TTR Elevator Conveyor - 3D Print New Belts?

10 - Alternative Layer Start/Finish Options

Tried the 'Nearest' start position for a new layer, where you are given the option for seam to be 'None' so slice again.

Arrgh, it still generates a seam of zits, also with 'Random', something here is fibbing!

If I was printing an angular object, that is one with corners the seams can be effectively invisible on those corners, but a thin, one row, continuous loop is another thing.

Well, there seems, pun intended, that there is no way out of this unless I can find the code that generates the seams.

Another option would be to coast the last millimetre or so before the end of a layer.

Coasting

This is when the nozzle keeps moving but the extrusion of plastic stops before the end of a layer, so this would reduce the excess flow when the nozzle stops at the end of a layer, increments up one layer and starts to extrude again.

Tried in increments up to 5mm (which shows a big gap when examined with the slicer pre-view), but makes little difference, except at 5mm I have two 'seam' points, albeit smaller than without coasting but almost twice the nuisance value.

Now What?

But then I said to Google "Cannot change print speed on raise3d e2" and the reply was :-

https://forum.raise3d.com/viewtopic.php?t=1669#:~:text=Re: Print Speed Not changing?&text=Under cooling, minimal layer time,minimum time of 15 seconds.

Profile -> Cooling -> Minimum layer time

Lo and behold, the setting was too short for what I needed.

Settings for print speeds are overriden by the 'cooling minimal layer print time', if they result in a layer print time of less than the specified minimum.

This is set to ensure a layer cools enough before the next layer is applied, with a simple single width layer as this belt, this parameter dominates.

I gradually upped this value to 95 seconds per layer after which the overall print time did not change so another parameter must be taking control, perhaps a minimum speed parameter somewhere.

The seam became smaller as the print speed was reduced, remember from earlier that the 'blob' which becomes a seam as the layers build up is due to the plastic still flowing after the nozzle has momentarily stopped moving horizontally to step up a layer in readiness to start printing the next layer.

So, by lowering the print speed the transition from moving to stopped is smaller as there is less speed to lose, also the plastic flow rate is correspondingly lower. So less plastic oozes out of the nozzle when it stops moving horizontally to step up one row, and hence a smaller bead of plastic is deposited and as the layers build up a smaller cross-section of seam.

The slicer illustrates where this will be on the print preview.

Phew!

This took quite a bit of sorting out.

Partial Nozzle Clogging

Flow issue also became compounded by partial blockages which made segments of the curves look like tear drop shaped stepping stones!

These were cleared by unloading the TPU, loading a length of cleaning filament, 50 to 100mm followed by reloading the TPU.

It took a while to twig it was a partial nozzle clogging issue and not a parameter change knock on effect.

Stringing Across the Belt Path from the Prime Line

The start position for printing the belt is reached from the end of the prime line by a direct movement, which results in a string cross the path of the first layer.

Eventually I have found out how to insert an extra bit of G-code into the start-up section of the code to re-route the extruder such that it misses where the first layer of the belt will be printed. So at least one less bit of stringing to contend with.

This site is useful for those with a desire for brain ache :-

https://reprap.org/wiki/G-code

The image below shows the new path of the extruder, from the end of the prime line at A, via B to C, originally it went directly from A to C crossing the path of the dark red line which is the first layer of the belt.




Skirts

These are rings surrounding your printed object which allow the flow from the nozzle to be stabilised over a longer run-up before starting to print your object.

However there appears to be a bug in the G-code the slicer generates as when the skirt was completed at a higer speed than the actual print job the first layer of the print job was laid at this higher speed and of course gave me a large blob at the start of the 2nd layer.

So, dispensed with the skirt which still leaves a little stringing at the start of the belts first layer, this is minor compared with a 'blob'.

After a few part prints to try and eliminate or at least minimise the seam a full print lasting just over 2 hours was run to completion.

 

11 - Things are looking up

The print of the belt is much more consistent, the reflections are even and the effect on the front edge is due to the slightly concave vertical surface, all very much better than previous attempts.








Extruded at 0.5mm wide for a single pass for each layer, the resulting dimensions (to one decimal place) are :-

Width 20 to 20.2 mm
Belt Thickness 0.5 mm
Seam Thickness 0.7mm​

This is more like it!

Considering the nozzle is 0.4mm , selecting it to extrude at 0.5mm and printing at 0.5mm wide is amazing, the seam is more visible than having any real physical impact.

Light banding is visible across the belt where it is printed around the corners as 30 segments but of no functional issue, did not fit the build plate to print as a circle.

Mind you printing a whole belt as a circular print would show these bands over its complete length even if the number of segments were raised to a much higher quantity.

So how does it fit in the frame as lengths have varied with different print set ups.

Fit

Width: Binds at the chute end at one part of the belt suggesting it is more an issue of the frame, probably better to tweak the frame than enlarge the gap at the belt sides by reducing its width by a print layer or two.

The conveyor belt I have which looks like an original measures about 19.6 - 19.8 mm wide.



With the belt lightly twisted to show its fabric backed nature.





This original belt is in pretty good order, when the tensioners are applied they sit mid range, however the serrated drive roller acts a against fabric which will inevitably wear through so I will be making sure I print a kit of spares for this conveyor along with the parts needed now to get the other well worn conveyer-elevator restored to full functionality.

I am not aiming at full restoration to original condition. One is rusted and chipped and the other had its motor replaced long before I bought it from a seller in the USA. I believe it was re-motored for a shop display, whether the original motor wore out or maybe just so it could be powered from 12V DC as opposed to 14V AC is not known.

Fitting of a 20 mm wide belt was a little opimistic, so a revision down to 19.5mm, (80 layers -> 78) may well be advisable depending on how repeatable the print process is, everything manufactured has some dimesional variation so design tolerances must take that into account.

Length: Runs well without the tensioners applied but when they do they almost run out of slack, suggests the belt needs to be slightly shorter in the event they do require tensioning after some use. So shortening by 4mm is worth trying.

Materials

TPU is a hard wearing and abrasion resistant plastic which is conveniently suitable for use as an operational conveyor belt, it's elasticty is enough to use as a belt with friction drive especially if the drive roller surface is a flexible rubber material such as an elastic cable sleeve.

The use of a tensioner on the conveyor was probably required as the fabric belt probably had very little elasticity, the elevator belt on the other hand was a moulded part, probably of a rubber and had sufficient elasticity not to need a tensioner.

The metal work particularly at the chute end is a little rough and rusty so cleaning that up will be beneficial to the belt edges and reduce motor and gear loading.

So I have a major step forward, I have at last a printing profile which matches the Overture TPU 95A to my R3D E2 printer, hopefully the next print will confirm this, for thin belts anyway!

Revised Belt Dimensions

Conveyor belt dimensions become as shown, corner segment count on the corners increased from 30 to 45, estimated print time 2 hrs :-



Next, print this modified belt.

 

Hi Phil, that is interesting, you could write and post about that in for instance the "Other Hobbies & Interests".

https://platform1mrc.com/p1mrc/index.php?forums/other-hobbies-interests/

They are also scenic items for larger scales too

Jim

 

12 - Hiccups, but then the desired result

Clean out the hot-end again

Here we go, the tear drop paving stones at the start of the print that I mentioned earlier, note it does not affect the prime line or first straight run along the build plate, seems to occur at vertices, left to run further it would repeat this mess at the next corner.



Cleaning Filament

By the way, the cleaning filament I have been using for the last 9 months is:-

eSUN Cleaning Filament 1.75mm, 3D Printer Cleaning Filament, Prevent Nozzles and Extruders Clogging, 100g Spool 3D Printing Filament for 3D Printers, Natural

https://www.amazon.co.uk/gp/product/B07FQ46L1L/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&psc=1

Rather more expensive than standard material per kilo but you only use enough for the job, on this cleaning job, I use about 2x 75mm, one to intially clean and leave in for a few minutes, then feed some more through, followed by the TPU or whatever you are printing with.

