PDA

View Full Version : Hand Ergometer



irving2008
23-06-2013, 05:36 PM
Not yet a project but one I hope to complete with, unfortunately, some considerable assistance as I am unlikely to have access to, let alone be able to use, my own machines for at least a year.

(for those that don't know see this thread (http://www.mycncuk.com/forums/general-discussion/5327-new-life-irving-reborn-after-near-fatal-accident.html))

Anyway I need to exercise regularly from my wheelchair, both for cardio vascular and arm muscle development and right now i get one 20min session a week. A hand ergometer - basically an exercise bike for the arms - is expensive, over 2000 new. Even a very old refurbished unit goes for over 700 on eBay and are physically massive as most do legs too (no use to me and have large flywheels), so I thought I'd look at designing my own.

Basically an ergometer is a calibrated load which can be adjusted. Because arms are a lot weaker than legs modifying an existing exercise bike isn't practical, nor is using a modified bike trainer resistance unit (the resistances are too large, the gearing all wrong and the load adjustment too coarse). I had hoped to use a stepper motor with DC energised windings but some experiments by Jonathan showed that probably wasn't going to work either.

The design I have evolved is based on an eddy current brake, basically an aluminium rotor passing through a magnetic field. The field is generated by 4 electromagnets (50kg/24v pickup units, about 12 each on eBay) mounted on a cast iron cage so that the resistance can be varied by changing the current through the magnets. The cage is mounted on bearings concentric to the axle the hand cranks act on and is allowed to rotate a limited amount as it reacts to the torque applied to the rotor. A load cell measures this torque (<10 on eBay).

The rotation speed of the rotor, nominally 60rpm, is measured with a photosensor and speed, torque and magnet current is all fed to an Arduino computer which then calculates the actual load in watts being applied and displays on a small TFT colour display. It will also provide time info and control the magnet current to vary the gear/terrain for interval training (none of the ones I use at the gym do this for the hand section, tho it is common for leg training units). Large button switches will change mode/settings/etc using a soft menu (touch screen is no good unless I have a stylus).

Attached are some parts drawings for comment/discussion. Unless spec'd all parts are 6061 ali or similar. They are not final, in particular I am going to simplify the handle design/assembly and I've not done the load cell or Arduino mounting, or the floor mount arrangement. They were drawn using Inard CAD Pro for Android tablets For 6 its not a bad package and exports PNG, PDF and DXF.

GEOFFREY
23-06-2013, 09:28 PM
Irving, I just don't know how you do it. Unbelieveable - keep it up. G.

Jonathan
23-06-2013, 10:05 PM
I had hoped to use a stepper motor with DC energised windings but some experiments by Jonathan showed that probably wasn't going to work either.

I think a DC motor could be made to work quite easily, particularly if it's separately excited. There are quite a few available second hand, the problem is people like using them to make (not very good) wind turbines, so the prices on eBay are inflated. Using eddy currents seems more elegant, but the DC motor does give the option of using the energy generated for something useful - e.g. powering laptop. From a quick google search it looks like we can expect about 30-60W from two arms.


The cage is mounted on bearings concentric to the axle the hand cranks act on and is allowed to rotate a limited amount as it reacts to the torque applied to the rotor. A load cell measures this torque (<10 on eBay).

I recently needed some load cells for a project where I need to measure torque, so I got some of the cheapest 20kg 'hanging scales' on eBay and dismantled them. I've not actually made the circuit yet, but it looks like it wont be difficult to use the load cells they contain. That makes is <4 for the load cell.


The rotation speed of the rotor, nominally 60rpm, is measured with a photosensor and speed, torque and magnet current is all fed to an Arduino computer which then calculates the actual load in watts being applied and displays on a small TFT colour display. It will also provide time info and control the magnet current to vary the gear/terrain for interval training

It looks like the radius of your disc at the center of the magnets is about 75mm, so at 60rpm that's 0.5m/s. You're going to need a high flux density in the aluminium disc to generate much force due to eddy currents at that low speed. It's annoying that there's not much information about the 500N electromagnets on eBay. I think we can work out the flux density at the pole face required to lift 50kg, estimate the reluctance of the magnetic circuit, then use that to work out the torque/speed characteristic. Another approach is I could model in FEMM (http://www.femm.info/wiki/HomePage), using the dimensions of the magnet and making an educated guess for the rest. Either way I suspect you might need either some sort of gearing between the handles and disc, a bigger disc, copper disc, or more electromagnets.

