as I was aware, timing gears are not cut in industry as that would be too expensive. There are two main methods of manufacture.

1) Sintering. Pressed from a powder and then "fired" in a kiln. Its cheap for large volumes, but a very restrictive process. Its difficult to produce flanges at the ends due to variations in powder density, so is generally limited to one flange or no flange designs. Used mainly for steel timing gears.

2) Extrusion. The centre part is extruded through a die in long lengths, then cut off and machined. The flanges are pressed separately and then crimped on. These are the ones we normally encounter in the reprap community. Generally for aluminium timing gears.

This gets me thinking that a way to produce gears would be to make a blank and then form the teeth by rolling a purchased steel gear into them whilst warming the plastic blank.

As an interim process, maybe the reprap could produce a rough tooth form, then by rolling it with a warmed steel gear, the teeth could be more accurately sized.

Edited 1 time(s). Last edit at 02/07/2012 05:33AM by martinprice2004.

droftarts Wrote:
-------------------------------------------------------
> My personal preference is slightly larger pulleys.
> I use 10-tooth T5 pulleys, simply because I find
> the fit a little better. I think the ideal is
> probably GT2 3mm belt: teeth big enough to be
> printable, designed to be used for linear motion,
> and plenty of teeth in contact with the pulley,
> but not so small that a loose belt jumps.

From reading around various producers/resellers docs, I tend to believe that only the Trapezoidal teeth pulley (T series and their non-metric cousins) have a backlash problems.

So all others profiles should be good regarding backlash : AT, GT, STD, HTD etc...

My question is : did anybody try some non-GT among those ? Have we compared the costs and pulley printability of those different profiles ?

---

Most of my technical comments should be correct, but is THIS one ?
Anyway, as a rule of thumb, always double check what people write.

DeuxVis Wrote:
-------------------------------------------------------
> From reading around various producers/resellers
> docs, I tend to believe that only the Trapezoidal
> teeth pulley (T series and their non-metric
> cousins) have a backlash problems.

There are pulleys with different tooth profiles for different uses even for T-series belts. See here: Tooth Gap Design. The zero tooth gap profile shouldn't have backlash even for T-belts.

Same site gives minimum pulley diameters. For T5 belts, 10-tooth pulleys and 30mm diameter idlers are smallest permitted. So even the 608 idlers are smaller than permitted. My printer (Orca v0.2) runs the belt over 13mm idlers and eats so much T5 belts that they could be classed as consumables. T2.5 belts allow 15 tooth pulleys and 15mm idlers, so they should last longer.

DeuxVis, I think everyone jumped on GT2 as there are metal pulleys available the correct size, and it's a newer and slightly better tooth profile than HTD, and available in a smaller size (2mm as opposed to 3mm). I have a friend who has quite a lot of AT5 belt, and was keen to use it, but at the size of pulley we need the pulley teeth end up very thin, possibly weak.

I didn't do the STD profile, as I thought it was close enough to HTD, though having read this I'm not too sure: [www.pfeiferindustries.com]
I'll add it if anyone wants it, and if I can find a dxf file of the tooth.

The section in the scad file that allows for tooth fit should allow for the fit described in ttsalo's link - if you really want backlash! The important figure is the pulley diameter, as it defines the fit of the teeth. I printed out a couple of T5 pulleys the other day, and I have to say tooth fit was very good. Unfortunately, they are too short to be usable as my electronics kept on resetting. I was having a lot of problems with my electronics, which ended up with me destroying my Gen6 board, so I can't print any more! Time for some different electronics...

The other thing I noticed was the tooth form got better and better the slower I went: I ended up printing using 1 perimeter, 100% fill, 0.3 layers, 0.5mm nozzle, 1.7 width over thickness, 20mm/s perimeter, 40mm/s fill.

GT2 timing belts are nice, but here in the UK they are not readily available as far as I can tell. You would have to phone up industrial suppliers to get them.

I like T5 as you can get them easily by shopping online and can have looped belts of any length made up. There are many suppliers. All the T series belts seem easier to get here.

Are they readily available in other countries?

Hi, and thank you for this great resource!