It is also worth using this material when you change between very different materials, especially if the next material uses a much lower nozzle temperature than the first one.

Remnents of the higher extrusion temperature filament may clog your hot end when subsequently extruding at a much lower temperature for another filament.

This cleaning material can be used at temperatures from 160C to 300C.

Using this for preventative maintenance can save you from dealing with a completely clogged hot-end which requires stripping down and a thorough clean out, it's no fun I can assure you, especially if you then have to then spend and hour or so recalibrating the heads, which is 2.5x more tedious on a dual head machine.

Start the print again

No problems this time, this just reinforces that you should always watch the first layer being printed before walking away, this layer is the one where most problems can occur and if not rectified can ruin a whole build plate of items in a matter of a few minutes.

Another hiccup

The first 10mm did not get laid, unfortunately the second layer did not overcome this and then subsequent layers came adrift from the build plate, another rule, keep an eye on the first few layers too.

This time applied some Magigoo to the front left corner where the print starts, so although the first few mms were a little thin they stuck.

I think I need to revise my customised start G-code so the head either moves faster to the start of the print or it carries on laying a further length of prime line to keep the filament flowing on the way back to the build plate x-y origin where it then lifts to travel to where the printing actually starts.

Fun isn't it!

Result

First impression very good, reflections show an even build of layers, slightly concave walls on curves as usual.








Firm on build plate but easily removed by peeling inwards to the centre of the print, seam quite flat as the last print, good even finish of the layers.

Extruded at 0.5mm wide for a single pass for each layer the resulting dimensions (to one decimal place) are :-

Width 19.7 to 19.8 mm
Belt Thickness 0.5 mm
Seam Thickness 0.7mm​

Fit

Tensioner sits mid range when applied, but no belt slip without them.
Belt runs without binding being slightly narrower, nearer the original belt width.

Seam is minimal in appearance and is some what less than that possible with ovelapped fabric or rubber belts.

Print Process

The settings to print this long TPU belt give a good compromise between seam thickness, eveness of layers and printing time, in this case 2 hours. Filament based 3D printing can be quite slow, but once running only needs occasional monitoring. It's not as though you need to babysit for that time.

High printing speeds promoted by many machine manufacturers are not always advantageous, it can still be limited by the materials and the nature of the object you are printing.

For instance the Raise3D E2 is what is colloquially called a 'bed slinger', because the higher speed machines move it at high speeds, even the E2 can move it quite fast.

The disadvantage of these is that tall objects with a small footprint can become dislodged late in the print process, as can thin items like these belts. So there are times when a machine with a bed that only moves vertically and the extruder/hot end assembly moves in the x-y plane , i.e. parallel to the build plate is advantageous.

I need to refine my customised start G-code due to the second hiccup above, but that is something I can look into seperately.

So subject to any issues arising during motorised testing I'll consider the conveyor belt as job done, I'm finally happy with it, all being well I will print a second as a spare for the conveyor that still has a serviceable original belt.

Time to move on to designing the elevator belt.

 

13 - Elevator Belt

TPU Drying

Whilst the designing starts based on the last plain belt design, a drying session for the TPU, this has performed somewhat better than expected over the last week with only a 50 gm desiccant sachet in the base of my external spool holder which is not a sealed box.

Increasing moisture absorption could be the reason for first print of the day failing at start up the last three days.

On with design.

Elevator Belt

The elevator belt is the more demanding belt to create due to its teeth for lifting the 'coal' out of the bin, it is however much shorter than the conveyor belt and so can be printed as a circular part which greatly aids the placement of the teeth.

Elevator Tower

This is a separate component as supplied and stored as can be seen in the instruction leaflet, first image of the introduction, this is the one with the replacement DC motor. As this motor was too long an aperture had been carefully cut out of the back of the hut on the elevator base.

Shown here with my first attempt at a replacement belt using a plastic tape, apparently an automotive PVC cable tape according to the package in the elevator box, all the teeth I glued on have now come off, the adhesive was not suitable.





Next, what remains of the once 'serviceable' elevator belt, extremely brittle, breaking up as I handle it, the underside has ribs to match the drive roller serrations.



The elevator drive does not have a tensioner and hopefully the drive method used on the conveyor will work with this belt too, however it's length could be rather critical, if necessary I may need to use a softer more elastic TPU or even a very different composition of material.

Initially I will endeavour to use the same TPU as for the conveyor belt.

The Process

I envisage this as a two stage process.

1. Determine and print a plain belt to the correct width and length
2. Design and add the appropriate number of teeth around the belt

Design the plain belt to be built from a number of segments which is a multiple of the number of teeth, create one tooth, and then copy and paste it onto each second or third segment around the periphery of the circle as required.

Some design iterations will be required to get this right.

Start by estimating the number of teeth that would have been on the original belt, I can judge this by the remnants I have, although possibly not a complete set.

Dismantling

Starting with the re-motored tower as below which has had the bin cut-off, which is not an issue.

Remove the screws holding the elevator to the base, then the tower from the base.

Next the screws retaining the elevator to the tower, after which the drive shaft from the motor can be detached, releasing the elevator from the tower.



Remove the circlip from the free end of the drive shaft, fiddly!

Remove the three 8BA hex head bolts and it all comes to pieces, the centre grey rectangular tube is two thin metal U channels.




Dimensions

The distance between the drive roller and the idler measures as 164 mm, serrated drive roller 12 mm diameter, plain idler 11.8 mm, both rollers as for the conveyor.

Pitch of the teeth is approximately 11 mm this varies due to the distortion of the belt remains.

This belt is thicker at 1.25 mm, so print as 1 mm, it is likely it will need to be stiffer than the conveyor belt.

Treating both rollers as 12.8 mm diameter and ignoring the thickness of the belt,

estimated belt length = circumference of one roller + ( 2x distance between rollers )

length = 11.8 x 22/7 + (2 x 164 mm) = 365 mm

So if printed as a circular part, diameter would = length x 7/22 = 116 mm

With an approximate tooth pitch of 11 mm this equates to 33 teeth, make it 36 which is good enough for a guide line on how many segments to construct the circular belt with.

The root of each tooth measures at a nominal 3mm.

This is my starting point, same width and profile as the conveyor but 1mm thick and conventional start up G-code, i.e. direct route from end of prime line to start of the print, estimated time 1hr 34 mins.

Dimensions of the first plain test belt is as shown below.



First the plain belt test

Too short add 8mm, thickness of 1mm too stiff, reduce the thickness to 0.5mm, can't ignore the belt thickness so much this time regarding length calculation.

Second the plain belt test

New diameter would be length 373 x 7/22 = 119 mm, internal diameter 118 mm.

Dimensions of the second plain test belt is as shown below.




Drive Roller

Use basic TPU profile with fixed start and scarf to see if this works for a small item.

Nope, scarf mode for smoothing out seams looks worse than an ordinary seam on this 10 mm diameter roller.

Back to a basic seam result with fixed point layer finish/start joins.

First attempt broke when fitting the elastic sleeve, must get the TPU filament dried, for the moment, upping the nozzle temperature from 210C to 230C prints better.

Second okay.

Push fitted the roller onto the drive shaft.

When re-assembling the elevator frame the long screws can be rather fiddly to insert through all the holes in the frame and rectangular tube, they also protrude only a little for attaching the thin 8BA nuts.

So I inserted the bolts from the gear side and taped them in place with masking tape.