I've got plenty of permanent magnets you're welcome to which would probably be suitable, although it would mean changing the design somewhat since you'd have to move the mangets to control the torque.

irving2008
24-06-2013, 12:05 AM
Hi Jonathan

I was half expecting you to investigate further :)

The radius is 90mm and there are 4 magnets. Fully energised they each generate 1.6 Tesla (core saturated magnetically) at the pole which is 20mm dia. I don't have the B-H curve (as you found, no data sheets online, but then this isn't their intended use) but I'm hoping to be able to calibrate it. The air gap to the rotor is 1mm. I kept it this size as think getting it smaller will require too much precision in mountings.

Now for some calcs....

torque = sigma * omega * B ^ 2 * p ^ 2 * a * d
sigma=rotor conductivity
omega=speed rads/sec
B = mag field Tesla
p = effective radius
a = pole area
d = disc thickness

sigma Al 6061 = 3.5e7
omega =6.3 rad/sec (60rpm)
p = .09m (200mm dia rotor, 20mm mag)
a = 0.01m^2*pi = 3.1416e-4
d=0.005m

using an online calculator whose URL I cant find right now the 4 magnets @ 1.6T generate a total of 1.36T yoked in a 7mm air gap. Therefore

max torque = sigma * omega * B ^ 2 * p ^ 2 * a * d
= 3.5e7 * 6.3 * 1.36^2 * .09^2 * 3.1416e-4 * .005
= 5.2Nm

Power= torque * omega = 5.2 * 6.3 = 32.8W

So will need to increase # of magnets to 8 x 40kg, or go to 40mm dia pole or increase rotor dia to 300mm to get to 60W. Also could consider some sort of epicyclic gear to rotate rotor faster e.g. 3 or 4 times but maintain it concentric to axle to minimise footprint.

edit: 27mm dia pole (70kg magnet) and 250mm rotor gives 91W

I did look at permanent magnets but the mechanical complexity of moving them and still measuring torque accurately got very messy (mechanical sizes were based on 3 pairs of 20mm x 10mm Neo magnets, I'd not recalculated based on electromagnets, my bad)

irving2008
24-06-2013, 08:02 PM
Been looking for plastic internal (annulus) gear rings, to create an epicyclic gear train, but can't find a UK supplier. Any ideas?

Alternatives are a timing belt to drive the rotor at, say, 5x crank speed so can use smaller lighter magnets but that adds equal complexity

FatFreddie
24-06-2013, 11:36 PM
Been looking for plastic internal (annulus) gear rings, to create an epicyclic gear train, but can't find a UK supplier. Any ideas?


Is this something that could be 3D printed? Not sure what sort of strength / wear characteristics you'll need...

I'm happy to have a go if you want to try it - max size is about 200 x 200 by 100 mm deep though print time / warping might be a problem if using the full envelope. I've got plenty of PLA in - not tried ABS or nylon yet but willing to experiment.

irving2008
25-06-2013, 01:23 AM
It could be 3D printed, in fact I was already thinking along those lines. Envelope would be around 80mm square x 6mm thick. I'd need the properties of PLA to check if its strong enough. What's the resolution of your printer?

Clive S
25-06-2013, 09:11 AM
I have printed with both PLA and ABS and not sure they would be strong enough but laser printed I suppose could be done even if it had to be in two 3mm pieces and bonded back together. Do you have some sort of drawing! ...Clive

FatFreddie
25-06-2013, 11:48 AM
It could be 3D printed, in fact I was already thinking along those lines. Envelope would be around 80mm square x 6mm thick. I'd need the properties of PLA to check if its strong enough. What's the resolution of your printer?
That size wouldn't be a problem, resolution is quoted at 0.0125mm and accuracy at 0.1mm (Mono Mendel - RepRapPro (http://reprappro.com/Mono_Mendel)). The current nozzle is 0.5 mm (best used with a 0.4 mm or slightly smaller filament thickness) but I've got a 0.35 mm one that just needs putting together for finer detail.

Some material specs - 9129

Interesting PDF on designing for FDM (http://ode11.com/publications/sme_rp_2001.pdf).

Because there is variable bond strength between the filaments, it's best to design so that the stresses are applied along the filaments but for things like gears that happens anyway since the slicer will trace round the perimeters.