Question: when I try to build an 180 tooth MXL pulley into OpenSCAD, the program gets to around 98% and then seems to freeze there, never advancing anymore... Is this simply a matter of waiting? Is there any other way I could get the teeth profile of my needed pulley?

Thank you,
Eros

I would love the get the STD profile too :-) STD S2M and S3M

Thanks

@erosnicolau: hi, thanks for liking my pulley! One year on and it's still very popular. I've never had a pulley fail before, but they can take a long time to generate. There may be a limit in Openscad with the number of faces it can handle. The tooth shapes are quite 'high resolution', so it may be necessary to edit the number of points in the tooth to make it simpler. I'll take a look tomorrow, let you know what to change.

@Kenzu: if you can point me to some CAD data for the STD tooth profile, and data on pulley sizes, I'll update the pulley script.

@droftarts My google fu is not good enoght :-( Can't find any cad files for STD profiles.

@erosnicolau: I built an 180 tooth MXL pulley, though on my core i5 Mac Book Pro, it took 27 minutes to render! It can be simplified to be quicker; under '// Tooth profile modules' (about halfway through the code) replace

module MXL()
	{
	linear_extrude(height=pulley_t_ht+2) polygon([[-0.660421,-0.5],[-0.660421,0],[-0.621898,0.006033],[-0.587714,0.023037],[-0.560056,0.049424],[-0.541182,0.083609],[-0.417357,0.424392],[-0.398413,0.458752],[-0.370649,0.48514],[-0.336324,0.502074],[-0.297744,0.508035],[0.297744,0.508035],[0.336268,0.502074],[0.370452,0.48514],[0.39811,0.458752],[0.416983,0.424392],[0.540808,0.083609],[0.559752,0.049424],[0.587516,0.023037],[0.621841,0.006033],[0.660421,0],[0.660421,-0.5]]);
	}

with

module MXL()
	{
	linear_extrude(height=pulley_t_ht+2) polygon([[-0.660421,-0.5],[-0.660421,0],[-0.587714,0.023037],[-0.541182,0.083609],[-0.417357,0.424392],[-0.370649,0.48514],[-0.297744,0.508035],[0.297744,0.508035],[0.370452,0.48514],[0.416983,0.424392],[0.540808,0.083609],[0.587516,0.023037],[0.660421,0],[0.660421,-0.5]]);
	}

This still took 16 minutes to render.

@Kenzu: I think I had the same problem; I couldn't find a good STD CAD profile.

I am the IT manager at B&B Manufacturing, Inc. www.bbman.com

Being a "pulley guy" interested in reprap, I thought I would add some information to this dialog.

First, I would like to thank droftarts for the post on Thingiverse. The information contained there is precise and I am sure is helpful to the reprap community.

One of the other issues that I would like to address is how pulleys are manufactured. There are indeed some companies that make sintered metal pulleys and there are some companies that use extrusion. However, it is more common for the pulley to be "cut" or "cast". At B&B, we do not extrude or use sintered metal because of quality issues.

Also, I would like to point out that there are many companies that sell directly to end users, including B&B.

I have been looking at 3D printers for awhile now as a potential pulley prototype producer. I have been quoted some 3D "metal" printers, but so far the cost is not worth the investment.

Hello

I’m new to this forum.
Being from Norway, please excuse my bad grammar and spelling, as English is not my native languish.

I'm trying to figure out a H-bot belt system, but there are really to many choices’ on belts an pulleys.

Can I please ask some questions:
- what are the preferred pulley profile for an H-bot and what belt type would one use?
- how many tooth’s would one use to get an acceptable resolution and speed?

When aligning the pulleys and idlers to get the belts 100% parallel and perpendicular to each other, what would be the offset to use between the pulley and idler, let say for a 3x6 mm belt, would it be the belt thickness minus the teeth dept? It would engage the teeth side of the belt on the pulley, but the backside of the belt on the idler.

How tight can the H be before the belts starting to jam?

Oh and one more thing, does anyone have a link or name to a manufacturer of pulleys where belts can go at different Z-levels on the same pulley, so one can have one short belt from the motor being perpendicular to the main belt(this is not for a H-bot)?

Regards
Terje

Thank you so much!
Eros