Turned it over and then mounted the tube, then the idler roller with belt.



Attached the frame and run the nuts on with a small nut spinner, at least my 8BA spinner fits these nuts, hopefully another for 10BA will arrive in the post today, my other one is a tad too small across the flats. Partly due to rust on the heads of the conveyor belt frame.



Well this size of belt grips the sleeved roller so I shall use it as the basis of an actual toothed belt.

It feels as though it could do with being a little tighter but when the teeth are attached it will stiffen and so take up a small amount of slack, this will be an iterative process, especially as this belt does not have tensioning device available.

Now the fun really starts!

 

14 - Creating the Elevator Belt

So based on the plain belt from above a tooth profile was created based on the original samples with additional footing added to compensate for a thin base belt, original approximately 1.2 mm, TPU belt 0.5 mm.

The following method allows distribution of a shape around a circle with relative ease.

However it is probably not suitable for gear teeth, I set the pitch of the teeth according to the angular spacing required to place the required quantity of teeth around a circle whose circumferencial length is defined by the length of the belt.

The actual gap between the teeth does not matter for the outer surface of this belt whereas for a gear it does matter, it must be compatible with a gear meshed with it.

The YouTube video I used to do find out how to do this, (with the free Sketchup Make 2017) is here, this guy explains it much better than I can :-




Let's get started.

Tooth profile of both the original and a deeper one for a thinner belt.




So having created a single tooth the lines making up its profile are grouped so it can be copy pasted as one item.

The core belt with a single tooth shown.




So next zoom into the top edge of the belt and place the first tooth in position.

The bottom right hand corner of the group (all those in the blue square) is located onto a vertex of belt outer edge.




Next rotate this tooth about that point to align it to the segment of the belt it is to be attached to, the Rotate tool is used for this.

When the segment turns magenta and a message confirms it (that didn't get caught in the snap shot) the tooth profile is correctly positioned on the belt segment, a mouse click fixes it in postion.




Next make the tooth and the segment it sits on and an adjacent one independent of the rest of the circle by drawing lines tangential at the vertices. Then delete everything else.




Explode the remaining parts and delete the line between the base of the tooth and the two remaining segments of belt, they then merge as one shape, note they have no thickness at this time it is just a profile or outline from which we create the solid part, i.e. a 3D part.




Then use the push-pull tool to give it depth, in this case 19.5mm.



So what we have here is 1/36 of a circle with a tooth attached, the next step is to group the lines which define this part so it can be manipulated and replicated as one single piece.

Now the fun part, if I can only get it right, is to 'rotate with copy' and 'multiplier' to copy this part a further 35 times about the centre of the original circle it was built on.

Perform a rotate with copy selecting the bottom right corner of the tooth, and move the copy until the selected corner is placed upon the left most corner of the original part.




Then without touching anything else type in *35 to apply this copy 35 times, if you are lucky this is the result, if not start the rotate with copy again!

One false move and it breaks.




Changing the viewing angle the elevator belt can be seen ready for cleaning up.




Coffee Time

That's hit the 10 image limit for a posting so time for a coffee and brain cell rest before continuing.

 

15 - Design Continued.

The next step is to combine all these parts as one homogenous item.

Select all of the belt parts and 'explode' them so the become one, group them all and run the Solid Inspector to deal with any internal or reversed faces, select Fix All, then all good to move on.




Next export the CAD model from SketchUp to an STL* file on disc, and then import it into IdeaMaker for slicing.

*STL: this is a universal format independent of the CAD file format which is used to convey the structure of a CAD model to the slicer application, in this case IdeaMaker.

It's not happy with it, complains of Non-Manifold edges*, which basically means the geometry of the CAD model is such that it cannot be converted into a real life object.

*Non-Manifold edges: Quoted from Google "Non-manifold geometry is defined as any edge shared by more than two faces"



This was actually expected but just thought I'd demonstrate.

Rather than using the slicer to repair it I prefer to correct the CAD model.

The lines at either end of the tooth, one end circled are surplus.



Whoopee, removing those makes entire faces dissapear, but all is not lost ...



Just redrawing one edge of each opening causes most openings to heal up as a surface, this seems to be very much a qwerk of using a civil engineering design tool for what are tiny components.

That done, run Solid Inspector again, all okay now!

Export from SketchUp to file, import the file into IdeaMaker.

This time the slicer is happy, so lets slice!

First look a the structure looks promising



However if I turn on display 'travel' which is the nozzle being moved from the end of one line of plastic that was extruded to the start of the next without extruding that is where stringing occurs. I managed to avoid most of that on the plain belts by eliminating 'travel'.

But here it is rife as the first attempt here jumps between teeth, just look at all the blue lines between each tooth which is likely to string, it will look like a bad birds nest, eeek!




Two and half hours printing and probably another hour fettling the stringing off.

First off can I persuade the travelling to be reduced at the expensive of speed?

Does not seem to be a way of getting the nozzle to go around the teeth.

Looking at the structure of the layers in the sliced preview it shows the teeth being printed very much as the CAD model, each segment joins another via a seam, the white vertical stripes.




If the nozzle followed the outer surface of the entire shape rather than treating it as 36 separate shapes the amount of travelling, and consequent potential for stringing would be substanitially reduced.

Note from the table that it spends 39 mins travelling and potentially stringing, this is also very inefficient printing.

Two options available

• Use a 3rd party tool to process the output from SketchUp to clean it up prior to passing it to IdeaMaker for slicing.
• Use another slicer to process the output from SketchUp better.

 

16 - Elevator Belt Slicing

This and the next episode are heavy going and is a distillation of the many hours I have spent trying to find answers and ultimately creating my own solution from the mass of similar questions asked by users of 3D printers, and their attempts to find solutions.

Hopefully what I present here will save you some hours should you want to take the murky road of using a 3D printer with 3rd party slicers.

So here we go ...

A quick look at other tools to post process the STL file exported from SketchUp are hellishly complicated but do not seem to do what I want, however I did notice that when the STL file was read into them it looked a lot better than expected, i.e. it did not appear that the file contained 36 fragmented parts.

After a momentary brain overload I decided to have look at what the Ultimaker Cura 4.10.0 would make of the exported STL file.

Ultimaker Cura 4.10.0

Just using a basic TPU profile it look substantially better in the pre-view, all the travelling is within the belt outline and not jumping between teeth, it is present all around the belt even though it only shows in patches below.

Note that blue lines representing travel do not pass across the gap between the teeth.

The Travel -> Combing Mode keeps travel within the printed area so retraction is not required.
Teeth look well anchored to the core of the belt.

Draft mode printing time estimate is 1 hr 20 minutes




Increasing the infill to 100% to get some bulk into the tooth base the flow of extruded plastic goes around the tooth well, estimated time now 1 hr 31 minutes.




I think this needs to be printed to find out how good it really is, how well the settings from Ultimaker suit the R3D E2, they are also for a different brand of TPU to what I am using.

Does, IdeaMaker have 'combing', apparently not at this time.

Ultimaker Cura Slicer

So looks as though Ultimaker Cura is the tool I need to slice with to achieve a minimal string print with TPU.

However porting via Wi-Fi to the R3D E2 can only be done via IdeaMaker, but when I import the G-code file from Cura 4.10.0 into IdeaMaker, the preview shows the belt sits approx 4 mm below the bed, not good.

Try again with a later version of Cura, 5.9.0 on my Win11 workstation (5.N.n requires Win 10+) and see if it is different, otherwise the G-code will need to be passed to the R3D by USB stick, this issue of having to use a manufacturers own slicer to port code via a network appears to be a common proprietary mine field.