We could do strength tests before you commit to the design...

irving2008
25-06-2013, 10:05 PM
Trying to work out how to draw the teeth for this in my CAD program, but attached is an initial sketch of the epicyclic gears.

9131

Current thinking is I basically want a mod 1.25 60 tooth internal ring with a couple of fixing lugs (the ring is static), a mod 1.25 30 tooth sun spur gear on a press fit 10 x 22 x 6mm bearing (this will be fixed to the rotor which will rotate freely on the 10mm axle using another 10 x 22 x 6 bearing), and 2 off mod 1.25 15mm planetary spur gears on press fit 3 x 9 x 5mm bearings (these will be fixed to a carrier thats connected to the 10mm axle ). Spinning the 10mm axle will spin the rotor at 1+60/30 = 3 times the rate or 180rpm. the torque in the gear train is around 6Nm

Clive S
25-06-2013, 11:37 PM
Just found this to get some ideas Plastic Gears (http://www.formtechservices.com/formtechservices.com/Plastic_Gears.html) ...Clive

FatFreddie
26-06-2013, 11:47 AM
Trying to work out how to draw the teeth for this in my CAD program, but attached is an initial sketch of the epicyclic gears.

Current thinking is I basically want a mod 1.25 60 tooth internal ring with a couple of fixing lugs (the ring is static), a mod 1.25 30 tooth sun spur gear on a press fit 10 x 22 x 6mm bearing (this will be fixed to the rotor which will rotate freely on the 10mm axle using another 10 x 22 x 6 bearing), and 2 off mod 1.25 15mm planetary spur gears on press fit 3 x 9 x 5mm bearings (these will be fixed to a carrier thats connected to the 10mm axle ). Spinning the 10mm axle will spin the rotor at 1+60/30 = 3 times the rate or 180rpm. the torque in the gear train is around 6Nm

That looks ok for printing - probably best if I get the finer extruder going but I've made some gears about that size which work ok. Could the unit be thicker? I'm thinking that it's maybe a bit marginal on strength - probably easiest to just build it and test it than worry too much at this point though.

If you (or anyone else) can point me towards a gear profile calculator (can't find one that does the outer ring) I'll knock up something for a strength test.

irving2008
27-06-2013, 12:03 PM
So finally managed to draw tooth profile and rotate/replicate it round a circle... Inard CAD doesn't have an easy way to do this, but what do you expect for 6!

9156915591589157

Clive S
27-06-2013, 12:28 PM
Irving Can't seem to open that link. ....Clive

irving2008
27-06-2013, 01:05 PM
Strength calcs: Calculate for input gear as worst case (Ref: Spur Gears (http://www.roymech.co.uk/Useful_Tables/Drive/Gears.html))

Input torque = power/revs = 90/6 = 15Nm, add 50% for safety = 22.5Nm
Tangential force on tooth Ft = 2 * torque/PCD = 2 * 22.5/.01875 = 2400N

Bending stress = K * Ft/(b * m * Y)
K=velocity factor, estimated at 1.03
b = face width = 6mm
m = module = 1.25
Y = Lewis Form Factor = from tables, for 15T 20deg gear = .289 @ tip .49 @ centre of tooth, use .289

stress = 1.03 * 2400/( 6 * 1.25 * .289) = 1140MPa = ~165kPSI

PLA = 500kPSI so should be OK!

irving2008
27-06-2013, 01:23 PM
Irving Can't seem to open that link. ....Clive

should be fixed now...

FatFreddie
27-06-2013, 01:52 PM
So finally managed to draw tooth profile and rotate/replicate it round a circle... Inard CAD doesn't have an easy way to do this, but what do you expect for 6!

Can you post those as DXF files?

Must give the CAD program a try - looks good.

irving2008
27-06-2013, 04:17 PM
Can you post those as DXF files?

Must give the CAD program a try - looks good.
Yes its not bad, fairly intuitive, and seems to be the only one that works properly on an Android tablet. I tried a few others but not much good.

Jonathan
27-06-2013, 07:54 PM
I'd go for a timing belt drive for three reasons:

It's not unlikely that we might find the ratio needs changing and to change the ratio you only need to change one pulley compared to two gears.
The torque ripple will also be less with a belt, unless maybe someone wants to have a go at making helical gears!
Noise.

irving2008
27-06-2013, 08:41 PM
I'd go for a timing belt drive for three reasons:

It's not unlikely that we might find the ratio needs changing and to change the ratio you only need to change one pulley compared to two gears.
The torque ripple will also be less with a belt, unless maybe someone wants to have a go at making helical gears!
Noise.

1. I suppose its possible it might need changing, but a 3:1 ratio means 180rpm on the rotor and thats more than enough, can even use smaller/fewer magnets as torque requirement reduced by same ratio.
2. Not sure what you mean by torque ripple in this context, its an artifact of a motor due to bad pole design aka cogging...
3. Maybe, but plastic gears shouldn't be that noisy at these low speeds...

Epicyclic is elegant solution to problem. Belt drive would need to find a way to route belt past magnetic pole piece each end of magnet carrier, or carrier would need to be vertical or at least at an angle making use of the U-channel mounting more difficult and making the mounting of the loadcell more complex. Also belt drive might increase the footprint.

Jonathan
27-06-2013, 10:07 PM
1. I suppose its possible it might need changing, but a 3:1 ratio means 180rpm on the rotor and thats more than enough, can even use smaller/fewer magnets as torque requirement reduced by same ratio.
2. Not sure what you mean by torque ripple in this context, its an artifact of a motor due to bad pole design aka cogging...
3. Maybe, but plastic gears shouldn't be that noisy at these low speeds...

Epicyclic is elegant solution to problem. Belt drive would need to find a way to route belt past magnetic pole piece each end of magnet carrier, or carrier would need to be vertical or at least at an angle making use of the U-channel mounting more difficult and making the mounting of the loadcell more complex. Also belt drive might increase the footprint.

1. It depends how well we need to control the torque, i.e. the resolution. It the ratio is wrong, so the magnets are never operated near saturation, then you've reduced the available range in which to control them. I guess the ratio would have to be a long way off for it to be significant in this application...
2. I wouldn't say it's due to bad design, more that it's a side effect of optimising a motor design for certain characteristics. In this context I meant simply that the handles will not feel as smooth to turn - there will be vibrations at frequencies dependent on the number of teeth on the gears.
3. True, I'd forgotten it was so low.

One of these would be fun to watch:

https://www.youtube.com/watch?v=35JrvpcKKTo

irving2008
28-06-2013, 01:53 AM
1. Surely you've got that backwards? The magnets are intended to be used at saturation in pick-up mode. Here we will run them well below saturation, hopefully somewhere on the linear-ish part of the B-I curve. I'm only using the saturated value of 1.6T to find the max possible retardation.

2. Ok I get you now. At the nominal design speed of 60rpm I guess there will be 'vibrations' at 90Hz from the planet/sun wheel interface and 60Hz from the ring/planet interface. I expect these will be small - I guess they'll depend on the quality of the tooth shape and the fit of the gears. We won't know til we try! :)

FatFreddie
28-06-2013, 11:49 AM
We won't know til we try! :)

Current state of trying is an interference fit :-(

I printed some calibration parts and they came out fine so I need to look at it in more detail.

I suspect that the plastic will reduce vibration to some extent but as you say, trying it is the real test.

irving2008
28-06-2013, 12:42 PM
Current state of trying is an interference fit :-(

I printed some calibration parts and they came out fine so I need to look at it in more detail.

I suspect that the plastic will reduce vibration to some extent but as you say, trying it is the real test.

Does that mean you've printed them? :)

I can't guarantee the accuracy of the drawings, I use the MOD1 profile found on the web then scaled it up 1.25 to get a MOD1.25 profile but the way I had to rotate it to get the final gear may be over/undersize. I suspect its the ring gear thats slightly under... maybe you could measure the OD and ID of the gears and let me know?

FatFreddie
28-06-2013, 01:05 PM
I just printed some 1mm thick test pieces.

I loaded the 15 and 30 tooth gears into cambam and generated a profile with the built in generator and it was exactly the same outer and root diameters (just a bit more lateral clearance in the root) so I suspect you're right about the ring gear - unfortunately cambam doesn't do ring gears (I've bumped a thread asking for them).

Measured OD on the 30t gear is 40.0mm and 21.3mm on the 15t and the tooth tip diameter on the ring gear is 72.7mm and the root diameter is 76.4 (near as I can tell - it's a bit flimsy)

irving2008
28-06-2013, 03:58 PM
I just printed some 1mm thick test pieces.