But if IdeaMaker cannot slice for TPU properly then it’s back to the dark old days of using a USB stick to transfer G-code files to the printer.

Ultimaker Cura 5.9.0

Using Cura 5.9.0 on my workstation an extra viewing mode is available, an animation of the plastic being laid, very useful, not available with Cura 4.10.0 for Win7.

But oh dear, the combing would not work, Cura 5.9.0 produced a similarly littered with travelling between teeth as IdeaMaker, rats, tried earlier versions of 5.N.n to no avail. But as soon as I installed a version 4.N.n on the workstation it worked.

It would seem that in the major version change from 4.N.n to 5.N.n something got broken, unfortunately not an unusual occurrence with software, M$ has been like that for more than a couple of decades now.

The animation tool appears on Cura 4.10.0 on Win 11 but not Win 7, another query to Google, appears that it is only available when you have a sufficiently powerful graphics processor on your machine, thanks Ultimaker, telling me that during installation would have saved me a couple of hours!

Okay, so update Cura on my Win 7 machine to the latest that will run on Win7, 4.13.1, it reckons one of my settings profiles are corrupt, arrgh!

Back to tried and tested Cura 4.10.0, oh boy, hopefully one day all these people will get their act together. Who am I kidding?

Back to slicing with Cura 4.10.0

This has combing that works, and if I want to see the animated display on how the plastic is laid I will just pass the STL file over to the Win 11 workstation with its high performance graphics processor.

G-code to the printer via Wi-Fi network

What I want to do now is see if the output file from Cura can be read into IdeaMaker for sending via the Wi-Fi network to the Raise3D printer.

First attempt the G-code loads into IdeaMaker below the table level, which suggests at best, only 3/4 of the belts height will be printed, at worst nozzle hits side of the bed, bad, I need to be careful here.

I suspect the start up G-code I had for Cura is incorrect, so I will copy that used by IdeaMaker to the Cura profile and try again, unfortunately I cannot modify the output from Cura with IdeaMaker, it can only be read in and then uploaded to the printer.

Alternative is to not add any code for generating a prime line, instead I will have Cura create two surrounding circles of skirt* before the actual print starts.

*skirt: an extruded line of plastic around but not touching the object to be printed allowing the flow to stabilise.

Continued ...

 

17 - Slicing Issues Continue

What do I have now?

With combing and with a skirt the displayed preview in Cura is




When imported into IdeaMaker




Which looks safe to upload to the printer.

Why any attempt to add code to set up my custom prime line or even why the default code can push the object through the build plate escapes me without an in depth knowledge of G-code and even more so the hidden from view proprietary operations of these tools, intentional or bug based!

Cura 4.10.0 Preview

Does not show the result of actioning custom Startup G-code on the preview, only what it will apply, IdeaMaker would, e.g. it just shows a travel from the front left corner (X,Y origin) to the start of the skirt.




When I load the G-code file into IdeaMaker it jumps straight into preview mode and shows




Which is a little worrying as it shows the first layer what I can only described as projected below the surface of the build plate, because it does appear to show it being deposited on the build plate.

The custom start up G-code has been processed, why does Cura ignore it, despite it being low in the Z axis.

View from underneath.



Close examination of the G-code file does not reveal, to me anyway, a demand to go below the build plate surface, i.e. negative on Z-axis.

When stepping through the layers the Z values look okay.

At this stage a modest workaround that gives the desired result is the most cost effective solution, as I do not want to reverse engineer what is going wrong.

Custom G-code for Cura exporting to a Raise3D E2 printer

For those that may want to use it, it is on the official Raise3D Community forum, wriiten in German, but not a problem as the author presents it as screenshots of where the data is entered which is laid out the same in English as in German.

https://forum.raise3d.com/viewtopic.php?t=22422

This should also be useful when exporting from other 3rd party slicers.

So the process becomes :-

1. Create the part in SketchUp Make 2017
2. Write the full model in a universal form (STL format) from SketchUp to disc.
3. Read the STL file into Cura 4.10.0 and slice with combing enabled.
4. Write the result in the language of the printer (G-code) to disc.
5. Read the G-code file into IdeaMaker
6. Use IdeaMaker to upload the G-code to the printer via the wireless network

Clear as mud, if so, never mind, I still get bogged down in the mud too!

What Now?

I cannot believe what either preview is showing me, if I remove all the custom startup code before slicing at Cura it looks fine at IdeaMaker, but it needs some, I am led to believe.

After much brain ache there was not anything obvious in the G-code that would crash the nozzle into the build plate, and also the huge bug fix lists that come with any update of Cura. (like why was combing broken in v5.N.n) does not fill me with great confidence.

Also IdeaMaker is not perfect either, despite the number of bug fixes applied to the latest version I decided to cautiously go ahead.

So, proceed with caution to upload this result to the printer and try printing it.

But first TPU dispensing needs improvement.

 

18 - TPU - Dispensing

This has had a thorough drying ready for use and is in a sealed bag with desiccant, so far I have dispensed it from my external spool holder with desiccant in the base, what I want to try next is to dispense it from a sealed container, well, except from where it goes to the printer.

I decided to try a Kingroon Filament Dryer* Box.

https://www.amazon.co.uk/dp/B0BTVCP2XC?ref=ppx_yo2ov_dt_b_fed_asin_title

* Please Note: they say it is a 'dryer' box, but, filament can only be dried by heating which allows the moisture to escape from the filament by breaking molecular bonds. The desiccant is used to absorb the moisture from the air in the container before it can be absorbed into the plastic material. So, hot dry first, then store in a low humidity container, kept low by the desiccant.

The advantage of a box like this is that it should be possible to keep the TPU in good order longer between drying sessions whilst attached to the printer. Although the box will still need to be opened to load or unload the filament into or out of the printer, but at least it can be minimised.

Or perhaps I can consider a modified approach, like feeding it directly by tube to the extruder with the machine lid open. This will also help keep the build plate cool. In which case the feed tube stays full of filament and could be capped off.

Hygrometers

The hygrometers supplied with these boxes are held in place by a large clear sticky pad which some reviewers say is a nuisance, e.g. it was stuck to the instructions.

Well, I had bought two of these boxes, one hygrometer had already come unstuck so I just rolled up and off the thick clear double sided sticky pad, and the one still attached to the box easily popped off of the inside of the box with the twist of a 9mm wide screwdriver blade. Once again rolling it up off the hygrometer, you need to be careful not to damage the LCD display.

Removal of the hygrometers would have to be done some time to change their button cell. Doing it now made it easier to put them in a sealed bag with desiccant and a calibrated meter as reference to get some idea of their accuracy below an RH of 20%, in fact they were within 1% at an RH of 10% reading of the reference meter.

This is very good for a low cost generic hygrometer because RH values are not easy to determine as it is also temperature dependent, after all RH is Relative Humidity and certainly not absolute, but it's good enough for this use.

Having confirmed with a calibrated hygrometer that the RH readings of the included hygrometers are pretty close to it I will set this up for the next print run.

Ultimately I will mount the hygrometers in a different way.

The setup of printer and dispenser becomes as below.






Supplied desiccant and hygrometer loose in the base of the container, the RH reading settled down at 10%, which is just about as low as these types of hygrometer can measure. Others of the same generic design typically show LL below an RH of 10%.

The supplied tube for the filament is rather short and stiff so I cut a length of 3rd party tube I had bought on Amazon to go directly to the extruder with just enough slack to reach across the build plate, the included cutter works well.