I loaded the 15 and 30 tooth gears into cambam and generated a profile with the built in generator and it was exactly the same outer and root diameters (just a bit more lateral clearance in the root) so I suspect you're right about the ring gear - unfortunately cambam doesn't do ring gears (I've bumped a thread asking for them).

Measured OD on the 30t gear is 40.0mm and 21.3mm on the 15t and the tooth tip diameter on the ring gear is 72.7mm and the root diameter is 76.4 (near as I can tell - it's a bit flimsy)

OD for MOD1.25 = (N+2)*1.25, so 30T = 40mm and 15T = 21.25mm.

tooth tip dia on ring gear should be (N-2)*MOD = 58 *1.25 = 72.5mm and root dia 78.125mm which is correct on drawing...

FatFreddie
28-06-2013, 05:25 PM
Ok, looks like it's a problem with printing the internal features - let me have a look at it over the weekend - trying to learn Hadoop (if you don't know, you probably don't want to) at the moment and it's not conducive to concentrating on other stuff :-(

FatFreddie
01-07-2013, 11:32 AM
Ok, some progress. I've now got a set of gears that mesh reasonably well but I'm not happy with the print quality so I'm going to try a different slicer - the one I'm using at the moment is Sli3er and it's fast and easy to use but it doesn't seem to be the best quality. I was using Skeinforge before which is better quality output but not very friendly - I'll have another go with that over the next couple of days.

irving2008
01-07-2013, 12:46 PM
Ok, some progress. I've now got a set of gears that mesh reasonably well but I'm not happy with the print quality so I'm going to try a different slicer - the one I'm using at the moment is Sli3er and it's fast and easy to use but it doesn't seem to be the best quality. I was using Skeinforge before which is better quality output but not very friendly - I'll have another go with that over the next couple of days.
Thanks for your efforts on this. I'm in the process of drawing up slightly modified gears, the main difference being improvements to better support/locate the bearings and space the sun gear off the rotor boss - I may increase the thickness of the gears slightly. I'm also redrawing the supports, the rotor bosses and the planet carrier. I was thinking of 3D printing the carrier but I don't think the boss would be strong enough to take a grub screw to fix it to the axle (but what if the boss was thicker and had a slot for a nut to be inserted?). The other thought I had was 3D printing other components, the rotor bosses for example. I'm not sure about the cranks or the handles from a strength perspective, need to run the numbers. Also do you have any way to calculate the cost of a printed component?

FatFreddie
01-07-2013, 03:04 PM
No problem, I'm enjoying having something constructive to do with the machine. Work is a bit binary at the moment so apologies if I go missing occasionally.

I had thought of printing some of the other parts - once I get something that can be strength tested we'll know better what's possible. I have had some unexpected breakages in the past but that was with mounting brackets where I think I underestimated the shock loads. One thing to bear in mind is that things like splines which would normally be difficult are relatively easy with 3D printing. The main problems are fine tolerances and overhangs so a spline would be best if it was tapered or with pinch bolt.

I haven't bought any filament for a while but I think it's around 30 a kilo. Partially hollow parts are the norm so material usage is quite efficient.

irving2008
01-07-2013, 05:38 PM
No problem, I'm enjoying having something constructive to do with the machine. Work is a bit binary at the moment so apologies if I go missing occasionally.

I had thought of printing some of the other parts - once I get something that can be strength tested we'll know better what's possible. I have had some unexpected breakages in the past but that was with mounting brackets where I think I underestimated the shock loads. One thing to bear in mind is that things like splines which would normally be difficult are relatively easy with 3D printing. The main problems are fine tolerances and overhangs so a spline would be best if it was tapered or with pinch bolt.

I haven't bought any filament for a while but I think it's around 30 a kilo. Partially hollow parts are the norm so material usage is quite efficient.


OK. Attached are image and DXF file for the 3D printable modded sun gear and rotor boss which now include a seat for a 15 x 10 x 4 magneto bearing. these go either side of the rotor and are joined by 25mm M3 cap headed bolts and captive nyloc nuts. Are these printable?

Crudely each contains 1.25e-5 cubic metres of material, which means 0.02kg of PLA
9175
9176

FatFreddie
03-07-2013, 10:43 PM
Shouldn't be a problem to print those.

The new slicer is better, both finish and sizing.

I'm trying to put something together for some destructive testing now :-)

Don't worry about the plastic usage, I've got loads in (as long as you're not too bothered about the colour).

irving2008
04-07-2013, 01:22 AM
Once again, thanks for your efforts & interest in this project.

Colour isn't an issue, tho I did read that different colours have slightly different mechanical properties. What colour(s) do you have?

Have new drawing nearly done for ring gear. Had to increase outer rectangle dimensions to make it mountable without machining a big chunk of ally to bolt it to.

Also been considering what else could be printed. Have drawn up planet carrier, the handle cranks and the main bearing supports but need to do some bending calcs to see if they're rigid enough. Will post new drawings tomorrow once completed and sanity checked. Reducing the components that need to be machined reduces the manufacturing costs dramatically.

FatFreddie
05-07-2013, 01:02 PM
I've got a variety of odds and ends - mostly silver, black and red but I've probably got enough fluoro pink to do the 15t gears if necessary ;-)

Can't say I've noticed much difference in strength between the colours but some manufacturers are decidedly worse for inter layer adhesion - I've got some that I only use for early prototypes.

Attached picture shows the (very crude) test rig - current limit is the stiffness of the ring gear - it bows out under pressure and skips. I'm printing one with a thicker outer now.

I've also got some issues with the first layer being oversized (the slicers tend to make the first layer slightly thinner to get good adhesion to the bed by squashing the plastic between the head and the base) which should tune out but which would be very easy to fix if Alibre was man enough to let me put a chamfer on the teeth :-(

9197

The 'feel' of the gears is a bit rough (though this will improve), if it's important for it to feel very smooth, maybe some belt drive arrangement would be better - it needn't be any bigger. Alternatively we could use double helical gears which should print ok as long as the helix angle is fairly shallow. Probably best for me to get something working and send it to you for comments - I'm away this weekend but should be able to do some more next week.

irving2008
05-07-2013, 02:57 PM
Looks good tho :) Are those full thickness?

When you say the ring gear 'bows out under pressure' what exactly do you mean? I've changed the design of that, see attached. It will be bolted to a pair of 10mm sq ally columns which will add considerable rigidity.

9199

FatFreddie
05-07-2013, 07:56 PM
Looks good tho :) Are those full thickness?

When you say the ring gear 'bows out under pressure' what exactly do you mean? I've changed the design of that, see attached. It will be bolted to a pair of 10mm sq ally columns which will add considerable rigidity.

9199

Yes - that's full thickness (6mm).

The majority of the deflection is where the gear is closest to the outer edge of the piece - the straight edges by the gears become convex and the ones on the side away from the gears become concave - it moves enough for the teeth to slip over the tops. Moving to three or four planet gears may be a simple way to alleviate this without having to make the ring gear much stiffer. Your new design will also be somewhat stiffer - is there much room to play with?

irving2008
05-07-2013, 09:05 PM
There's a bit more room, I can go wider and taller and thicker if need be. This new ring gear is 8mm thick and the new gears are also 8mm as I needed more space for bearings & bolt heads.

irving2008
06-07-2013, 01:46 PM
I'm surprised it deflects that much, a quick calculation, crudely treating the side as a 6mm deep 'edge' x 6mm thick beam of length 90mm with fixed ends shows a deflection of .01mm per N of load... so deflecting a whole tooth is a 250N load, larger than expected but not impossible I suppose.

The v2 design (100mm long, 12mm deep 'edge' x 8mm thick) is .0013mm per N, so only 0.32mm deflection.

If need be I could add a surround of 8 x 8mm ally angle to constrain the PLA.

edit I was using the Youngs Modulus of PLA as 3450MPa. That article suggests a 50%+ reduction depending on how the filaments are laid down. The max strength requires cross-bonding, preferably at 45deg. So it could be that the forces are much lower, but the relative calculation above still holds.

FatFreddie
08-07-2013, 01:57 PM
A couple of complicating factors - I forgot to say that the part is printed with 30% infill to save plastic so it's not as stiff as it could be (there are three solid layers on the top and the bottom (1.1mm) and on the perimeters (1.5mm - I'm extruding filament 0.5mm wide and 0.35mm thick)) and the concave bending of the sides away from the planet gear effectively doubles the deflection.