Super Print 3-Meters PFA Teflon Bowden Tube 260 Celsius High Temperature Transparent PTFE tubing 2 X 4mm for 1.75mm Filament for Ender 3 CR-10 MK3 MK3S Mega 3D Printer ect.9(PFA)

https://www.amazon.co.uk/gp/product/B0BZC3W3HL/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&th=1

Bit of a struggle to push the TPU through this tube from about 2/3rds in, flexing it a few times as I fed the TPU helped and finally got there, pulling was fine, and pushing back a few mms to mimic a long retraction did not cause a jam.

Connected this tube directly to the extruder as I loaded the TPU into the extruder and ensured flow was okay ready for the next print.

Nervously, as the slicer pre-views were not encouraging, time to print ...

 

19 - Print with Caution

The moment of truth, transmitted the G-code file created by Cura to the printer via IdeaMaker.

The first difference to note when a G-code file was not created by IdeaMaker is the lack of detail about the print job which normally appears on the printers touch screen. Obviously much more is transmitted normally from IdeaMaker. Most of it is eye candy but the printer is able to determine total layer count and estimate remaining build time as it goes along, a similar issue occurs when loading G-code into the DaVinci printer not created with the manufacturers XYZPro slicer.

So, bearing in mind the different pre-views provided by Cura and IdeaMaker I need to be prepared to stop the printer quickly, so finger hovering over the power switch of the printer I started the print.

Normally the level sensing system just checks enough of the print area , but without this information from IdeaMaker it performed a minimal 9 point check of the whole build plate, okay so far, then the machine stopped, oh dear.

Quick look at the screen, paused due to filament detector, no filament detected.

Whoops, my fault, bypassing the normal entry point with the Kingroon dispenser I had forgotten to disable the detector, once disabled, selected continue and then back to hover over the power switch again.

Well, it executed the custom startup G-code which applies a prime line and started printing the skirt, sigh of relief, so both of the pre-views were incorrect.

The cross-section of the teeth is very small and a few did not stick directly to the AJOYIB sheet, the next pass started ripping some up again, okay, stop the print.

Second Attempt

Applied some masking tape to mark out the area the print was being placed, removed the failed print material and applied Magigoo within the taped area. This is best used on a warm build plate so heated it up to 40C for a minute then let it cool off down to 30C which allows the Magigoo to dry quicker.

Started the print again and the tiny detail printed on the first pass adhered this time, and for the next few passes, and next layer, looking good.

Leaving the bed heating off it just settled to about 25-30C for the remainder of the print.

The print continued to successful completion in 3 hours, after which I left it to cool down for 30 minutes before peeling from the build plate.








Result

The layers looked consistent and the belt quite pliable, dimensions to 1 decimal place.

Width 19.5 to 19.6mm
Thickness at core 0.7mm​

Stringing present on the inside of the belt as strings will not go around corners but relatively easy to rub off, if there had been one per layer (97 of them) between each tooth it would have been a nightmare to remove them all, so the combing feature of Cura has worked well.

A little stringing also near the roots of the teeth but minimal.

The stringing on the inside of the belt took a few minutes to rub off, plus some Magigoo that stuck to the TPU better than the build plate surface.

With the nozzle travelling around the outside perimeter of the teeth it has eliminated the string between the teeth and formed them much better than without combing.

Fit

Placing the belt in position




Side frame fitted






There is sufficient grip on the drive roller, although it feels as though it could do with being a little tighter but being much shorter than the conveyor belt and without tensioners it will be challenging to get spot on, the only real give is in the elastic sleeve around the drive roller.

The belt flows well around the drive roller and the teeth appear to be tough enough for the job which is encouraging, belt seam is well hidden.

Motorised testing of the parts is required to find out how well the belts printed so far function in practice.

 

20 - Motorised Testing

This is where the functionality and fit for purpose of these new belts are tested in earnest.

Re-assembly

This is a fiddly exercise especially if the base is complete with the bin, but if like this example then the elevator can be detached and re-attached without removing the tower from the base plate.

Assuming the base plate is complete.

With the elevator re-assembled, engage the drive shaft into the bearing bracket on the elevator as you manoeuvre the elevator into position against the tower.

Attach the elevator to the tower by two screws but not tightened yet.

Attaching the tower to the base plate needs to be done with the hut in position if you have one to ensure the hut will locate properly on its pins against the tower.

Angle the elevator to engage with the bin (not present on this example).

Fit the tower back onto the base, engaging the bottom of the drive shaft into the bearing in the base whilst making sure the gears are meshed correctly.

If yours has a bin then also ensure the bottom of the elevator is positioned correctly in it.

Refit the screws that hold the tower to its base and tighten carefully to ensure all is square.

Then the two screws at the base of the elevator are fitted and all four elevator fixing screws tightened.

Although the original AC motor probably only ran in one direction a DC motor can run in either direction.

So if a DC motor is fitted ensure it runs the correct way otherwise the drive shaft will lift, jam the gears at the top of the tower and/or cause the gears at the base to disengage depending on the alignment of the gears.

First motorised test

Running the DC motor up to 6VDC the elevator belt runs easily but does seem to slip a little at one part of the belt, stringing issue causing a lack of grip?

This was first noticed when operated by hand before motorised testing.

Later I found out that between two teeth the belt is smooth, every where else it is a little rough.

So, otherwise fine. Moving on, the conveyor is attached as per the instruction sheet shown in the introductory posting, ensuring the gears are all meshed correctly.

Second motorised test

The teeth on the elevator belt just miss the conveyor belt where it would discharge onto the conveyor belt, rather close but okay.

Ran up the motor again and ensured it runs without any issue between the two belts and run for about 10 minutes without any problems, looking good, but could be better.

Next, I need to shorten the elevator belt slightly to aid grip between the drive roller and the belt.

 

21 - Reducing the elevator belt length

I thought earlier (13) that maybe this belt need to be a bit shorter, maybe 1 or 2mm shorter at the most, so what are my options, well two come to mind.

1. Re-draw from scratch.
2. Re-scale it minutely in the X & Y planes keeping the Z plane (belt width) fixed.

Re-scaling in the X & Y axes is potentially the quickest so let's try that first.

Re-scaling the elevator belt drawing

Internal belt length was marked earlier as 373 mm which included a contribution of the belt thickness making a nominal diameter of 118 mm. So the earlier drawing is marked incorrectly, shoot the engineer, er, that's me, okay an off-day.

Once I could call it a senior moment, but when you are a senior what do you call it, a senile moment maybe !

So the first design was



Internal diameter = 117mm

Internal length = 117 * 22 /7 = 367.7 mm

Want to decrease length by 1mm, new length 366.7 mm

New internal diameter = 366.7 * 7 / 22 = 116.7 mm

Scale X & Y axes by 116.7 / 117 = 0.997

Result, internal diameter of 116.65 mm



I wonder if this will actually print noticeably smaller, really pushing what is possible here!

Cannot be really sure what the finished print internal diameter was and what it will be because measuring the diameter of a flexible part such as this is difficult, just because it flexes. This time I will measure it while it is still attached to the build plate.

Anyway, whatever it was and becomes, so long as the internal circumference is 1mm shorter that is all that matters.

Now this will also reduce tooth height, but insignificantly, and besides making them a little smaller would give more clearance between them and the conveyor.

Will it change the belt thickness, probably not as the change would be miniscule.

Okay, let's print it.

Loading into Cura

Model errors, highlighted, turn it over, oh dear.




These were not flagged up by the Inspector whilst in SketchUp.

Reload into IdeaMaker which is a bit more informative.

Twit, I included the two teeth profiles (circled in red) in the STL file says IdeaMaker, two unprintable items as they have no thickness, an example of a mesh* error, which includes non-manifold edges.

*mesh: the model written in STL format represents the object to be printed completely in triangles, the resulting mass of triangles is a mesh.




So back to the CAD model, re-export the model correctly.