I tried the new ring gear (still 6mm and 30% fill) and it moved but didn't skip (at moderate loads - see below) so I don't think it will be an issue - more something to bear in mind for any future modifications - I'll know more when I've managed a proper load test.

Testing at 2KG/metre resulted in a broken gear carrier and splintering of the wooden batten attached to it so they need to be improved next - I might make an aluminium carrier as I've just converted the milling machine to a CNC-USB controller but haven't cut any metal with it yet :-)

I think I've pretty much tuned out the oversized first layer problem by putting a slight relief on the side of the teeth next to the bed - the 15t gears look good and the 30t isn't far off - same technique should work on the ring gear as well and has improved the feel of the gears quite a lot.

irving2008
08-07-2013, 02:53 PM
Interesting...

I've designed a 2-gear planet carrier, see attached, but whether it'll be stiff enough in PLA to grip the 10mm shaft or not remains to be seen. The slot in the boss is for a nut as you can't put a thread in PLA.

I can go to a 3-gear solution, but a 4-gear will need 64/32/16 tooth gears else it can't be assembled.

9219

FatFreddie
11-07-2013, 12:17 PM
I think you're right - I can't see the PLA being strong enough to take the stress. One possibility might be to have a hole drilled in the shaft and a bolt through the hole so the torque is transferred to the outer skin via the bolt. As an aside, square nuts are best for small captive nuts in PLA - much more resistant to spinning in the slot. M3 are 5.5mm A/F and I've got some in stock, M4 and above are ok in hex.

I've put some proper bearings in and the gears are working much better. There is one tooth on each gear that is slightly malformed and that causes sticking but a couple of minutes sanding eased it considerably. Progress is a bit slow at the moment because I'm trying to get the printer working again after adding some hardware to allow me to print without it being connected to a PC (it keeps dropping the connection which is a pain) - all seems ok apart from it not heating the bed which, unfortunately, is a show stopper. If I can't fix it I'll revert to the old firmware but it's probably something fairly simple.

Your new ring gear design is much stiffer and I don't think skipping will be a problem. Engagement seems fine as long as it's within certain limits but when the engagement zone of the teeth moves to the more angled zone near the tip, the forces increase rapidly causing the skipping so a moderate increase in stiffness (and improved tolerances from the new slicer) has made a big difference.

FatFreddie
18-07-2013, 08:43 PM
Hi Irving,

Sorry I've been quiet - took a while to get the printer going again and I've been busy - the new firmware needed all the accelerations and speeds sorting as well.

Where do we go from here? Two options I think -

1. I've got a very simple set of gears working which I can send you to have a look at (I gave up on the idea of destructively testing them, the originals weren't far off requirements and these are a lot better). PM me an address if you want to do this.

2. I can just build some gears to your spec - should have some time on Tuesday. If you want to do this, could you send me the latest drawings.

Mark

irving2008
22-07-2013, 03:04 PM
Hi Irving,

Sorry I've been quiet - took a while to get the printer going again and I've been busy - the new firmware needed all the accelerations and speeds sorting as well.

Where do we go from here? Two options I think -

1. I've got a very simple set of gears working which I can send you to have a look at (I gave up on the idea of destructively testing them, the originals weren't far off requirements and these are a lot better). PM me an address if you want to do this.

2. I can just build some gears to your spec - should have some time on Tuesday. If you want to do this, could you send me the latest drawings.

Mark

Hi Mark,

Sorry not got back to you sooner, hasn't been a good few days given the heat (people with high-level spinal cord injury can't regulate body temperature easily).

I'm trying to finalise drawings for the gears and have beefed them up by increasing thickness as well as making the ring gear even more substantial to try avoid needing a seperate ally support, but will retain fixings for that in case. Overall assembly needed a rethink as can't get gearbox and magnets on same side of rotor without much more complex magnet carrier so going for magnets on one side only. Fortunately 3 x speed increase plus reduced torque requirement introduced by gearbox means can use smaller magnets anyway. I'll post that up shortly.

Also doing a 3-gear planet carrier to reduce gear tip loading. Going to make the carrier boss more substantial and have 2 grub screws at 90 deg so will give it a go in PLA. Worst case it'll do for testing and I'll get one made in ally later if need be.

Hope I'm not imposing too much.

FatFreddie
23-07-2013, 09:58 PM
Sorry not got back to you sooner, hasn't been a good few days given the heat (people with high-level spinal cord injury can't regulate body temperature easily).

Makes my whingeing about being a bit hot seem rather pathetic.

No rush on any of this from my perspective but I'm away for the middle two weeks in August and will be busy in the run up and I don't want to become the bottleneck.


Also doing a 3-gear planet carrier to reduce gear tip loading. Going to make the carrier boss more substantial and have 2 grub screws at 90 deg so will give it a go in PLA. Worst case it'll do for testing and I'll get one made in ally later if need be.

Worth a try with the carrier - as you say, you can always fall back to ally.


Hope I'm not imposing too much.

Not at all - it's been a very useful exercise and I've improved both my understanding of the process and the printers hardware and software and am now able to produce much better prints more easily.

Mark

irving2008
01-08-2013, 03:24 PM
Not been a great week what with heat and other issues but trying to get back on track.

Been prevaricating over how to retain bearings in the planet gears The revised gears are 10mm thick and the bearings I'm going with are 3 x 9 x 4. The planet carrier has a 3mm boss and with a 3mm caphead screw as an axle the bearing will be centrally located in the gear against a recess. but that doesn't stop the gear sliding off the bearing. Normally you'd have some sort of bearing retainer but the only way I can see to do this with a 3D printed gear is either a groove 'not printed' to take a circlip which means an overhang (do-able but not ideal) or some sort of printed protrusion which will deform enough to allow the bearing to be pressed into place (is this possible?).

Thoughts?

FatFreddie
02-08-2013, 04:15 PM
Been prevaricating over how to retain bearings in the planet gears The revised gears are 10mm thick and the bearings I'm going with are 3 x 9 x 4. The planet carrier has a 3mm boss and with a 3mm caphead screw as an axle the bearing will be centrally located in the gear against a recess. but that doesn't stop the gear sliding off the bearing. Normally you'd have some sort of bearing retainer but the only way I can see to do this with a 3D printed gear is either a groove 'not printed' to take a circlip which means an overhang (do-able but not ideal) or some sort of printed protrusion which will deform enough to allow the bearing to be pressed into place (is this possible?).

Thoughts?

The bearings should be ok on a press fit but a small ring could be glued in place. I've been using 5 x 11 x 4 bearings for the planet gears and they press in quite firmly.

mekanik
11-08-2013, 12:59 PM
Hi M8
Just been having a look @ your project and although you have expended considerable man hours on the prototype, i was wondering if you had given any consideration to using a Shimano/sram geared rear hub ?
Just a thought.
Take care and i sincerely hope your circumstances improve in the near future.
Regards
Mike

mekanik
11-08-2013, 01:16 PM
Shimano Nexus 7 hub outer epicyclic ring gears. | eBay (http://www.ebay.co.uk/itm/Shimano-Nexus-7-hub-outer-epicyclic-ring-gears/251312695277)

Shimano Nexus 7 hub epicyclic gears. | eBay (http://www.ebay.co.uk/itm/Shimano-Nexus-7-hub-epicyclic-gears-/251312693668)

http://www.ebay.co.uk/itm/SRAM-S7-7-speed-internal-gear-hub-with-coaster-brake-shifter-18T-cog-/181187731193

irving2008
11-08-2013, 01:41 PM
Hi M8
Just been having a look @ your project and although you have expended considerable man hours on the prototype, i was wondering if you had given any consideration to using a Shimano/sram geared rear hub ?
Just a thought.
Take care and i sincerely hope your circumstances improve in the near future.
Regards
Mike

Mike,
Great minds etc... it was one of my first ideas. Sadly the maximum ratio on a Nexus hub is 1.5:1 and I need at least 3:1, plus the hub is too wide @ 110mm

regards,
Irving...

mekanik
11-08-2013, 03:08 PM
Would the ring gears on the e-bay links not be better than plastic ?
those listed are cheap enough.
Could you use two spur gears with the correct ratio ?

irving2008
11-08-2013, 06:55 PM
Would the ring gears on the e-bay links not be better than plastic ?
those listed are cheap enough.
Could you use two spur gears with the correct ratio ?
Epicyclic gears have very specific relationships. To get 3:1 the number of teeth on the ring is 2 x teeth on planet + 1 x teeth on sun gear. Without knowing the number of teeth, circular pitch and tooth form of the Nexus ring gear the chances of finding other gears to suit off the shelf is diminishingly small...