Now okay in IdeaMaker and Cura, it seems that the light blue underside is normal, not sure where the tooth profiles were in the Cura view. Nevermind, back to business.

Sliced in Cura - estmated time 2 hours, really, took 3 hours last time.

Pre-view looks okay, regarding travelling, let's go with it, not sure I have got the hang of the way Cura presents profiles yet.

Export the G-code to disc, load into IdeaMaker and upload to printer.

Three attempts to get the delicate first layer laid, I think a tweak to the first layer settings is required, or was it just moisture absorption over the last 24 hours as even the prime line wasn't so good, purged some filament through before the 3rd attempt started and ran okay.

Result

3 hours later, version 2 of the elevator belt completes, unlike IdeaMaker the authors of Cura do not have the detailed performance data of a Raise3D printer so cannot predict run time accurately, however the printer updates the estimated remaining time at it processes the G-code file.

As this item is the same at each layer it can quickly determine the expected run time and was predicting an accurate end time well before 50% done, it also displays the quantity of layers still to be printed which suggests this data is read from the G-code file well in advance of being used to deposit plastic.

Stringing: Noticeable string on the inside of the belt but the teeth are largely clear of string thanks to Curas combing mode, so some clean up is required.

Dimensions

This time measured before removing from the build plate, key dimensions to 1 decimal place

Internal diameter: 116.6 to 117 mm
Core thickness: 0.6 to 0.7 mm
Width: 19.4 to 19.5 mm​

That is pretty good considering the nature of the material and how thin the belt is relative to its height.

Although the internal diameter was difficult to measure due to flexibility of the part even near to the build plate, my caliper jaws are not deep enough to reach the build plate, the printed part is very close to the designed dimension.

Fit

I have just found out that the elevator can be removed without detaching the tower if the bin is not there, just by removing the four screws which retain the elevator to the base and the tower.

Belt fit is better regards tension, but there is still room for improvement, but anyway on to more motorised testing.

 

22 - Motorised test of shorter belt

Elevator belt (v2) and conveyor belts run smoothly, so now try with some artificial coal.

Motorised Loaded Test

For this test I positioned the cut-off bin and try it for real with a granulated load.

I have two types

• a 3mm granulated rubber from Goodwood Scenics, looks and smells like re-cycled car tyres
• an unknown hard material from EezyLoads, has rounded edges to it

The granulated rubber is not dense enough, smaller pieces stick to the bin and also to the belt and easily jams the elevator belt.

The EezyLoads material is a hard denser material and flows better than the granulated rubber, although this Goodwood Scenics material looks better as a simulated coal it is not suitable for this elevator-conveyer. I suspect it would not discharge well from my train of re-wheeled Tri-ang operating 21T hoppers either for the same reasons as above.

With this unknown specification motor running at about 8V DC it runs the elevator well, at 6V it is prone to material jamming at the bin.

Still it's a good start, but I may have to tweak the design of the teeth to minimise jamming.

This slightly shortened belt grips better, but could possibly lose another mm of its length, however too tight will add significant drag due to the simple bearings of the elevator and increase gear wear.

Also in the event of a jam at the bin, belt slip is preferable to nylon gear damage, can' t replace those!

EezyLoads Artificial Coal

This is the EezyLoads coal I bought a few years ago

Model Railway Budget Grade Artificial Coal for OO/HO Gauge

https://www.eezyloads.co.uk/online-shop/artificial-coal-oo-ho-gauge/

They also produce this now

Model Railway Professional Grade Artificial Coal for OO/HO Gauge

https://www.eezyloads.co.uk/online-...sional-grade-artificial-coal-for-oo-ho-gauge/

which is more of a jagged look, more realistic looking, which I think I will also try shortly.

Although being jagged it may also jam in my elevator-conveyor and operating wagons, a trade off required here, play value, sorry, I mean operational efficiency vs. realism.

I think I need to consider ribs across the belt to increase grip on the drive roller.

Ribbed or Internally Toothed Elevator Belt

Two options available with the materials in use to improve grip between belt and roller without adding undue load to the very basic bearings of the elevator spindles.

Add to the inside surface of the belt

1. Ribs, to increase friction between the belt and the drive roller
2. Teeth to engage in teeth on the drive roller

Ribs

Ribs would create higher localised force on the elastic sleeve of the roller and so increase the grip without increasing the tension of the belt. This is because the tensile force the belt applies to the drive roller would be spread over a smaller area, however for this to work it also depends on those ribs being able to indent the surface of the roller.

The printing process will not print sharp edges, it is all curves because the plastic is extruded from the nozzle in an unconstrained manner, unlike plastic extruded into a metal mould.

So a compromise is required between what is printable and what will add bite to the belt.

Teeth

Now this is me possibly getting overly ambitious, what about the inside surface of the belt being toothed along with a matching drive roller.

Starting off with one segment of belt as before, don't worry I am not going through the detail again, a little bit of jugging of figures gives me the possibility of the teeth being rectangular, a nominal pitch of 2.5mm and depth of 0.4mm.

This maps quite well to the drive roller having 15 teeth of nominally the same pitch.

I say nominally, because 3D printing with filament is not that controlled dimensionally, especially as I am really pushing the bounds of the class of machine I am using, and with a flexible material as TPU into the bargain.

What I will try is to print the belt first and tweak the drive roller to match, it may also be better to print it in a more solid plastic like ABS.

Okay, going forward with an internally toothed elevator belt and matching drive roller.

 

23 - Internally Toothed Elevator Belt

Belt Design (v3)

The starting point is the tooth profile used to date and the current internal diameter of the elevator belt which from (21) is 116.65 mm (design value).

So what I have come up with as a starting point is using those same overall dimensions as the previous designs but mounting the 36 outer teeth on a 144 segment circle which is used to space the internal teeth, differentiate and call them drive teeth at one quarter of the pitch of the outers.

It's easier to see below.



This is mapped into a complete belt by the process of rotate with copy and multiply as before in (14).






The pre-view with Cura, travels not shown, the source of most stringing.








Travels, stringing in blue




Toothed Roller Design

By a couple of CAD iterations I found that a close approximation of a matching tooth pitch for the drive roller was for it to have 15 teeth, now all this is approximate because the teeth on either part at their overall physical size is very optimistic to print with any precision, but then again I am not designing a cam belt!

Once again the teeth are laid out based on angular position and 15 divides easily into 360 as 24 degrees.

Set the depth between the teeth to be 0.5 mm, slightly deeper than belt teeth height.

According to IdeaMaker the estimated print time is 3.5 hours, previously without drive teeth it was 3 hr 11 mins, so not as bad as I thought, the Cura time estimates without detailed data about the R3D printer is under 3 hours which is not likely.

Belt Test Print

I think a belt test print would be prudent before printing a whole belt so a short length that would completely wrap the drive roller plus a bit would be the best option.

So let’s say 12 x 22 / 7 = 37.8 mm minimum

One tooth segment is approx 10.3 mm

So just print a length of belt 5 tooth segments long as a try out.




Next to detail an initial design of toothed drive roller.

Toothed Drive Roller

A first try at this as below, how well the belt and roller will mesh is very much dependent on the printing process, so I start with nominal dimensions and will adjust as necessary.



I will print this in TPU to start with as it secures well to the drive shafts without glue or screws, if I have to go to ABS for the roller then I would probably use the grub screw from the original serrated drive roller to secure it well enough. Anyway that's a plan B, I always like to have a plan B and even a C.

Result

No marks for engineering precision with well rounded edges but the two parts do engage to a limited degree, tooth pitch acceptable, a full belt should grip the drive roller well enough to be worth the effort to print it.

The idler roller probably needs to be toothed as well but I can verify that when I have a full belt and drive roller assembled into the elevator frame and driven.

Parts as printed, this shiny black plastic is difficult to photograph under artificial light, no, I’m not going build a photographic studio or room, no space.




Full belt print

Estimated 4 hour run using the settings and methods used previously.

Stopped print third of way through, needed to go out for some tea and cake

Well, at least It's enough to get a feel for whether the full 4 hour print will be worth doing or not.

Fitting the toothed drive roller

The TPU printed drive rollers are a firm push fit on the drive axle, only problem is getting them off again, the nylon gear is a tight fit on the axle splines, and the splines will not fit through the axle hole in the side frame.

Drifting the axle out from a TPU printed roller is difficult with the nylon gear in place so I really need to be able remove the gear first.

Ah, but I don't need to drift it out, from my tool cabinet, a large loco wheel puller, closed up a bit removes the nylon gear, a heavy duty worm puller removes the TPU drive roller.




When Trix designed this toy it was quite heavily engineered, the nylon gear is a firm push fit onto splines and for extra security is also retained by a grub screw which bites into the splines, add 50 years plus a little rust it becomes a tight fit. Anyway it's off without damage and I'll clean up the splines before I refit it.

As with the conveyor drive shaft, confirmed that the splines are too large to pass through the bearings so don't try it otherwise the 'bearings', stamped holes in sheet steel will be irreversibly enlarged.

Next, fitting the one third wide internally toothed belt and toothed driving roller to ascertain how well this is going to work and whether the idler roller should also have teeth or just be smaller in diameter by the height of the teeth.

Must avoid the external teeth fouling the bin by being pushed too low by the idler roller.

 

24 - Compensating for the printing process

As expected the drive teeth profile is rather curved but also varies from curved to almost flat so does not engage that well with the toothed drive roller as the full width sample in (23) suggested.

Being narrow it also stretches more easily so jumps off the drive roller teeth.

I need to improve the tooth profile, in any manufacturing process where materials are moulded or shaped at elevated temperatures what follows is shrinkage as that item returns to ambient temperature.

How much this has to be allowed for depends on the temperatures, the coefficient of expansion of the material and the precision to which it is to be made to perform its intended function.

Some parts of course run at elevated temperatures which also needs to be taken into account at the design stage too, luckily the items I am attempting to make only run at room temperature.

Revision of the drive teeth

Having designed for perfect prints so I can see how much they deviate from the perfect I can estimate how much I need to distort the design in an attempt to counter the effects of the printing process.

The teeth are rounded from square so I should aim to make them wider at their tips.

They are shallow, so make them taller

The current sides of the drive teeth are tangential to the inside curvature of the belt as printed as a circle, so I will take the a single segment of the belt and flatten the teeth back to the belt and then raise them with the push-pull tool as rectangles so only their centre lines are tangential to the circumference of the belt as printed and increase their height to 1 mm.

One segment of the belt now becomes this




This may not be enough but it is a starting point.

Creating the complete belt from this single segment in the usual manner gives the following.




Seven segment test print for a quick check as a whole belt will now take well over 3 hours with all this extra detail.




Drive Roller

Initial re-work as an attempt to reduce the rounding of the the tips of the teeth.




Build plate adhesive change - GeeeTech JT16

This time I will try the GeeeTech JT16 glue stick, applied to a warm bed, in this case 40C, unlike the Magigoo which goes on as a liquid, this is very much like the typical glue stick. Its main advantage is it is a lot cheaper than Magigoo, but I have found Magigoo more effective for earlier tasks in the Arkitex project where the bed was heated up to 50C for some parts.

The main difference is that the Magigoo releases its grip as it gets colder, JT16 doesn't, as I am printing on a bed at starting at 30C and dropping to 25C it will be interesting to see how it works for these TPU parts.

Result

Both parts removed easily from the JT16, easier than from the Magigoo, but stayed firm during printing, also JT16 lasts a few prints, Magigoo I tend to use once per print with TPU because the TPU often pulls it away from the build plate.

Belt

The belt teeth are much better formed this time, they look like serviceable teeth.

Drive Roller

Not so good, the teeth have become rather mushroom headed and will not mesh with the belt, the printing path when creating them has distorted their shape so they need to be reworked again.

They have taken on an exaggerated (mushroom) form of the design, so revising in the same way as the belt may help, in other words flatten the teeth to the core and pulling them out as simple rectangles.

The next image shows, left, design just printed, right, revised form, retaining the pitch and depth but thinner tooth cross-section.




Print preview of the revised drive roller.




Next print the revised drive roller and a full elevator belt, time taken for belt 4 hr 49 mins, the detail of the drive teeth are greatly increasing the printing time now.

Result

The drive teeth are much better formed, on both the drive roller and the belt which is shown loose fitted here, they actually mesh much better when the belt is stretched slightly than this image suggests.



On testing, with the belt fitted around the original idler it now fouls the bin where it scoops out material so the idler roller must be reduced in diameter to accommodate the height of the drive teeth.

Idler Roller

Two options, toothed or plain.

The original TTR idler roller is made of two parts, the core that includes the bearings which fit into the elevator side frame and a tube that floats on it, only this outer tube needs replacing.

Toothed idler sleeve push fitted onto the idler core.




In practice this adds friction to the drive as the belt teeth squeeze into the idler teeth, fine at the drive roller but bad here.

Re-work as a smooth sleeve shown fitted here with the toothed one alongside.




Motorised test with tower and bin

This assembles and runs well, the belt is now more aggressively driven when dealing with scooping material from the bin, but if it jams completely the drive roller will slip on its axle or the belt will jump.

Next I need to check the discharge from elevator onto the conveyor is okay, what can go wrong here is if the elevator teeth are too tall they will foul the conveyor belt.

Motorised test with tower, bin & conveyor

Performed well, no issues between elevator teeth and the conveyor belt, but the odd jam against the bin base even before adding 'coal', I really need to check this out with a complete original base plate to be sure all is okay.

Placing the separate bin against the base with the tower is subject to misalignment and not much is enough to be a problem, ideally I want the gap between the elevator teeth and the curved base of the bin to be as small as possible, but not touching.

As the belt is being pulled from the top the lower edge as it arrives at the bin will be a little slack, and therefore droops a little, if the leading edge of a tooth hits the edge of the bin then it is likely to dig in causing a jam.

If necessary I will ramp the entry to the bin, my preferred approach is not to, I want the upgrade to an elevator-conveyor be purely a kit of parts independent of a buyer needing to cut metal, should I try and sell it.

But that's wishful thinking, the time to print 3D parts makes them expensive to produce, but then I could outsource manufacture, but of course the 3rd party will want their cut too. So I would save myself the effort but the added cost would be a turn off to a buyer.

Back to reality, I want the two machines I have had for 20+ years running again.

 

25 - Trapezoidal Drive Teeth?

There are many forms of profiles for flexible belt teeth to suit many applications, but I have to keep it simple due to printing in filament with a 0.4mm nozzle. Although the R3D printer can be fitted with a 0.2 mm the likelihood of being able to print TPU would be low, 0.4 mm is not easy as it is.

But the further I go the more I understand what this printer can do with this TPU and the further I look to improve the belt and the drive roller.

Belt Examples

Here are some professionally made drive belts.

https://www.bbman.com/blog/2013/08/...YAt_hKNgeTGRpMhCH3XKxWC_ZJh8QP8gG2Z6h7E4wqj2l

https://www.pfeiferindustries.com/timing-belt-tooth-profiles-and-pitches

So, I will try one of the easier profiles and that is the trapezoidal, the nature of the drive roller requires belt teeth that are tapered, it just about works with flat sided teeth because the material used can deform as it is stretched, not ideal but it works quite well.

But, can I do better, well let's see.

The Engineering

Ideally the contact between the teeth of the belt and drive roller will at a minimum be perpendicular to the tooth surfaces and tangential to the radius of action of the drive roller.

Even better if it is in rolling contact as with perfect gears, no friction, but that is hypothetical.

If they do not make perpendicular contact then there will be a tendency for the belt to ride up the drive roller teeth. Under duress it moves too high, loses grip and jumps by a tooth or more, at the moment I will settle with getting near to perpendicular.

What I have now

Viewing the belt, the current angle of the tooth sides on the drive gear is approx 5 degree from the vertical at mid height, remembering the sides are parallel to the angular centre line of 24 degrees.

When the belt is wrapped around the drive gear the tips of the teeth are brought closer together, calculating what their angle would be is tricky but perhaps I can measure it?

Working from the latest toothed drive roller.

The drive roller teeth are now of an acceptable working profile.

Copy pasting a pair of teeth into another part of the drawing and viewing them square on the flank angle of the teeth measures 12 degrees from the vertical relative to the root of the tooth.

Which, actually, is logical as the teeth centres are spaced angularly at 24 degree intervals.




Starting from a segment, created perpendiculars from each end of the roots, then measure in 12 degrees from the base of each tooth to create the required guide lines.

Join up the corners to produce this version 4, the notes at the right contain development information.




Create a single solid segment 19.4 mm deep, then a full belt.




Then a test segment.




Test segment pre-viewed in Cura




Test segment pre-viewed in IdeaMaker, just before upload to printer, looks promising.




Print the test segment.

One needs to remember these details are small, in the pre-view the 0.4 mm diameter nozzle is extruding plastic as a 0.3 mm line width and 0.2 mm deep layer, so the initial deposition on the build plate is critical, but so far so good.

Result

Very good, despite what they look like in the CAD models they mesh extremely well, so the compensation put in place for the teeth profiles and the angle placed on the belt teeth has come out far better than I could have hoped for, what of a complete belt.






Print a whole v4 belt

Next it's time to print a whole belt with the trapezoidal drive teeth which will no doubt take some time, I think it was about 48 mins for the 7 tooth segment, so looking at around 5 hours for a full belt. Looks like being a late night printing.

The use of the GeeeTech JT16 and the AJOYIB sheet on the Raise3D E2 are proving to be a very good combination for printing with this Overture TPU 95A.

I applied another coating of JT16 between the tape marks width wise, (first depth wise a couple of days ago) before the start of this print which resulted in an almost perfect first layer.

The specks of plastic that are deposited in the first pass almost look like stringing they are so small. The odd bit that didn't go down did not impede the next layer as larger adjacent rings of filament covered up for the missing specks.

Estimated run time of just under 5 hours indicated by the printer after the first few layers laid.

Time for a break while the printer is working.

 

26 - Fitting & testing the v4 elevator belt

Result

After 4 hrs 36 mins belt was completed, but as it was late it was not removed for another 8 hours, but it removed very easily from the build plate coated with the JT16 adhesive. I will keep to the JT16 for this job as it is the most effective of the two used this time.

Belt is a good fit to the drive roller, however almost 'too good' as seems to bind occasionally, but also the friction fit of the TPU drive roller to the drive shaft is susceptible to wear as the traction between drive roller teeth and belt has improved, slippage of the drive roller on the drive shaft is becoming apparent.

Alternatives considered for the drive roller

Could be that the drive roller needs to be of a harder plastic so that I can secure it to the drive shaft with the original 6BA grub screw, this will probably affect the tooth profile which may need modifying to suit the belt.

The belt design appears to be fit for purpose now, famous last words?

Drive Roller - tougher material?

There is a choice between ABS and the HatchBox PLA Pro+, ABS being more elastic than the PLA but unpleasant to print due to the fumes. Fume extraction on the R3D is poor, but the DaVinci with my fume extraction hood may not be capable of printing the part accurately enough.

Really must improve or rather fit a fume extraction hood to the R3D E2.

First I will try reworking two aspects of this TPU roller.

Drive Roller

Adjust

1. tooth profile - reduce thickness by 0.1 mm on each flank for extra clearance
2. bore - reduce by 0.1 mm for tighter fit

New dimensions




Cura pre-view




IdeaMaker pre-view - created from Cura output G-code file but tooth detail is different!

Not sure what I can really believe in these pre-views now!




On to print this v4 drive roller, while that is happening a close look at why the belt seems to stick at intervals.

Inspecting the v3 belt

Operating the elevator belt drive by hand, off the tower, I marked the belt with some masking tape as it started to stick and then dismantled the elevator assembly. Eventually I found that there was a small piece of 'coal' stuck between two belt teeth, was not much, but there is not really any spare space when the belt and drive roller are meshed together, with that removed the belt ran smoothly.

This highlights two needs

Sift out the fine particles which can drop off the side of the elevator belt and end up on the underside of the belt as it travels down to the bin. It gets squeezed into the valley of the belt teeth as it passes around the idler roller, which could possibly cause a the elevator teeth to rub the bin and cause the belt to judder.

If it is still wedged in the belt when it reaches the drive roller it gets jammed in tighter, and this is where having too good a fit between the belt and drive roller teeth can also cause a belt judder, so some clearance is required between the meshing teeth.

A juddering belt tends to flick small pieces of the 'coal' off the belt.

The revised drive roller

Although only 0.1 mm was taken off each tooth flank more was left off the resulting part, unlike machining parts the resolution of the print is relatively low resolution, sometimes in jumps of 0.2 mm or more even if only a change of 0.1 mm was required. Had a lot of fun with that on the Arkitex project.

So the revised tooth profile printed much thinner than the design.

The bore of the roller was reduced by 0.1 mm and the circle defining it changed from being a 24 sided polygon to a 30 sided polygon, SketchUp creates curves as a series of straight lines but these can be so short that in practice they appear to be continuous curves.

These tiny change greatly increased the tightness of the fit, in fact it needed the touch of a light hammer and a tube to position it.

Testing

Shown below the v4 belt with the modified drive roller, difficult to photograph as the free end of the roller is pulled to the left by the belt tension, but it does give an idea of the meshing of the two. It was noted that there was a slight curve to the roller tooth profile so I have orientated that to point in the direction of travel where it will bite into the belt harder rather than less.




Hand testing: it operates smoothly, tension good, at least I do not need to tweak the belt length again.

Motorised testing: a bin jam can almost stall the motor, I need to be careful because the two worm drives in series create a very high gear ratio, the belt did not jump and roller did not slip. I might need to fit a current trip as a safety feature in case it jams in operation when unattended for a short while.

This is however still with a 3rd party DC motor, my other elevator tower which is very clean and smart but without a serviceable elevator belt has an original Trix AC motor fitted. I do believe that this motor can be run from DC too, time for a Google...

TTRCA - TTR Collectors Association

https://ttrca.co.uk/LOCOSERV.PDF

Yes, it's a universal motor, 12V DC or 14V AC to suit the power sources available in the early 1930s when not all houses had electricity, 12V accumulator or transformed AC, as the necessary low voltage AC to DC rectifiers were not readily available then.

A lot has changed in the last 90 years!

Summary

So at the moment this modified TPU drive roller is tough enough for the job.

With a significant increase in the torque which can be transmitted to the belt from the drive roller, what gives if the elevator belt teeth jam in the bin, it needs to be the belt, by slipping or jumping.

I should also sieve out the tiny pieces of coal which end up in the conveyor frame and between rollers and the side frames, sieved material is probably the best starting point.

Perhaps I should also consider whether the belt should be wider, a closer fit to the frames, I'll sleep on that ...